Quanta Magazine

Why the Human Genome’s Tangled Physicality May Confound AI

Philip Ball — Thu, 18 Jun 2026 02:12:26 GMT

Samuel Velasco and Hannah Waters/Quanta Magazine

Since its molecular structure was deduced in the 1950s, DNA has been hailed by many biologists as the secret of life. They’ve read and studied the information stored in the DNA found in the cells of living organisms, known as their genomes, and claimed that this genetic database must be some kind of blueprint, code script, or computer. But if DNA really does harbor some greater secret about how life works, biologists have yet to find it.

In fact, the human genome is less a script than a puzzle that gets harder the closer they look. Knowing the entire sequence — the order of all 3 billion or so of our DNA’s chemical building blocks, nearly fully deduced by the international Human Genome Project between 1990 and 2003 — hasn’t helped much. That investigation showed that barely 2% of the human genome consists of actual genes, the information-coding sequences of DNA.

It’s now clear that understanding the human genome is no longer a matter of figuring out what each gene does. The deeper and much harder question is how those genes are used, or regulated, a question that seems to involve some and perhaps much of the rest of the genome. By switching suites of genes on and off, the many different cell types in our bodies can all be created from the same material. Cells also regulate their genes from moment to moment in response to a constant inflow of signals from their neighbors and surroundings. But the processes that govern gene regulation are proving so complex that some biologists wonder whether a full understanding of it — of how the genome really works — will ever be within the grasp of our puny minds.

Some are counting on outsourcing the analysis to artificial intelligence. Genomic “foundation models” such as Evo 2, Genos, and Google DeepMind’s AlphaGenome are trained on vast quantities of genomic data, which biologists use to make predictions about how differences in DNA sequence affect biological processes and ultimately the traits (including disease risk) of a whole organism. These algorithms don’t worry about the complicated regulatory stuff going on; all of that is supposedly subsumed by the algorithm’s “training,” through which it deduces correlations from cases we already know about.

This approach is likely to be useful, but for those who crave real understanding of how the genome, and ultimately life itself, works, a computational black box will never suffice. And perhaps more to the point, the genome might not submit to the kind of straightforward input-output approach that such AI models ultimately assume.

That’s because the genome is no blueprint or algorithm. It is something else.

The Old View

Given that it’s the product of around 4 billion years of evolution, perhaps it’s not surprising that our genome is complicated. The surprise has been what those complications are. “Our genome is not what we might make it if we sat down at the drawing board,” said the biologist Karen Adelman, who studies gene regulation at Harvard Medical School.

The traditional view posits that a small proportion of our DNA holds the code for making the protein molecules that orchestrate our cells’ chemistry. Each instruction for a protein is held in a corresponding gene — we have around 20,000 of these — and gene sequences can range in length from a couple of dozen to almost 3 million DNA “letters” (representing molecules called nucleotides). Making a protein from its gene is a two-stage affair. First the DNA is read, letter by letter, by an enzyme called a polymerase, which creates a copy of that code in a related molecule called messenger RNA (mRNA). This is called transcription. The mRNA is then read by a piece of molecular machinery called the ribosome, which constructs the protein — a process called translation. The proteins made by the ribosome then go off to do their jobs in making and sustaining the organism.

This picture is still more or less correct. But it turns out that “the genes are probably not the most interesting part of the genome,” Adelman said.

What matters more is how our genes, many of which we share with simpler organisms, are regulated: turned on and off. Which proteins a cell needs changes over time and according to cell type: muscle, brain, skin, and so on. How the genes that encode those proteins are regulated depends on some of the genome that doesn’t code for proteins.

Biologists have known about gene regulation, and the involvement of “noncoding” DNA, since the 1960s. But for many years, most of what they understood about this came from studies of simple organisms like bacteria, where the principles are generally straightforward. It has gradually become clear, though, that in complex eukaryotic organisms like us, gene regulation is far more complicated, involving overlapping systems of oversight and control, each with its own intricacies.

Transcription Factors

Transcription gets started by proteins called transcription factors, which are like the operations managers of gene regulation. These proteins stick to sections of DNA (typically close to the target gene) and recruit the polymerase enzyme to make an mRNA copy. In bacteria, transcription factors are rather like keys that fit the locks of unique binding sites on DNA. But that’s not how they work in complex organisms. In us, the logic of transcription factors is more difficult to parse.

For one thing, our transcription factors don’t show strong preferences for particular DNA binding sites. What’s more, they tend to work in pairs or groups. And a given transcription factor might have different effects in different contexts, such as activating gene transcription in one cell type but suppressing it in another, depending on which other transcription factors are around.

In bacteria, regulation tends to have an “OR” logic, Adelman said, whereby a particular signal turns a gene on or off: It’s either this or that. But in the human genome the logic is more like what computer scientists designate “AND.” Many signals are integrated to reach a regulatory decision: this and that and also that other thing. In this case, regulation can be more responsive to nuances of context, and the regulatory knobs are tunable rather than being just on/off. “This is part of the beauty” of our regulatory complexity, Adelman said.

When they interact with the genome, transcription factors bind to pieces of DNA called enhancers — which present a puzzle of their own.

Enhancers

Enhancers are gathering points for transcription factors, and they are thought to be the decisive influence on transcription: They deliver the “go” signal for a waiting polymerase to make an mRNA version of the DNA sequence. Seems simple enough, but mapping enhancers to their respective genes is far from straightforward. Our genome has hundreds of thousands, perhaps millions, of enhancers. That means we have many more of them than we have genes. Each gene might be influenced by many enhancers, and each enhancer might influence multiple genes.

“It’s embarrassing that 25 years after the Human Genome Project, we don’t know where all the enhancers are in the genome, let alone what they do when they act and which genes they control,” said Wendy Bickmore, a genome biologist at the University of Edinburgh.

Biologists do know that most enhancers won’t respond to a single transcription factor. Their activation “requires a cocktail,” Bickmore said. “That’s what gives [an enhancer] that exquisite specificity — because it’s only in a particular cell at a particular time that you have the right combination of factors to bind and activate that enhancer.”

Some enhancers are, as you’d expect, close to the genes they regulate, or even sit on DNA inside a gene. But others sit far away from the gene — perhaps millions of nucleotides away, with more genes in between.

The existence of such so-called “distal” enhancers “seems bonkers,” Bickmore said. “How do you get that information from over there to over here, to the gene that needs to be activated? That’s a largely unanswered question.”

One of the answers comes in the form of a loop.

Loops and Hubs

Distal enhancers are brought to the gene they regulate on great loops of DNA or, more strictly, of chromatin, the combination of DNA and its packaging proteins that are unraveled as if from a ball of wool. The loops are created by a protein motor called cohesin, which runs up and down the DNA strand and extrudes it as needed.

Once cohesin has formed a loop to bring elements together, what then? It was once thought that they then stick together or assemble into a molecular machine, but they don’t. Rather, the components appear to form a loose but dense blob in which they interact rather weakly, fleetingly, and indiscriminately — a sort of committee, sometimes called a condensate.

These transcription hubs are extremely fluid and differ from one cell to another. “There’ll be a bit of loop extrusion going on over here, in the next cell it might be over here, and the whole thing is turning over incredibly fast,” Bickmore said. Even if the cells are notionally identical — both skin cells, say — exactly what the gene-regulatory machinery is up to at any moment is never quite the same in any two of them.

Chromatin loops are just one reason why a gene’s transcription depends on the shape and structure of the chromatin around it.

Chromatin Shape

The textbook image of a chromosome — one of the 46 units into which our genomes are divided — is of a compact, X-shaped cluster of chromatin. But any time a cell is not actively dividing, its chromatin is unwound into what looks like a tangled mess. There is order to the chaos, however. Some parts of chromatin are densely packed into a form called heterochromatin. The compacted DNA there is relatively inaccessible to transcription factors; the genes it contains are typically silenced. Meanwhile, other parts are relatively loose, open, and accessible: This is called euchromatin.

There are special enzymes involved in packaging and repackaging chromatin, thereby controlling transcription. In other words, what matters is not just the encoded information in the DNA but also how it exists physically and dynamically in space. “We’ve stopped thinking about the genome as a linear piece of DNA code,” Bickmore said. “Thinking about this incredibly dynamic three-dimensional folding as absolutely inherent to regulation is a very exciting change.”

One aspect of this 3D organization is the clustering of segments of chromatin into compartments called topologically associating domains (TADs). Within a TAD, the genes seem to be coregulated: switched on or off in groups. Such groups keep suites of genes active or silent together to form and provide function in different cell types. Cohesin is also involved in the shuffling of chromatin to construct TADs — a dynamic process in which the chromatin is constantly rearranged in our cells.

Chromatin shape can also be influenced by chemical modifications called epigenetic marks: small molecules attached to DNA packaging proteins called histones or stuck directly to DNA. Some of these epigenetic modifications can alter the electrical charges on histones, which changes how the proteins attract or repel one another and so rejigs the chromatin packing. Epigenetic modifications to chromatin are like annotations of the DNA script that change its meaning in a given context. When cells divide, the epigenetic annotations are copied, too.

How and when the marks get added and changed, and what each type of mark means for gene activity, are complex questions with no simple answers. Some researchers talk of an “epigenetic code” governing this aspect of gene regulation, but it’s far from clear if anything so systematic really exists.

All of these processes and others can determine whether a gene gets transcribed into mRNA. But there are further layers of regulation that determine whether the mRNA is then translated into a corresponding protein — and which protein arises.

RNA Interventions

This post-transcriptional regulation is often controlled by RNA molecules that are said to be noncoding. These short-lived molecules aren’t templates for proteins, as mRNA is, but have other jobs of their own. While mRNA is produced from the protein-coding areas of DNA (so-called “coding genes”), noncoding RNAs are transcribed from other DNA regions now generally described as noncoding genes. These noncoding RNAs are versatile, taking on varied roles in a cell. Researchers are learning more about what they can do every day, and many if not most of them seem to be involved in gene regulation.

Small noncoding RNAs called microRNAs, for example, can silence mRNAs before they can be translated into proteins. They do this by guiding special enzymes to a particular mRNA to degrade or chemically modify it. The microRNAs don’t do this job alone but, not unlike transcription factors, act combinatorially, in groups, and in a rather promiscuous manner: A given microRNA might regulate many mRNAs, and a given mRNA might be regulated by many microRNAs.

Why make an mRNA only to stop it getting translated in a protein? This sort of post-transcriptional regulation is like having another checkpoint: Does the cell really need this protein? MicroRNAs can be mobilized to allow cells to adjust gene expression depending on the immediate context. In this way, the workings of the genome are less like a program’s inevitable progression and more like an adaptive and responsive process.

Another post-transcriptional complication is that mRNAs get translated to protein only after they have been reorganized. Fresh from transcription, an mRNA contains sequences that encode bits of protein, called exons, as well as sequences that shouldn’t be translated and need to be snipped out, called introns. (Strictly speaking, this pre-edited RNA is called pre-mRNA.) The job of editing introns out and splicing exons together is done by a molecular assembly called the spliceosome, which is made from several proteins together with various noncoding RNAs.

The spliceosome too can be sensitive to context, so that it might splice the pre-mRNA to encode one protein in one cell type and a slightly different protein in another. Sometimes these different protein “isoforms” can have very different roles. Transcription factors, for example, are often alternatively spliced in this way, and their isoforms can take on different regulatory tasks — some might activate gene expression, for instance, while others repress it.

Checks and Balances

All told, these and other regulatory mechanisms show that the genome is far from some automated program running in the background to build us and keep us alive. Our cells are, in effect, making complex decisions about how to use their genes — both the information they contain and the structure they assume.

Thus, cells need to assemble a rather loose and fuzzy committee of components, such as transcription factors and enhancers, to get transcription underway, which also depends on how the chromatin strand is shaped and molded at that moment. Then there are further layers of decision-making and action-taking in between mRNA and the final, functional protein.

Remember, too, that all the players — from transcription factors to noncoding RNAs — are themselves produced from the genome in the same kind of context-dependent process. That makes the genome rather like a recursive, self-referential system that the computer scientist Douglas Hofstadter dubbed “a strange loop.” It acts on itself, mindful of its own history (which determines chromatin conformation and epigenetic markings, say) and heedful of messages from inside and outside the cell. Not, then, a blueprint.

And for that reason, not at all easy to understand. “I wouldn’t have designed it this way if I was God,” Bickmore said. “But here we are!”

Why is gene regulation in animals like us so darned complicated? One potential answer is that evolution doesn’t have the foresight to design with efficiency and transparent logic, but merely tinkers with what it has already available. Maybe so — but eukaryotic gene regulation isn’t just a messy version of what happens in bacteria. It has different principles, and there’s surely a reason for them.

Bickmore suspects that the complexity of regulation and of genome organization might have been the only means of generating complexity in the organism. For example, organisms with many tissue types and varied lifestyles required more control over which genes were on or off in a given cell. One thing this demanded was more and more noncoding regulatory sequences in DNA. But then they couldn’t all fit close to the gene itself.

“As you get more complexity, you need to add more and more enhancers,” Bickmore said. “But where are you going to put them? You start to put them farther and farther away. Once they are [far enough], you start to need TADs and three-dimensional [chromatin] folding to allow those things to work.”

We also need regulatory complexity because, over evolutionary time, the human genome has accumulated DNA from parasitic viruses in the form of jumping genetic material called transposable elements. These sequences have inserted themselves all over our chromosomes and are good at replicating themselves. To sift the good DNA from the bad, we needed additional layers of regulation to ensure that cells weren’t translating RNAs they don’t really need or that could be actively harmful.

With so many context-dependent checks and balances in the workings of our genome, it is evidently not a program or algorithm that predictably generates the same outcome in every situation. It’s an open informational system that responds to external inputs and the genome’s dynamic internal conditions. This poses a challenge if AI relies solely on the genetic sequences within genomes to predict what genomes will do.

“A Highly Sensitive Organ”

Researchers developing AI-based genomic foundation models such as AlphaGenome hope that all these layers of regulation — transcription factors, splicing, epigenetic marks, loops, chromatin packing, and so on — will be implicitly included in the correlations that the algorithms learn between genetic sequence and organismal traits. They’re content for the complexity described above to be in a black box, so long as the model generates accurate predictions. But will that work?

“I’m sure [AlphaGenome] is going to be useful, but with limitations,” Bickmore said. “To me the big gap is in the complexity of the human body — in all the cell types and how they change over time in development. And all that data is missing.”

Fundamentally, the challenge is that the genome is not a set of static, linear instructions. It is highly dynamic, and it uses its information contextually, with combinatorial and promiscuous logic. “Whether we’ll ever be able to capture that aspect” in algorithms like AlphaGenome, “I don’t know,” she said.

Yet the problem goes even deeper because the functioning of specific organisms, including each of us, doesn’t just depend on genomes. Other factors, such as diet, environment, microbiome and, for us at least, culture, can matter hugely, too — not just in terms of how we act and how healthy we are but also in the state of our genome itself. The biologist Adrian Woolfson, co-founder of California-based biotech company Genyro, which aims to use AI systems for so-called “generative biology,” calls this information cloud the “informiome.”

“While the human genome forms the foundation of the human informiome, other layers of extra-genetic information are equally important,” Woolfson wrote in his book On the Future of Species, published in April 2026. Genomic foundation models won’t even be able to predict all the consequences of genetic mutations, he argued, because the relevant information is not in the genome sequence in the first place.

So how should we think about the genome? Maybe the only metaphors that can capture the way the genome really works must come from biology itself. In 2020, the biological historian Evelyn Fox compared the genome to “an exquisitely sensitive reactive system.” Rather than a sequence of genes leading to the formation of traits, she said, it’s more of “a device for regulating the production of specific proteins in response to constantly changing signals it receives from its environment.”

That sounds close to the picture painted by the geneticist Barbara McClintock in the address she delivered upon being awarded the 1983 Nobel Prize in Physiology or Medicine for her discovery of transposons. The genome, she declared, is “a highly sensitive organ of the cell, monitoring genomic activities and correcting common errors, sensing the unusual and unexpected events and responding to them, often by restructuring the genome.”

Research since that time has fleshed out this image, revealing how the shape of chromatin can matter as much as the information its DNA sequences encode and how an army of molecules collaborates to reorganize it and make collective decisions about how to use its genetic information in context-dependent ways. There is no human technology that works this way, so metaphors such as blueprints, programs, or computers will always fall short.

Bickmore is optimistic that the workings of the genome are understandable, despite its complexity. “We’ve got a handle on it now,” she said. “We might not know the details, but I think the whole field is coalescing now into a framework where we’re thinking along similar lines.” AI can surely help with this sense-making, but in the end, human reasoning will be needed to discern the fundamental principles.

“McClintock was far more on point than people realized at the time,” Adelman said. “What she said was that the genome isn’t static — it’s living.”

Seven Perfect Shuffles Randomize a Deck of Cards. But How Many Sloppy Ones?

John Pavlus — Wed, 17 Jun 2026 02:35:35 GMT

The magician Noah Levine performs a riffle shuffle.

Samuel Velasco and Chris Young/Quanta Magazine; all inset illustrations from The Expert at the Card Table by S.W. Erdnase.

In 1992, mathematicians famously proved that seven “riffle shuffles” — the kind where a player splits a deck of cards into two piles, then uses their thumbs to interleave them back together in a zipperlike motion — are enough to mix up the deck.

When Dave Bayer and Persi Diaconis came up with this proof, they also revealed something surprising about what happens along the way: At first, the cards stay relatively orderly. But with that seventh shuffle, the deck suddenly tips into a highly unstructured state. This kind of behavior, called a cutoff phenomenon, is of interest beyond cards, and many dynamical systems — including “spin glasses” in condensed matter physics — are believed to exhibit it.

Unfortunately, Bayer and Diaconis’ proof — referred to by some as a mathematical miracle — only works if you adhere to some rigid constraints about how to cut and shuffle the deck. If you shuffle more like a middle schooler than a magician, the result doesn’t hold.

Now three mathematicians have finally extended the finding to less precise shuffles. Mark Sellke, a Harvard University statistician currently on leave to work at OpenAI, along with Jialu Shi and Jiamin Wang (graduate students at the University of Cambridge and Princeton University, respectively), proved that a cutoff phenomenon exists for riffle shuffling even when you don’t cut the deck into two nice, even piles.

Diaconis was effusive about the update to his work. “It’s a fresh idea, and it’s remarkable that something like that would work as effectively as it does,” he said. “It’s a brilliant piece of mathematics.”

Mixing Cold Spots

To call the humble riffle shuffle “complicated” sells it absurdly short. The number of possible arrangements for an ordinary deck of cards is 52 factorial — that is, 52 × 51 × 50 × … × 3 × 2 × 1, or (roughly speaking) an 8 followed by 67 zeros, close to the estimated number of atoms in our galaxy. Another way to put the figure into context: Every time you shuffle a deck of cards, you produce a configuration that has almost certainly never existed before, and never will again.

But mathematical interest in card shuffling goes beyond its combinatorial complexity. Back in 1981, Diaconis and Mehrdad Shahshahani discovered cutoff phenomena in the context of card shuffling — after which mathematicians started to uncover them all over the place.

Persi Diaconis ran away from home when he was 14 years old to work with a magician. He returned to school 10 years later and became a professional mathematician. Card tricks continue to play a role in his research.

Caroline Gates

Cutoffs are similar to phase transitions in physics, such as the sudden crystallization of liquid water into solid ice at zero degrees Celsius. But cutoffs occur in the specific mathematical context of “Markov chains,” mathematical models that probabilistically describe how a system (like a deck of cards) moves between different configurations.

Cutoff phenomena, as their name suggests, happen in much the same way as Ernest Hemingway famously described going bankrupt: gradually, then suddenly. And while cutoffs are ubiquitous — they’re expected to occur in “most large, complex systems,” according to Sellke — it’s also hard to prove general theorems about them. “For most problems where one thinks there is a cutoff,” said Laurent Saloff-Coste, a mathematician at Cornell University who has collaborated with Diaconis, “one doesn’t know how to prove it.”

That’s why the “seven shuffles are enough” theorem was such a big deal. Bayer and Diaconis — who as a teenager ran away from home to apprentice with a magician specializing in card tricks, before becoming a renowned mathematician — didn’t just prove the existence of a precise cutoff in a real-world system. They provided a single formula for where that cutoff should be, and that formula worked for decks of any size.

Yet terms and conditions also apply. One: The riffle shuffle has to follow a realistic but strict model where cards are randomly interleaved from the left or right pile one by one. (Each card gets dropped from either the left or the right pile with a probability that’s proportional to the number of cards remaining in that pile. This means that the cards don’t simply alternate between left and right, which would result in a predictable structure; instead, the order might go “left, right, right, left, right, left, left.”)

Two: The deck has to be cut more or less in half before shuffling.

“All of our analysis depends on those details,” Diaconis said.

In 1999, Steven Lalley, a mathematician at the University of Chicago, attempted to loosen those constraints by seeking a cutoff proof for riffle shuffles that didn’t start with roughly evenly cut decks. “It seemed natural to me to ask — there are some people who tend to cut the deck a little higher or a little lower,” he said.

These less evenly cut decks have sets of cards that tend to stay in the same relative order even after multiple shuffles. While the rest of the deck looks well mixed, these particular sets of cards — which Lalley called “cold spots” — still retain information about their original locations in the deck.

Imagine, for instance, that you label your cards 1 through 52. After multiple shuffles, cards 16 and 17 will no longer appear right next to each other in the deck, but 16 might still tend to appear before 17 more often than it would in a random deck. If many pairs within a section of the original deck — say, cards 15 through 25 — show similar biases, then that set of cards forms a cold spot.

Lalley hoped to prove that when those cold spots disappeared, so would the last traces of order in the deck — giving him a way to show the existence of a cutoff.

But he couldn’t prove it.

Tracking Labels

Two decades later, in 2019, the son of Lalley’s collaborator Thomas Sellke — Mark, then a graduate student at Stanford University — found himself in one of Diaconis’ classes, where he learned about the original seven-shuffles result. “He mentioned offhandedly that if you don’t cut the deck in half, then nothing [about the proof] works anymore,” Mark Sellke recalled. “I was like, ‘This is it? … Come on, we must be able to do this.’”

By 2021, Mark Sellke had pinpointed the cutoff for decks cut much more unevenly than those in Bayer and Diaconis’ original work — including for decks cut into more than two piles. But the deck still had to be cut in the same way between each shuffle. He wanted a more realistic result, where the cuts from one shuffle to the next might look very different. And so in the summer of 2024, he teamed up with Shi and Wang, who had also expressed interest in the problem.

The trio first assigned each card a barcode. It starts when you cut the deck. All the cards in the left pile get assigned the number 1; those on the right, zero. Now shuffle, randomly interleaving the cards from the two piles one by one. Cut the deck again. If a card ends up in the left pile, add a 1 to its label; if it ends up in the right pile, add a zero.

As this process repeats through more riffle shuffles, each card builds up a longer and longer barcode of ones and zeros, which encodes its path through the shuffling process as it hops from left to right and back again. For instance, if the 17th card has a barcode of 0110 after four shuffles, that means it started in the right pile, ended up on the left twice, and then landed back on the right.

These numbers create a unique tracking label for every card in the deck. If two cards that started out in the same relative order — say, 16 and 17 — end up with the same barcode of ones and zeros, that means they took the exact same path through the shuffling process and are still in the same relative order.

To prove the presence of a cutoff, you have to show that very few of those matching barcodes remain after a certain number of shuffles — no matter how many cards you started with, or how the deck was cut. But comparing every barcode is time-consuming. Fortunately, the cold spots offer a shortcut, just as Lalley had hoped. Since those are the regions in the deck that tend to resist mixing, they’re the only places you have to check for barcodes that match.

Start with a deck of n cards and list the barcodes of all the cards in the deck’s cold spots in ascending order.

Mark Belan/Quanta Magazine

Then use that information to build a collection of points and lines, called a graph. Represent each card (labeled with its barcode) as a point.

Connect two points with an edge if they have the same barcode, to reflect the fact that the corresponding pair of cards has still not gotten mixed up.

Repeat this process with a second deck of n cards. Then line up the graphs representing both decks and identify where they overlap.

These overlaps indicate how long both decks continue to resist mixing. Sellke, Shi, and Wang then showed that after a certain number of shuffles (which depended on how many cards were in the deck), the overlap between the decks’ regions of unmixed cards disappeared at an exponential rate. This so-called exponential tail, according to Sellke, gave them an upper bound on the number of shuffles it would take to dissolve the last bits of orderliness in a typical deck.

To Clumping and Beyond

According to the new proof, that ceiling is roughly 14 riffle shuffles for a 52-card deck, if you cut your deck in a random place with each shuffle. Beyond that point, the cards will be fully mixed.

“For me the motivation is pretty simple,” Sellke said. “I occasionally play poker with my friends, and I want to know how many times I should be shuffling my cards.”

“It’s spectacular,” Lalley said. “I’ve known Mark ever since he was a young kid. The problem sat there for 26 years, and finally he cracked it.”

Technically speaking, there’s still work to be done. The model of shuffling that the new result depends on, like Bayer and Diaconis’ before it, still assumes that the cards riffle down one by one, rather than in clumps. That level of shuffling skill is beyond what most casual card players possess (although well within Diaconis’ own powers, as he made sure to note). Sellke said he’s “very interested” in attacking this version of the problem as another way to generalize the cutoff phenomenon in card shuffling.

“I really like this clumpy shuffle question,” he said. “I haven’t made progress for a while. Maybe after someone makes a bit of headway, then we can try again to adopt the understanding we’ve already produced.”

How Many Elementary Particles Are There, Really?

Natalie Wolchover — Mon, 15 Jun 2026 03:07:59 GMT

Kristina Armitage/Quanta Magazine

Every time I write about particle physics, I encounter a moment of uncertainty about a quantity that, at first glance, ought to be clear. How many kinds of elementary particles should I say there are?

In experiments at the Large Hadron Collider, physicists smash together beams of protons, breaking them up into all possible elementary bits and pieces. Meanwhile, they have an incredibly accurate set of mathematical equations for describing these building blocks and all the ways they fit together. So, since the known particles of nature can be both empirically observed and theoretically described, you would think they could also be counted. But alas not. I knew that, for reasons we’ll see, the census is not so easy as it seems.

So I recently emailed a few physicists to ask how each of them personally tallies nature’s fundamental constituents. The first indicator of just how complicated the issue is came in a reply from David Tong, the University of Cambridge physicist and textbook author, when we were scheduling a video call: “P.S. I think the true answer to your question is not an integer!”

We’ll get to that (it comes from a mysterious calculation from 2011), but let’s enter this rabbit hole from the top.

The known elementary particles and their interactions obey a set of equations called the Standard Model of particle physics. The Standard Model is a “quantum field theory,” a mathematical description of reality in which entities called quantum fields permeate the universe. Ripples moving through these fields are what we call elementary particles; some behave like matter, while others impart forces. The quantum fields and associated particles in the Standard Model underlie all known physical phenomena other than gravity, dark matter, and dark energy (all of which take unknown forms at a fundamental level).

In posters on classroom walls, the Standard Model displays 17 particles. There are 12 matter particles, or fermions: the electron, muon, and tau; three neutrinos; and six quarks. Each of them has a distinct set of sensitivities to various forces. There are also four force-carrying particles, or “bosons”: the photon (which imparts the electromagnetic force), the W and Z bosons (the weak force), and the gluon (the strong force). Finally, there’s the Higgs boson, a so-called scalar particle that’s neither matter nor force; rather, it imbues other particles with mass through its interactions with them.

Samuel Velasco/Quanta Magazine

It may just be this simple. “I think 17 is the right answer,” Melissa Franklin, a professor of particle physics at Harvard University, told me.

But every particle physicist, Franklin included, recognizes that there are caveats.

From 17, you can keep counting. Where you stop depends on your taste for complexity and mystery. The question of how many particles there are brings us to the edge of what’s known about the most basic levels of stuff.

There is one glaring problem with 17. To satisfy special relativity, each of the Standard Model’s matter fields supports both a particle and an “antiparticle,” which is identical to the particle except for having the opposite electric charge. This is what we popularly know as antimatter. So instead of 12 matter particles, there are really 24. Likewise, W bosons come in oppositely charged types known as W+ and W−. (This doesn’t happen to the Z bosons, photons, or gluons; they’re electrically neutral.)

Franklin excludes antiparticles from her census, she said, because mathematically they more or less mirror their particle versions. (Bizarrely, antiparticles are equivalent to particles moving backward in time, and vice versa.) Neither is possible without the other, so they shouldn’t be counted twice.

But I find that rationale unconvincing. Particles and antiparticles are undeniably distinct, even if they are secret twins. They can’t transform into each other (with the possible exception of neutrinos, which may or may not be their own antiparticles), and far from being functionally equivalent, they play totally different roles in reality. Matter is so dominant in our universe that any antimatter typically encounters matter quickly and annihilates. The reason for the cosmos’s matter-antimatter asymmetry is a major physics mystery.

Antiparticles bring the total up to 30.

But the notion that there’s only one gluon is another oversimplification. Really, the strong force is conveyed by eight gluons (and their associated fields), each possessing a distinct blend of charges known as “colors” and “anticolors.” The different gluons are impossible to distinguish experimentally, so Franklin, being an experimentalist, scoffed and shook her head when I asked if all eight should be tallied individually. Yet in the mathematical equations that define the Standard Model, the eight gluons are distinct from one another in the same way that the W and Z bosons differ. For consistency’s sake, we probably have to count all eight. So now we’re at 37.

Quarks come in colors, too — the three possibilities are dubbed red, green, and blue — and antiquarks have anticolors, called anti-red, anti-green, and anti-blue. (Don’t try too hard to picture anti-red; these aren’t our familiar optical colors, though they combine in a manner that’s analogous mathematically.) The colors reflect how gluons and quarks interact with each other.

For matter to exist in stable isolation, it must be color-neutral. So, just as red light, green light, and blue light blend to make white, so do red, blue, and green quarks form color-neutral protons and neutrons (the building blocks of atoms).

So there aren’t six quarks and six antiquarks but rather 36 in total. And that makes 61 elementary particles. But there’s more.

Matter particles also come in left-handed and right-handed varieties, a quality known as chirality — arguably a crucial distinction. “I insist on left- and right-handed particles,” Chris Quigg, a senior particle theorist at the Fermi National Accelerator Laboratory, told me. “I can’t account for this. Blame my parents.” (Far more idiosyncratically, Quigg leaves the force-carrying particles off his list, as he considers them to be transformations of matter particles rather than particles themselves.)

Chirality is a quantum version of the handedness that chemists see in molecules or that we see at the ends of our arms. It is not a geometric arrangement like those, but mathematically the two states are mirror images of one another; you can’t rotate one to turn it into the other, any more than you can with left and right hands. The force-carrying particles have an analogous distinction, known as a polarization state. Photons and gluons can be either left- or right-polarized, while the W+, W−, and Z bosons have a third, “longitudinal” polarization state as well. (That extra state has a complicated origin connected to the Higgs field and events during the Big Bang.)

Not everyone counts these different chiral and polarization states as distinct particle types. Yet it’s logical to do so, because they affect how particles behave and interact. The weak force, for example, affects only left-handed matter particles. For related reasons, neutrinos appear only in a left-handed form in the Standard Model. These are physically distinct states with different roles in nature. Counting each chirality and polarization state separately gets us to 118 particles — from a right-handed, anti-red, anti-charm quark to a green–anti-blue, left-polarized gluon, to a longitudinal W− boson.

“Now,” Tong said, “comes the weird stuff.”

Physicists call all the ways that particles can vary “degrees of freedom” — with a different degree of freedom for each state a particle can hold. Color, for example, comprises three degrees of freedom: red, green, and blue. But those differences go beyond the states we have already described. We might consider the tally of all these degrees of freedom as a more precise, mathematical version of the question of how many elementary particles there can be.

Physicists have long noticed a pattern in the degrees of freedom: The number of them depends on the scale at which you count them. On the scale of our everyday reality, objects are describable with fewer variables than it takes to specify the states of all the microscopic constituents. When you zoom in on, say, a proton, and reveal its constituent quarks with their colors and various other properties, you’ll observe more ways of moving or varying — more degrees of freedom. This is one of the main reasons it’s so difficult to pin down the particle population. The closer you get, the more their categories splinter.

Furthermore, the beginning of the Big Bang might have abounded with additional, high-energy particles that can’t form in our current, low-energy universe and aren’t part of the Standard Model. For instance, many extensions of the model to the high-energy early universe posit the existence of heavy right-handed neutrinos, but these would never arise now. “As you go down in energy scale,” Tong said, “you’re losing particles as you go, because they’re so heavy,” and therefore only possible at much higher energies. “As you go down in energy scale you lose knowledge of those particles.” If we continue to follow this idea, at very low energies only one particle is left: the photon. Because they’re massless, photons can approach zero energy.

It’s natural to wonder if a full accounting is possible. How many fundamental degrees of freedom are there, including all of those at the very highest energies and most microscopic distances that we can’t possibly detect? This brings us to the fascinating 2011 calculation Tong told me about, by Adam Schwimmer and Zohar Komargodski.

Komargodski, a theoretical physicist at Stony Brook University, walked me through it. I just mentioned the trend in which, as we zoom out in the universe, we’re able to detect fewer effective degrees of freedom. In 1989, the physicist John Cardy conjectured that this is an inviolable rule that any quantum field theory must follow. The rule had already been mathematically proved true of quantum field theories with one space and one time dimension, which describe particles moving along lines. But what about theories like the Standard Model, which involves three spatial dimensions plus time (called 3 + 1D)?

Schwimmer, an emeritus professor of physics at the Weizmann Institute of Science, and Komargodski proved Cardy’s conjecture. Their “a theorem,” acclaimed among quantum field theorists, says that in 3 + 1D quantum field theories, the number of effective degrees of freedom must always decrease as you zoom out. They showed that this is universally true by exploring how quantum fields must respond to gravity tugging on them in four different places.

Their proof also yielded a strange conclusion about how many fundamental degrees of freedom there must be in 3 + 1D quantum field theories such as the Standard Model. Quantum fields, the proof showed, cannot have just any number of variations. To the contrary, only specific values are allowed: Scalar fields such as the Higgs field have just one degree of freedom. Matter fields must each have 5.5 degrees of freedom. And force fields each have 62 degrees of freedom. These figures emerge mathematically, without regard to the specific particle states we’ve been discussing to this point. “And nothing else works,” Komargodski said.

“One, 5½, 62 — they pop out of the theorem,” he added. “I have no idea why this is what nature chose.”

Tong explained that fractional degrees of freedom (like that extra half degree possessed by matter fields) are variations that aren’t fully independent from those of other fields. What’s possible with one particle might depend on the state of another. “You kick that way, and suddenly all hell breaks loose, and the field is oscillating all over the place,” he said.

So assuming the respective number of degrees of freedom for each scalar, matter, and force field in the Standard Model, how many does that make? Komargodski paused our conversation to ask ChatGPT, providing the relevant numbers, and then checked its work. The answer: 995.5. That’s apparently how many degrees of freedom there are in the Standard Model.

I can’t help but feel flummoxed. And apparently that’s the general reaction.

“Underlying all of this is the statement that quantum field theory is unbelievably hard and we’re not very good at it,” Tong said. “There’s still a lot we don’t understand.”

Personally, I find myself to be a maximalist on the question of how many particles there are, even though (or because) it is a path to mystery. But I also see the appeal of 17.

Where Did Earth Get Its Oceans? Maybe It Made Them Itself.

Robin George Andrews — Fri, 12 Jun 2026 02:04:20 GMT

Ada Zejun Shen/Quanta Magazine

At this moment, a spacecraft is headed from Earth to Europa, an ice-veiled moon of Jupiter thought to contain an ocean similar in some ways to one of our own. NASA engraved a metal plate affixed to the spacecraft with a poem, commissioned from Ada Limón during her time as poet laureate of the United States. It reads, in part:

And it is not darkness that unites us,

not the cold distance of space, but

the offering of water, each drop of rain,

each rivulet, each pulse, each vein.

For decades, NASA’s exploration of the solar system has been dominated by the search for water in places like Europa, because as far as we know, water is essential for life.

It may come as a surprise, then, that scientists don’t really know how water first arrived here on Earth.

For years, the top theory was that water came to our planet via comets — objects made of frozen matter that orbit the sun, often decorated with sparkling tails. In all likelihood, these icy relics, which came into being at the dawn of the solar system, did bring water with them when they rained down on a primeval Earth. But in recent years, several spacecraft caught up to comets to examine them. What they found was that cometary water didn’t match ours; the chemical signatures were different.

After that, “comets sort of fell out of favor,” said Ashley King, a meteoriticist at the Natural History Museum in London. Asteroids — rockier and more metal-rich than comets — then became the most popular choice. Asteroids impact Earth far more frequently than comets do, and their water reserves (while not as voluminous as those of comets) look a lot more like those on our planet.

But asteroids have their own problems, and a radical new idea about planetary water is gaining steam. Through careful observation of worlds orbiting other stars, along with some explosive laboratory experiments involving diamond anvils and lasers, scientists have realized that rocky planets like Earth have a way to make water all by themselves. All you need is an ocean of magma, a whole lot of hydrogen, and a little bit of geological alchemy.

A Showdown Between Comets and Asteroids

Earth formed about 4.54 billion years ago. Through geologic fire and brimstone, much about its earliest eon has been lost to history, but the basics are agreed upon: It began as a ball of mostly molten rock. Then it became a blue marble. How?

Comets provided a well-motivated answer. They often linger far from Earth in a doughnut-shaped highway of icy objects beyond Neptune called the Kuiper Belt, or in the even more distant and nebulous Oort cloud. But when a comet passes close enough to the sun, its ice and frozen gases turn to vapor, creating a tail that can stretch for hundreds of millions of kilometers (in one known case, more than a billion). Compared to asteroids, comets give you “a lot of bang for your buck,” said James Bryson, a meteoriticist at the University of Oxford.

Scientists thought comets could have crashed to the Earth and provided its water. But nobody could prove that comets contained Earth-like water — until the 1980s, when the European Space Agency (ESA) decided to check. Giotto, their first deep-space mission, was truly ambitious: It would be the first spacecraft to get an up-close-and-personal look at a comet’s icy heart.

In 1986, it caught up to Halley’s comet, famous for appearing in Earth’s sky as our paths intersect roughly every 76 years. Giotto managed to send home both dramatic images of the comet’s nucleus and measurements of the cloud of material around it. What raised scientific eyebrows was Giotto’s measurement of something called the D/H ratio.

ESA’s Giotto spacecraft (top) took the first close-up images of a cometary nucleus (bottom) as part of its mission to Halley’s comet.

ESA (top); Courtesy of Dr. H.U. Keller

ESA’s Giotto spacecraft (left) took the first close-up images of a cometary nucleus (right) as part of its mission to Halley’s comet.

ESA (left); Courtesy of Dr. H.U. Keller

Almost all the water on Earth is made up of two hydrogen atoms and one oxygen atom: H₂O. But there is another form of water, called heavy water, made up of one oxygen atom and two atoms of a heavier form of hydrogen called deuterium.

If comets are responsible for our oceans, one might expect Giotto to have found that the water on Halley’s comet had a similar ratio of deuterium to hydrogen as the water on Earth. That’s not what it found. “It didn’t match at all,” said Karen Meech, a planetary astronomer at the University of Hawai‘i. In fact, Halley’s D/H ratio was twice that of most of the water on Earth.

More cracks appeared in the comet theory during the 1990s and 2000s, when spectroscopic observations of other comets, like Hale-Bopp, also found evidence of heavy water. But the hammer blow arrived in 2014 when the spiritual successor to Giotto, ESA’s Rosetta mission, made history by orbiting and sending a lander to the surface of 67P/Churyumov-Gerasimenko, a comet shaped like a giant rubber duck. During its orbits of 67P, Rosetta made the most precise measurements of a comet’s composition to date — and found that it contained the highest concentration of deuterium of any comet we’ve measured.

If Earth’s water didn’t come from comets, perhaps it came from asteroids. These rocky objects mostly hang out between Mars and Jupiter, and they impact our planet all the time as meteorites, though most of their material burns up in the atmosphere or lands in the ocean. Scientists have collected tens of thousands of meteorites and found that the water molecules contained in a particular group closely resemble those in our world. One meteor that plunged into the sleepy British town of Winchcombe in 2021 — leaving a sizable dent in a family’s driveway — was found to have a D/H ratio that almost perfectly matched that of Earth’s oceans.

Comet 67P/Churyumov-Gerasimenko was photographed in 2014 by ESA’s Rosetta mission, which also sent a lander to the comet’s surface.

ESA/Christian Stangl

Meteorites, though, can be contaminated during their fiery dives and crash landings. That’s why scientists have flown spacecraft out to asteroids and collected material in orbit for forensic analysis. In some cases, they have found that asteroids still moving through space also seem to have Earth-like water. A study published in 2023 revealed that water from the asteroid Ryugu, which Japan’s space agency visited in 2018, had a D/H ratio similar to that of most water on Earth. When it comes to the provenance of Earth’s water, “the community is probably more favorable of asteroids than comets these days,” King said.

But the D/H ratios of asteroids did not close the book on the question of Earth’s water. Asteroids also contain small amounts of noble gases like argon, krypton, and xenon — inert elements that act as tracers of various geologic processes — and scientists have found that those mixtures do not usually correspond to what we find on our planet. In addition, theories based on comets and asteroids have the same fundamental problem: The ability of either type of object to give the planet its oceans relies on luck. Multiple asteroids or comets would have had to impact Earth after its superhot magma ocean phase to produce the inundated world we live on today. This was taken for granted in the past, but the existence of this late-in-the-day bombardment is heavily debated in the scientific community.

There is another possibility, one that relies not on cosmic chance, but on our planet’s own industriousness: Earth made most of its water by itself.

The asteroid Ryugu, visited by the Hayabusa2 spacecraft in 2018, contained water similar to the most common kind found on Earth.

NASA’s Goddard Space Flight Center/University of Arizona

Hydrogen, Meet Magma

When astronomers look at exoplanets — worlds outside our solar system — they see a diversity of atmospheres. But when they simulate the ways the planets took shape, scientists find that many of them could have started out brimming with hydrogen. Could Earth’s formative years have been similar?

Scientists used to think that the early Earth had little hydrogen. They reached this conclusion after examining meteorites called enstatite chondrites that have a suspiciously similar chemical makeup to Earth. Because of this similarity, scientists think the two probably formed from the same material, Bryson said. These meteorites seemed to lack hydrogen, so scientists thought the same went for our planet.

But some studies, including one co-authored by Bryson, found that there was hydrogen in the meteorites all along. It was just hidden in their organic molecules, silicate glasses, and sulfur compounds. Perhaps, then, Earth was also awash in hydrogen in its early days.

Earth’s ocean of magma was full of oxygen. In a paper published in 2023, three scientists wondered what might happen if the hydrogen in a planet’s atmosphere and the oxygen in its magma were to mix — somehow. Hydrogen doesn’t just spontaneously bind to oxygen, so they aren’t the most willing chemical partners. Still, the researchers concluded that such a process would let a planet make its own water; they just weren’t sure how much.

Two years later, they were thrown a lifeline by an ambitious set of experiments built by the researchers Harrison Horn, a physicist at Lawrence Livermore National Laboratory; S.-H. Dan Shim, a geophysicist at Arizona State University; and others.

Among other things, they wanted to know how sub-Neptunes, commonplace exoplanets two to four times the diameter of Earth, can have atmospheres rich in water, as telescopic observations suggest, even when they hew close to their scorching-hot host stars. Could a reaction between a hydrogen atmosphere and a magma ocean be enough?

They suspected it could, but only if a huge amount of hydrogen put the magma under a sufficient amount of pressure. “That higher pressure is a big part of what facilitates the water production,” Horn said. “It actually enhances the chemical reactions.”

To test their model, the team wanted to re-create the extreme (and extremely dangerous) conditions present on adolescent sub-Neptunes. They needed to put hydrogen, a highly flammable gas, under intense pressure using special tools called diamond anvils, and then combine it with rock samples melted with lasers. It took them five years to develop the techniques they needed to conduct these experiments safely and effectively. “We broke a lot of diamonds,” Shim said. “It was an exciting journey.”

They had hoped the hydrogen and oxygen would react to make water. And that’s what happened, to the extreme: The reaction of high-pressure hydrogen and laser-melted rock was so efficient that it made up to 1,000 times more water than scientists predicted. (A second laboratory study, published around the same time, reported similar results.) “It doesn’t seem unreasonable [that you could] produce a huge amount of water quite quickly,” said Paul Byrne, a planetary scientist at Washington University in St. Louis. And “this is all homegrown, indigenous water”— no comets or asteroids required.

Does that mean Earth created its own oceans? This is where the waters get a little murky. “The paper doesn’t make strong claims about Earth,” Horn said. But both he and Shim think it’s a valid link to make. “It could happen,” Shim said.

Other scientists agree that some amount of water could have formed on Earth — but perhaps not nearly enough to produce its oceans. “I’d say it’s certainly possible that some water could be generated by reaction with hydrogen early on,” said Quentin Williams, an experimental geophysicist at the University of California, Santa Cruz. “How much might be generated is, however, pretty enigmatic.”

The issue is that nobody knows if there was enough hydrogen in Earth’s early atmosphere to create the pressure the reaction seems to need. Sub-Neptunes are far more massive than Earth, and their intense gravity is better at holding on to hydrogen. “Earth is right on the edge of where that kind of thing can start happening,” Horn said.

Some scientists don’t think Earth had the heft to manufacture its own water at scale. “I’m a little bit doubtful whether you can have this for an Earth-mass planet,” said Anders Johansen, an astrophysicist at the University of Copenhagen in Denmark and at Lund University in Sweden. But, Byrne said, the sub-Neptune experiments suggest that the reaction wouldn’t need to last long to create an enormous quantity of water. Earth might have been a water factory for only a moment, but that moment may have been enough to forge oceans.

If that’s the case, then the implications stretch far beyond our own solar system. Perhaps countless planets meet what may be a necessary condition for hosting life because, as Byrne said, they “are born water-rich.”

Drowning in a Sea of Possibilities

It’s possible that at least some of Earth’s water came from processes on the planet. But that’s not the end of the story: Comets are making a comeback.

By the time Rosetta met the duck-shaped comet 67P in 2014, scientists had studied the water of 11 other comets. All had D/H ratios unlike Earth’s — except one. In 2011, ESA’s Herschel Space Observatory found that the water in the comet Hartley 2 had a much more Earth-like signature.

Comet Hartley 2, studied by the Herschel Space Observatory, has an Earth-like ratio of heavy to normal water.

NASA/JPL-Caltech/UMD

The observation seemed anomalous. But in a paper published in 2024, scientists reinvestigated Rosetta’s analysis of comet 67P — and found that space may have tampered with the data. At the time of the sample collection, Rosetta had been flying through dust that contained heavy water. The icy comet body itself, surrounded by this dust, may have been fairly Earth-like after all. Then, in a paper published in 2025, observations of comet 12P/Pons-Brooks detected a D/H ratio much like that of Earth’s oceans.

So, was it comets all along? Or asteroids? Or did Earth turn on the taps independently? “I suspect it was a combination of all of them,” Meech said. Perhaps it’s hard to find a perfect extraterrestrial match for our planet’s watery chemistry because it’s such a diverse cocktail — one that’s also been tweaked and filtered over time by Earth’s geologic, atmospheric, and biologic processes. “We may never know,” Meech said.

But that doesn’t mean scientists will stop trying to answer one of the most fundamental questions about Earth and, ultimately, our own existence. “It’s like asking, what’s the origin of life?” Meech said. “The more you learn, the less you know — but the story becomes richer and more exciting.”

What’s the Future of Gene Editing?

Janna Levin — Thu, 11 Jun 2026 01:37:21 GMT

Chanelle Nibbelink for Quanta Magazine

One of the most surprising and remarkable discoveries in recent scientific history has been CRISPR. Short for Clustered Regularly Interspaced Short Palindromic Repeats, CRISPR is a form of immune system that evolved in bacteria more than a billion years ago to defend against persistent viral threats. Under attack, bacteria can snip a small fragment of a virus’s DNA, store it in the CRISPR region of their genome, and then use it to recognize and destroy the same virus if it returns. The CRISPR-Cas9 system, to give it its longer name, consists of a short strand of guide RNA that identifies where to cut the DNA and a protein that acts as the molecular scissors.

What made this system truly revolutionary was the demonstration in 2012 that it could be reprogrammed with different pieces of guide RNA to edit virtually any genome in any species, and at a level of precision and ease that far surpassed existing gene-editing tools. Since then, the editing capability of CRISPR has been tested on everything from developing disease treatments to engineering drought-resistant crops to resurrecting genes of extinct species. The possibilities have expanded so rapidly that researchers, ethicists, and regulators have found themselves struggling to keep up.

One person acutely aware of the power of CRISPR is Jennifer Doudna, co-developer of the technology. Doudna, who received the Nobel Prize in Chemistry in 2020 with Emmanuelle Charpentier for this pioneering work, has been a prominent voice not only for its vast potential but also for its responsible and ethical use. In this episode of The Joy of Why, Doudna tells co-host Janna Levin how her early, “rebellious,” decision to study RNA led her on a serendipitous path to one of biology’s most transformative discoveries. They also discuss the breakthroughs, barriers, and frontiers that will define CRISPR’s true impact.

Listen on Apple Podcasts, Spotify, TuneIn or your favorite podcasting app, or you can stream it from Quanta.

Transcript

[Music plays]

JANNA LEVIN: Okay, here we go. I’m Janna Levin.

STEVE STROGATZ: And I’m Steve Strogatz.

LEVIN: And this is The Joy of Why.

STROGATZ: A podcast from Quanta Magazine, where we discuss some of the biggest unanswered questions in math and science today.

LEVIN: Hi, Steve. Here we are.

STROGATZ: Hi, Janna. It’s a new season.

LEVIN: I know, this is fun. Season Five. I’m pretty excited to talk to you about CRISPR today. Have we ever had this conversation?

STROGATZ: No, we have not.

LEVIN: Do you remember first learning about the CRISPR mechanism for gene editing?

STROGATZ: Well, I, I have heard of CRISPR, but I barely know anything about it. Should I think of it as some kind of molecular scissors that can do chopping of bacterial DNA by the bacterium itself?

LEVIN: Yeah, gosh, now you’re going to be challenging me, but yes. CRISPR, it’s a mechanism that can chop the DNA and then insert it. So it’s a combination of a cut and a paste.

STROGATZ: Aha!

LEVIN: And, I so distinctly remember hearing someone describe to me that there was a naturally occurring mechanism in bacteria which indicated they could edit their own genome and splice in the DNA of an invading virus, for instance, and store it for later so that it was more effective as an immune system if that same virus attacks.

STROGATZ: It’s a really cool idea in itself. I mean, aside from any applications it might have, I think I remember from high school biology that bacteria don’t have any immune system.

LEVIN: Yeah. I mean, pretty simple organism. I think about this also, if you just imagine its molecules acting as prescribed, right? Just moving around. When you hear it from this perspective, it sounds like a stroke of genius. Yet, there’s really nobody doing the thinking. It’s just molecules responding.

STROGATZ: You know, thank you for saying that. Because it’s so easy when you hear biologists talk about this or that mechanism. It’s good to remember there’s nobody home. This is, this is molecules.

LEVIN: Yeah. It’s just a little bit of positive charge making it move a little bit towards this. It is incredible that through this kind of iterative steps of just very simple application of basically electrical attraction that something this sophisticated could emerge, and is essential to the survival of an organism, and even the definition of an organism. Absolutely fascinating. Of course, this is our history. We come from very simple organisms ultimately all the way back down, and yet we don’t have a CRISPR mechanism.

So, let me tell you about our guest. Jennifer Doudna is a professor of biochemistry, biophysics, and structural biology. She shared the 2020 Nobel Prize for her incredible pioneering work on CRISPR. First time two women have won the Nobel Prize together, I believe, by the way. She’s at UC Berkeley, and she leads the Innovative Genomics Institute and does so much beyond that.

And I’ve been wanting to speak to her for many years because I find the work so fascinating, and she’s an incredibly productive and prolific scientist. Really incredible person.

If you will, here is Jennifer Doudna.

[Music plays]

LEVIN: Welcome to The Joy of Why, Jennifer, it’s such a pleasure to have you.

JENNIFER DOUDNA: I’m delighted to be here, Janna.

LEVIN: Thanks for joining us. I was saying I’m a big fan of your work. I’ve been following it for years as much as I can. For you, you have had such an unbelievably storied career. You’re so accomplished, incredibly productive, both in academia and outside in industry. You literally discovered a means to rewrite the code of life. It’s a discovery of almost unfathomable ramifications. I think everyone wants to know the secret of your success, or at least what drew you to your subject? How did you know you were such a natural match for this subject?

DOUDNA: Well, Janna, I’ll start by saying I certainly did not know that I was a natural match for my subject area. I happened to be growing up on a rural island in Hawaii. I got fascinated by chemistry in high school. I had a great chemistry teacher and I was amazed at the variety of life I saw on the island there. And I guess I put all of that together and said, “I wanna understand the chemistry of all of that, of how life evolves.”

And I didn’t really know about that from a chemical perspective until I read The Double Helix by James Watson. And I think it was the realization that science is a process of discovery. It’s not about memorizing a bunch of facts. It’s about figuring things out. I remember thinking clearly when I was in high school that it would be a really fun career to be paid to figure things out, and I think that’s what I’ve always pursued.

LEVIN: Now The Double Helix is a fascinating story of discovery. It really is a great classic book. How they go from chemistry to life is so exciting. I feel that a lot when I read about your work. It’s really descriptions of molecules and bonds and enzymes and protein folding. How you get from there to life just seems to still be tremendously elusive. And is that still a huge source of curiosity for you?

DOUDNA: It is a source of curiosity. My line of work is biochemistry. We’ve always worked with purified molecules and tried to figure out how they function, what they’re doing inside of cells. But taking knowledge like that and trying to weave it into a story that explains evolution or even just explains biology as we experience it in our own bodies and in our world. That’s a big stretch. So, we’re still working on that one.

LEVIN: Yeah, fascinating. When you were doing your PhD work, it was in the mid-’80s, and your work went on around the time the Human Genome Project started to become a really viable possibility and was grabbing a lot of people’s attention… all of this work on DNA. But you took the direction to study RNA. And you’ve even described that yourself as a kind of bold and risky move. Why was it that you were motivated to move away from where the crowd was going and to look at RNA instead, even knowing that it was risky?

DOUDNA: Well, it felt just a little bit rebellious, I guess. If I’m honest, that’s part of the reason. But I guess I feel that when you go to graduate school and I was, you know, very young, I was in my early 20s, I didn’t know anything and I had the really good fortune to work with an amazing mentor, Jack Szostak, who was a yeast geneticist.

So he studied how chromosomes divide in yeast cells. Sounds kind of esoteric, but actually a lot of fundamental discoveries were made from that system that ended up relating to things like how human chromosomes go awry and give rise to cancer. So that’s been an interesting line of work for sure.

However, when I arrived in Szostak’s lab, he said, “Actually, I’m changing my field of research because I’ve gotten very interested in evolution and specifically in the origin of life.”

LEVIN: Exactly the question.

DOUDNA: And I thought, wow, I can’t think of a bigger question than that. And not only that, but he had a very specific experimental path to discovery there. He was curious about how RNA molecules might have, in fact, given rise to modern life by preexisting DNA being around on our planet before there was DNA, and that perhaps RNA could have played an original role as a self-replicating form of genetic material.

So, you know, I didn’t know, again, anything about that, but it certainly sounded amazing, and that’s how I got into the field in the first place, was really through his encouragement and my ability to jump onto a seemingly kind of rebellious project at a time when nobody else, for the most part, was working in that space.

LEVIN: Yeah, RNA was highly underrated at the time. I mean, here he proposed this suggestion that seems quite grand, but that wasn’t really a popular thought about RNA at the time, was it? I mean, RNA was kind of underrated.

DOUDNA: Well, to be fair, there were a few visionaries who were absolutely thinking about that. Tom Cech is one of them.

He, with Sid Altman, won the Nobel Prize in 1989 for their discovery of catalytic RNA, RNA that could function like an enzyme. And then there were quite an interesting collection of people who were also very interested in questions about the origin of life and were investigating curious examples of RNA molecules that have either catalytic properties, they can function like enzymes, or seem to play other very interesting roles in biology. For example, serving as the genetic material of viruses. For me, it was really those colleagues, my superiors really, but it was that whole generation of scientists who were interested in these questions that were not really in the mainstream at the time, who had a huge influence on me.

And in particular, a scientific conference that I went to when I was a second-year graduate student in 1987, where I had the chance to see a number of those folks giving lectures and meeting them for the first time, hugely influential on my decisions for the future of my career.

LEVIN: And how did it play out? Does RNA have that role of possibly preceding DNA in the emergence of life in evolution? Is that a question we can answer?

DOUDNA: Well, it’s hard to answer it definitively because unless we build a time machine, we can’t really go back and check, you know? But I think what’s fascinating is that over the years, I think there’s only been increasing evidence that that theory is probably correct, or at least that’s an important piece of the story of evolution on the planet.

Where RNA came from in the beginning is still debated. You know, did it arise here on the planet, or did it come from somewhere else in the universe and arrive on our planet as a seed? People still debate that kind of thing. It’s an interesting speculation, but there’s a lot of evidence that RNA was probably the first kind of self-replicating biological molecule that gave rise to life on the planet.

LEVIN: Hmm, I mean, the idea of a panspermia is fascinating, but it also just sort of kicks the question down the road. It emerged somewhere. But it is a fascinating possibility. So here you are, you’re trying to understand these deep questions. How does that lead you to the CRISPR story?

DOUDNA: It was a circuitous route, if I’m honest, and this has really been my experience more generally in science, is that I think you start off in one direction, and if you are open to interesting ideas and results that come up along the way, the path is never straight.

In my case, that was through a process where initially we investigated catalytic RNAs and, in particular, understanding their molecular structures to try to find out how they could actually function in an enzymatic way, which was a very interesting question. Still is, frankly. And then we started to look into how RNA molecules control the way that gene expression works, and that simply means control the levels of proteins that are made in different kinds of cells.

It turns out that that’s something that is very fundamental to all of life. It probably influences not only organismal behaviors, but also the way that certain tissues form, the way that viruses function, of course. Fascinating aspects of gene regulation that really boil down to understanding the levels of proteins that are made at any given time.

There’s a lot of evidence that RNA molecules in different ways are a very important part of that story. They help control those expression levels of genes. And so we were investigating this in viruses and in different types of cells. And at that point I had started my career at Yale. I moved my lab to UC Berkeley in 2002. And I was fascinated to make the acquaintance of Jill Banfield here at Berkeley, who had discovered evidence at a computational level of an adaptive RNA-guided immune system in bacteria. So this was, for me, yet another fascinating example of RNA molecules controlling the expression of genes, and we wondered, how does that work?

And that was really my entrée into the CRISPR system and all of the CRISPR biology that came from that.

LEVIN: Wow. Now CRISPR is an absolutely fascinating, I guess I would say mechanism. How would you, how would you best describe CRISPR? I mean, maybe it would be fair to the uninitiated to tell us what the acronym stands for.

DOUDNA: Uh, let’s see if I can pull it out. Clusters of Regularly Interspaced Short Palindromic Repeats. Ooh! Don’t ask me to do that again! Yeah, it’s a bit of a mouthful!

LEVIN: I had a little cheat sheet somewhere if I had to look it up. It’s part of the genome of bacteria, is that right?

DOUDNA: That’s right. It’s in the genome of bacteria, and it’s a very special part of the genome because it actually allows bacteria to create a genetic vaccination card.

They capture little pieces of DNA from viruses and insert them into this special place in the genome called the CRISPR locus that stores that information from viruses over time. It, it makes an amazing, you know, recording really in real time of infections that are happening. And not only that, it’s not sort of dead information, it’s actually information that gets reused in the form of RNA molecules that are produced from those little templates in the DNA to make molecules of RNA that can go out and search for matching sequences in DNA.

When those matches are found, they recruit proteins that can come in and snip the viral DNA and protect the cell.

LEVIN: That’s just unbelievable actually, right? So the RNA is playing a really active role going out and then annihilating the virus that it might previously have contracted. But it’s hard to imagine that all of this is just molecules interacting electromagnetically. It’s, it really is such a sophisticated mechanism. I think one of the interesting questions is, why did humans not develop this amazing vaccination system?

DOUDNA: Well, it’s hard to say, you know, why something doesn’t exist, or at least as far as we know. But I guess what I would say is that humans have other ways of defending against viruses that are in some ways more advanced, in the sense that they are protein-based and they allow very sophisticated defenses against viruses that themselves have clever ways of trying to avoid immunity.

And I think what we see with CRISPR systems is that because they’re based on direct recognition of a viral DNA sequence, it means that viruses can avoid being detected by simply mutating their DNA sequences. And this is probably one of the reasons why we see so many different types of CRISPR systems in biology. There’s a lot of active evolution of those systems going on over time.

LEVIN: They have to keep ahead.

DOUDNA: They have to keep ahead. Right? Yeah. And so I think when you have rapidly growing cells, like bacteria that are reproducing on a scale that is very similar to the rate at which viruses are reproducing, that kind of works. But when you have viruses that reproduce much faster than the cells they’re infecting, like in us, I suspect that kind of a mechanism just can’t keep up if it’s a CRISPR system. And so we evolved other ways of being able to defend against viruses that avoid those immediate escape mechanisms in viruses.

LEVIN: Now, because the CRISPR mechanism also involves cutting the DNA of the host, it introduces the potential to damage the host as well. And so how does a repair mechanism get involved to make sure that it’s not a more damaging system than it is protective?

DOUDNA: Well, in bacteria, of course, that’s kind of the point, right? The cutting is the way that the immune system functions, so it helps the cell to find and then cut up viral DNA sequences. But what’s very interesting is that it turns out that in animal and plant cells, these cells respond to DNA cutting differently. They detect cuts in DNA, and they tend to try to fix them, and they can fix them because they have time, and that’s, again, because the cells are dividing much slower than if the cell is a bacterial cell.

And as a result, when there’s an insult to the DNA, like, say, a double-stranded break that gets introduced, for example, by CRISPR, cells can find the break and fix it. And when they fix it, as you just said, that’s an opportunity to also introduce a change to the DNA sequence, and that’s fundamentally how CRISPR works to induce gene editing.

LEVIN: Now, here you’re studying this esoteric mechanism in bacteria, might be relevant for evolution. Clearly really fascinating. But then there’s a big step forward, which is to contemplate how you might alter this mechanism to allow editing for, for the human genome. Was that something you intentionally sought out, or was it kind of an accidental realization that this was possible?

DOUDNA: Well, it certainly wasn’t something that was the motivation for the project in the beginning. The project was designed to ask and answer a question about how bacterial adaptive immunity was operating.

However, as soon as we understood the chemistry of that RNA-guided DNA-cutting activity of a protein known as Cas9. It was, you know, an amazing example of how when you do fundamental research, it leads in unexpected directions. That understanding of the chemistry of RNA-guided DNA cutting immediately suggested a very interesting application of that activity, namely to induce precision editing in cells like ours, or like plants and animal cells, that have this capacity to repair double-stranded DNA breaks.

LEVIN: You mentioned the Cas protein. What was so important about the Cas protein specifically. Proteins abound in these systems. So what was so important? Why is it often paired, CRISPR-Cas9?

DOUDNA: Well, it turns out to be the real engine of gene editing, and the reason is that it’s the enzyme that does the DNA cutting. It uses the RNA molecule that comes from the CRISPR sequence as the zip code. It’s the molecular guide that tells that protein where to go and where to cut DNA.

But Cas9 is the actual machine that does the cutting. And so you really need both together and the two together provide a very powerful tool for programmable gene editing in different kinds of cells.

LEVIN: Once you’re editing genes, you immediately realize that you have the potential to radically alter life on Earth, to participate in the process of evolution. But there were other gene-editing tools also at the time. What was so special about this gene-editing tool that really made it transcendent and ubiquitous in a way that the other gene-editing tools didn’t really take on?

DOUDNA: Well, you bring up an important point because you’re right, that there had been a fairly long-standing effort among molecular biologists to figure out how to manipulate genes in a precise way.

There were a whole series of discoveries that were made that were instrumental to that capability. Partly it was the understanding of how double-stranded DNA-break repair works in cells, and the other was figuring out how to introduce a double-stranded DNA break in the first place, especially at a place that you might want to induce a gene-editing event.

And so because that knowledge was preexisting, I think it created a very nice path for CRISPR because what CRISPR offers is an easy way to generate double-stranded breaks. And not only that, back to the role of this Cas9 protein, what’s really interesting and kind of crazy about the CRISPR technology is that we can use exactly the same protein to manipulate genes in wheat, rice, human liver cells, the brain, you name it, right? It’s the same enzyme. And the reason that works is because we can simply change the guide RNA that tells it where to go, and we can redirect its activity to a gene of interest in any cell type.

Because of that, it just makes it a very easy technology to deploy, and that’s really what we saw in the field. As soon as that original article with my collaborator Emmanuelle Charpentier was published in the summer of 2012, immediately there were many labs that started using it and testing it for gene editing in different systems. And that set off an enormous race, and then of course a trajectory of many labs adopting the technology for all kinds of applications.

LEVIN: I mean, this is the discovery of a lifetime. I mean, it really is. You described, in response to receiving the Nobel Prize with Emmanuelle Charpentier, that this was a joyous time of discovery, as though it was singular, as it stood out. And I guess I’m wondering, would you describe it as a moment of a realization or was it more the process of the discovery?

DOUDNA: Well, it wasn’t instantaneous, but it was pretty fast actually. Because, you know, and that, that’s sort of been my experience in science over the years is that, you know, when you discover something that is of real import you kind of know it right away in a sense.

With CRISPR it’s not as though we could foresee everything that was to come, of course, from the technology. But we could really pretty immediately see how this could be a very powerful tool because of the ease of deployment, how easy it was to alter this RNA molecule and send Cas9 to different places in a genome. And all of the potential uses of that kind of technology. It was just very exciting to think about and contemplate and imagine what could be possible.

LEVIN: So, has the technology changed significantly? And what do you think the most impactful technological advances have been since its discovery?

DOUDNA: Well, since the discovery of CRISPR, what’s happened is that it’s become a whole toolbox, and the way that’s happened is that it’s been possible to take advantage of, again, the fundamental chemistry of the way the CRISPR system functions as an RNA-guided mechanism of recognizing and cutting DNA.

It’s been possible to change that into a mechanism of recognizing and changing DNA in different ways. And so that’s really made it an incredibly versatile technology that can now be used for all kinds of different types of genetic manipulations. And I think that I’m just excited about all of those, to be honest, because I think that it gives scientists a very rich set of technologies that can be deployed as they’re needed in different settings, and it’s only gonna continue. I mean, every time I go to a meeting about CRISPR, I’m continually blown away by you know, by that expansion of the toolbox. And so I just think it continues to get better and better and better.

[Music plays]

STROGATZ: Wow, this is making me try to remember some of my biology classes because, for instance, the phrase double-stranded break. I’m not sure I fully appreciate what’s going on here. So let’s just remember, maybe you can correct me if I’m getting this wrong.

DNA is a double helix, we all learned that. It has these two strands, and you can break one strand and leave the other strand intact. There are enzymes that do single-stranded breaks, and that’s not super dangerous from the point of view of the integrity of the DNA molecule or the gene because you’ve still got one intact strand. There’s still all the base pairing along the whole double-stranded structure. You put a snip in one strand, but you haven’t broken the back of the molecule. A double-stranded break is literally chopping the molecule, DNA in half. Really very dramatic move.

LEVIN: Right. In principle, it should be very damaging to the cell.

STROGATZ: Yeah, and so to be able to have access to genetic machinery that can not only do these double-stranded breaks, but do it in a manageable way, and this is the part that got me, it’s like it’s a sort of universal scissors. It can work in any organism.

LEVIN: Yeah.

STROGATZ: And you can just guide it to any place.

LEVIN: Yeah, it’s insane

STROGATZ: Right? Like, in the old days, there were enzymes that they’re good at snipping one strand, but only if the sequence was such and such, you know, like much more restricted kinds of scissors. This is like a really magnificent all-purpose gadget.

LEVIN: I think she really says it well when she says, “It was just so easy to deploy.” And you saw it right away in use in other labs immediately. There was very little barrier to its application. I think this point about the double-strand breaking is a single-strand breaking, as I understand, is more easily repaired. Yeah. And you can, but you don’t, in principle, change the DNA. But if you double break, you can now insert new base pairs.

STROGATZ: Okay.

LEVIN: And that’s really what CRISPR is doing. It’s, for instance, taking the DNA from an invading virus. It’s cutting its own DNA and putting the viral DNA in its own strands. It’s inserting the base pairs, and you need the double break to do that. And the reason why that’s interesting is you’ve essentially made an immunization card, a record of your own ability to immunize against that invader. At least that’s the case for bacteria.

So here now we can adapt this from the bacterial toolkit and implement it in human beings, and fundamentally change the genetic material.

STROGATZ: It’s incredible.

LEVIN: It’s pretty incredible.

STROGATZ: It’s, it’s not the biology I ever learned, and I guess the real experts are just as shocked, right? It was a really monumental discovery.

LEVIN: I’ve got to say, this sort of excitement over CRISPR, I think, is among the most fascinating scientific discoveries that I’ve ever heard of, and it has the potential to change fundamentally the human blueprint.

STROGATZ: It’s just astonishing now what’s possible. But it sounds like it’s been discovered in the laboratory, tested in the laboratory. Is it making its way to the bedside, to the clinic? Is it helping real people?

LEVIN: Yeah, exactly. Jennifer discussed cases, real patients, living human beings who are alive precisely because of CRISPR therapies. So, we’re gonna get right into that after the break.

[Music plays]

LEVIN: Welcome back to The Joy of Why. We’ve got biochemist Jennifer Doudna with us here today to discuss CRISPR and the future of gene editing.

It’s not quite 20 years, but we are living in the time where there’s these really impactful technological advances. You have this work with Baby KJ. Why don’t we talk about Baby KJ? Maybe you could tell us, it’s a very concrete example of what’s actually being done therapeutically.

DOUDNA: Well, Baby KJ was born in August of 2024 and he had a rare metabolic disease that was diagnosed right away after he was born. He couldn’t digest protein properly, meaning that he was extremely sick. He couldn’t eat a normal diet. He wasn’t gaining weight. He was in the neonatal intensive care unit. You can imagine that his parents were distraught and, you know, desperate to do something to help their boy.

Fortunately, his clinical team at the Children’s Hospital of Philadelphia realized that he probably had a rare genetic disorder and were able to quickly get a sample and sequence the DNA. They figured out that this boy had mutations in both copies of a gene encoding an essential enzyme required for protein digestion.

And not only that, they realized that this was a type of mutation that could in principle be fixed using a version of CRISPR that would have that capability. And so they reached out to a number of groups, including the Innovative Genomics Institute out here in California, about helping them to create a version of CRISPR that could treat this boy. And incredibly, incredibly, I still can’t really believe it, but it did happen in an eight-month time period.

LEVIN: That’s incredible.

DOUDNA: And the baby was treated and today he seems to be thriving, which is absolutely wonderful. So, you know, it’s just an extraordinary story of teamwork. It’s an extraordinary story of using off-the-shelf technology. No new research had to be done. We could use existing versions of CRISPR and a delivery tool that had been developed originally for the COVID vaccine, actually. And using that in the patient, it was possible to create a therapy I don’t know if anyone’s ever created a therapy that quickly tested it and delivered it to a patient. But now we know it can happen, which is really exciting.

LEVIN: It’s fascinating. I have so many questions, but when you deliver this kind of a therapy, since it’s a gene editing therapy, is it a one time you deliver it? The genome is edited? Or is it a therapy that has to be re-administer over time?

DOUDNA: Well, in this case it was a little of both in the sense that it was three times into the patient, but not since then. And I think the hope is that sufficient editing of that patient’s cells in his liver, that are essentially repopulating his liver over time, have been edited such that he now has a normal functioning liver that’s producing the kind of digestive enzymes that are needed for his health. That will just have to be monitored, of course, over time. And because it’s a one patient situation, we don’t have any way of actually testing whether and how much editing occurred in his liver. It’s just, looking at his physiological properties now and trying to assess what his health is.

But it is quite impressive that it took just this, you know, kind-of very succinct delivery. It doesn’t require treating the patient every day or every month. He had three treatments with this therapy, and we hope that that’s sufficient to give him a normal lifespan with a normal outlook.

LEVIN: It’s incredible. I mean, there are other areas in terms of human health therapeutics where you would see this kind of possibility, cardiovascular disease or, I dunno, altering the microbiome. Where do you see the most sort-of productive direction for thinking about gene therapies?

DOUDNA: Well, you just mentioned two big ones that we think about a lot. So I think, you know, the cardiovascular angle is fascinating. It might not be obvious to someone listening to this, you know, why would CRISPR be useful for treating heart disease? And yet it is. And the reason is that many studies have shown that people that have a particular form of an enzyme in the liver that processes cholesterol differently than others have protection against cardiovascular disease because they don’t tend to accumulate plaques in their arteries over time. So, wouldn’t it be great if you could actually use CRISPR to give everybody that form of the gene? And that’s what the principle is for using CRISPR in that fashion.

And in fact, there was a company that was founded to do this, a company called Verve that has demonstrated enough potential for this approach that they were actually purchased by Eli Lilly last year. And so, you know, there’s a lot of interest on the part of even big pharmaceutical companies in pursuing a strategy that could give people an option that doesn’t involve taking a daily pill or getting, you know, frequent injections or something, or having to radically change their diet, but instead having a one-and-done therapy that just gives them a genetic fix to the problem of high cholesterol.

LEVIN: Now, this is the upside, the success stories. But there are also barriers to developing these treatments at scale. What are the barriers? Are they all just financial barriers or getting FDA approval, or are there actual barriers to scaling up these kinds of treatments?

DOUDNA: Well, certainly the financial and barriers are there. What’s exciting about the case of Baby KJ in particular is that those barriers were overcome and that sort-of speaks to what’s possible. On the flip side, we know that that strategy isn’t going to work for everybody. It’s very hard to scale that. For example, how would we replicate that particular path for other patients that have rare diseases?

So, I think it’s worth really for the field to think about what are the approaches that could just radically reduce the cost and make it a lot easier for other patients to get access to this type of a therapeutic. And so I think it’ll take not only getting creative with engineering and the way that these molecules are manufactured, and that’s already underway to try to reduce costs there. But it also goes back to the science and the technology. For example, you know, Baby KJ was lucky that his disease affected his liver, so it was possible to use an off-the-shelf delivery technology to introduce the CRISPR molecules to cells in his liver. But that’s not going to be helpful for people that have a lung disorder or a muscle disease or a brain disorder.

And so, one of the real forefronts in the field right now is figuring out how to solve the delivery problem for all these other tissue types. I think it’s gonna be solvable. You know, I’m very bullish on this. But it’s gonna take real work. I mean, you know, it’s not going to just happen. I think we have to really focus on it.

And fortunately, many people recognize that this is an important challenge. And so we’re seeing more and more efforts in this regard. A lot of our young students that come in to the Innovative Genomics Institute here in California certainly are very motivated by this. They’re excited about it. It’s a hard problem. They wanna grapple with it. They wanna figure it out. So, that’s the kind of energy and innovation that I think will solve a hard problem like delivery.

LEVIN: If you have a newborn baby with a terrible genetic disease, who is not going to have a long life prognosis, you can imagine risking anything to treat this child. But for somebody who has alternative therapies, how scared are people of doing something as radical as editing their genome? And are there negative consequences? Are there possibilities of having mistakes in what’s edited, or how it’s edited, or immune response?

DOUDNA: Yeah. Well, you know, with any technology of course there’s always risk, right? And with gene editing in particular, you could imagine, right, you don’t wanna have something that’s not accurate, or editing sites that are unintended or even that are harmful. You certainly don’t want to be in a situation where you have consequences of editing that lead to undesired outcomes.

I think a good case in point is actually the situation with sickle cell disease, because that’s a disease where the presence of the sickle cell mutation in the human population is probably in part because it gives some protection against malaria infection. So, people that have one copy of the so-called sickle cell gene, phenotypically they’re normal, but they have some protection against malaria infection. So you could argue that for them a bit of an advantage to have that in parts of the world where malaria is endemic.

And so that’s just kind of a good reminder that our genetics are complex and genes, you know, aren’t necessarily good or bad. They could be a little of both depending on the situation. So I think, gene editing just, we have to employ it cautiously because it does require a lot of knowledge about what effect a genetic change is going to have on a person over the course of their life.

LEVIN: Yeah, you raised this fascinating possibility that a gene we think is simply harmful, actually has a protective purpose. And we talk a lot about, you know, maybe myopic people also have some correlation with abilities, or you can’t fix one thing without possibly damaging another. I think that’s just generally true about human beings.

It also leads to some ethical questions, and I know you’ve thought a lot about the ethical questions. I feel like we have to talk about the somewhat shocking case of the Chinese scientist in 2018 who used CRISPR to genetically alter human embryos, which seemed like it was really crossing a line, resulting in the birth of two twin girls. It’s my understanding that he was trying to make the babies resistant to HIV. Now, when you heard about that were you shocked, or did you feel it was inevitable that somebody would transgress across this unspoken ethical line?

DOUDNA: It was shocking. No, it was definitely shocking. It had already been on my mind though that, you know, this was certainly a possibility. And this particular individual had been going around and attending meetings on gene editing, so he wasn’t unknown to the genome editing community.

But that all being said it was certainly shocking to find out that this wasn’t just chat, it was actual action that he had taken. And once the details were revealed, it was pretty clear it was an extraordinarily unethical thing to have done for multiple reasons.

You know, if there’s a silver lining to that story, it’s that I think internationally people recognized right away that this was wrong and they took a stand against it. And in fact, that scientist was arrested and his lab was closed and he was jailed for a few years. So, we’ll see what happens going forward, but I was pleased, I guess, that there was a very strong and concerted response by the international community about his action.

LEVIN: And the big issue being that it was in embryos, it was editing the germline. So in other words, it could be passed down. Was that the big line that was crossed versus another therapy?

DOUDNA: Well, for me, even more than that was that it was first of all used in a way that was medically unnecessary because there were other proven ways to protect those babies from transmission of HIV during their development and birth.

Secondly, I don’t think from the evidence that I saw that the parents were aware particularly of what they were actually agreeing to, which is also very shocking. And then as you mentioned, the third piece is that this is a permanent technology. And not only that, when you perform it in embryos, you are making changes that are heritable. So those changes will now be passed to future generations, and we really don’t know what the impact of that will be.

LEVIN: Do you think that there are others that are unscrupulously performing these kinds of experiments? As you said, it’s an incredibly flexible, programmable, swift, and not terribly expensive technology, which also makes it kind of scary.

DOUDNA: You know, it’s possible. I think that my assessment is that the original perpetrator, I guess you could say of that germline application of CRISPR, a lot of his motivation I think was frankly for publicity. And so I think that part of the deterrent now is the idea that publicity would be pretty negative for somebody forging ahead with something like that today.

And yet, you know, we do know of companies, for example, I’ve heard of a few around the country that are exploring again the possibility of germline editing and offering that as a service to people. So, it’s not as though this is off the table or no one’s thinking about it anymore. I think it’s still very much in the milieu and we’ll have to see what happens in the future. But it, to me, just underscores the continued importance of public engagement, of scientists being involved in the conversation around CRISPR and how it should be used.

LEVIN: There are implications for using this technology for climate, for plant life. I mean possibly even food, to address food scarcity, or diseases like malaria, where you stop it at the level of the insect, not at the human-body level. How do you see those advances progressing? Is that an area that’s very active at the moment?

DOUDNA: Yeah, it’s pretty active and I think there’s a real upswing in the applications of CRISPR for those kinds of things that we’re seeing right now. Especially for addressing challenges that are coming with the changing climate, both in terms of food security, how we ensure that we have plants that are robust with respect to drought, with respect to pests, that have improved nutritional value. All of those things are interesting applications using CRISPR.

And then the other is thinking directly about carbon release and applications that involve changing the microbiome in cattle to avoid the emission of methane. Cattle are one of the major sources of methane emissions around the world every year, and CRISPR in principle can dial that back by changing the genes in those bugs to reduce methane emissions potentially permanently. So that’s, I think, something that I’m very excited about, and is an active program here at the Innovative Genomics Institute.

LEVIN: What do you see as a primary focus in your research in the coming decade? Do you see it as being more focused on industry and application, or back to exploration in the lab?

DOUDNA: Well, it’s a little of both. I think that, you know, in my own research lab, I continue to have folks that are doing fundamental discoveries, and there’s a lot of exciting work, frankly, coming out of that effort right now. And then we also appreciate the value of figuring out this delivery challenge. I think it’s a big challenge.

We’re not engineers in the lab. We love engineers, but I’m certainly not an engineer. But the opportunity to understand fundamentally how cells take up new molecules, how these molecules can access specific kinds of cells. There’s a mechanistic basis for a lot of that, that is something that we do love to dig into in a lab like mine. So those are gonna be two areas that we’re gonna focus in, for sure.

Beyond that, I really want to continue to serve as a mentor. I’m enjoying the fact that at the Institute here, we’ve been able to hire in a number of younger faculty who are kicking off exciting research programs of their own that align with the kind of overall goals and mission of the Institute. These are folks that are here in large part because they love working on big, hard problems. They love doing that collaboratively. They love doing it here in the Bay Area, where we have access to incredible resources of all types.

We love being literally right across the bay from Silicon Valley. You know, as AI continues to advance and accelerate the pace of our work, we’re increasingly integrating that into what we’re doing. So that’s been really fun and I wanna do more of that.

So it’s a really exciting time, I would say.

LEVIN: When you were working on this originally, in the early days when you were making the transition from studying RNA to studying CRISPR to realizing its incredible power in terms of rewriting the code of life, so to speak. When you look back at that time, is there a time that you miss of, you know, before all of this, before the success, the attention, and also seeing the impact it’s having on so many other researchers and so much other work?

DOUDNA: Well, yes, um, my life certainly changed dramatically right around 2012. And, I often joke… my husband’s also a professor here at UC Berkeley… and I often, uh, joke to him that there was my life BC — before CRISPR — and then everything changed.

And you know, do I miss it? Well, yeah, some parts of it I definitely do. I, you know, there’s a joy in just coming into the lab every day and spending time with my students. I try to do as much as that as I can, but, you know, I’m doing things like, this, which is fine, you know, but it’s different.

And yeah, I love science so much. I love the process of discovery. I love working with scientists who are just starting out in their careers, you know, and they’re creative. They’re fearless. They want to figure things out. And it’s, you know, science is always a struggle, right? It’s always hard. And so I do enjoy going through that struggle with them, and I don’t do that as much as I used to. And I do miss it.

LEVIN: All great stories have to have a struggle.

DOUDNA: For sure.

LEVIN: No great book was written without a struggle to drive the plot.

Thank you so much, Jennifer. It really is such a mind-blowing topic. It’s so exciting to see it moving and that it’s happening and it’s actually happening fast. We’re gonna live to see the implications of this. Thank you so much for joining us.

DOUDNA: Thanks for having me, Janna. Great to be here.

[Music plays]

STROGATZ: Wow, I’m hit by so many things as I listen to that. The first is something that I think I heard Stephen Jay Gould, the old evolutionary biologist and writer, say when I was sitting in on a lecture of his one time, which was that it was the age of bacteria, it is the age of bacteria, and it will always be the age of bacteria.

You know, we don’t, see them. We don’t think about them much. But they’re so important, and you can learn so much about life as Jennifer Doudna and her collaborators have done by focusing on bacteria.

LEVIN: Yeah. But it’s also fascinating to talk to someone who’s had such a direct impact on technology, therapies, the potential for improving the human condition, but that’s not really why she got started, and this is something people keep forgetting. It really was just curiosity-driven science, childlike enthusiasm that she maintained her whole life. And how do we convince people that we need to encourage that to have the impact on humanity?

STROGATZ: Hmm. Well, it might help if we could convey the history of science in a way that was as engaging to people as it really is. You know? I mean, throughout the history of science, we hear about these stories of serendipity, where someone discovers something so important, and it’s often described as being by accident.

But it was pointed out in some place that I read that you shouldn’t think of it as exactly by accident. Like in her case, she was really looking for something very focused and thinking about RNA in bacteria. But then she ended up finding something she wasn’t looking for and somehow putting your mind in that state where you’re curious and alert — I mean, it’s that old line about chance favoring only the prepared mind.

LEVIN: Mmm-hmm. Oh, and you definitely get that in her story. “And then I met so and so, and then we talked about this.” So it’s not as though she simply sat down and it was just a matter of time. There is that serendipity. There is that making decisions, choosing to be open to somebody, choosing to have a dialogue on something a little left of center of what you’re working on, and being open to pursuing that with everything that you brought to the table.

STROGATZ: It’s something, too, that I think about as a scientist or a mathematician in the broader collective of our enterprise that will lightning strike for me personally? You know, there’s ego in what we do. And I sometimes have to remind myself that it doesn’t matter if it happens to me as long as it happens to somebody.

LEVIN: Yeah, absolutely. Well, Steve, to be continued. Let’s both get back to our work.

STROGATZ: All right. Get to work, Janna. I’ll see you next time. Bye-bye. [laughs]

LEVIN: Bye.

[Music plays]

STROGATZ: If you’re enjoying The Joy of Why and you’re not already subscribed, hit the subscribe or follow button wherever you’re listening. You can also leave a review for the show. It helps people find this podcast. Find articles, newsletters, videos and more at Quanta Magazine [DOT] org.

LEVIN: The Joy of Why is a podcast from Quanta Magazine, an editorially independent publication supported by the Simons Foundation. Funding decisions by the Simons Foundation have no influence on the selection of topics, guests or other editorial decisions in this podcast or in Quanta Magazine.

The Joy of Why is produced by PRX Productions; the production team is Caitlin Faulds, Jade Abdul-Malik, Genevieve Sponsler, and Merritt Jacob. The Executive Producer of PRX Productions is Jocelyn Gonzales. Edwin Ochoa is our project manager.

From Quanta Magazine, Simon Frantz and Samir Patel provided editorial guidance, with support from Samuel Velasco, Simone Barr, and Michael Kanyongolo. Samir Patel is Quanta’s Editor-in-Chief.

The episode art is by Chanelle Nibbelink and our logo is by Jaki King and Kristina Armitage. Special thanks to Garth Avery at the Cornell Broadcast Studio.

I’m your host, Janna Levin. If you have any questions or comments, please email us at [email protected]. Thanks for listening!

An Early Step on the Long, Strange Road to Photosynthesis

Carrie Arnold — Wed, 10 Jun 2026 02:57:19 GMT

Son of Alan for Quanta Magazine

Every second, trillions of watts of solar energy — more than 10,000 times the energy used by modern humans — blast the Earth’s surface. Around 2.4 billion years ago, life took an evolutionary leap when bacteria learned to harness these photons to break apart water molecules and stitch carbon atoms into sugars. Along the way, they flooded Earth’s atmosphere with oxygen and rewrote the rules of life.

“The oxygen-evolving capability was a big innovation. I sometimes call that a singular event,” said Robert Blankenship, a retired biochemist from Washington University in St. Louis. “By all accounts, it only happened once during the process of evolution, and that really set up the world for becoming oxygenated and the wholly aerobic world that we live in now.”

However, the set of chemical reactions we call photosynthesis has bewitched and befuddled scientists for generations. It requires the coordination of dozens of proteins and hundreds of pigments that harvest photons, all embedded in a cellular structure less than one-thousandth the width of a human hair. Electrons pinball across membranes and between compounds to drive molecular turbines that rebuild air and water into sugars to provide the energy and raw materials that cells need to grow.

We now know this process in fundamental detail; advances in microscopy and cell biology mean that researchers can essentially track a single electron through photosynthetic proteins to illuminate the full molecular mechanism. This level of detail dims, however, as scientists attempt to travel back in time to understand how photosynthesis could possibly have first evolved in single-celled organisms called cyanobacteria over 2 billion years ago.

“It’s now pretty clear that all the photosynthetic [protein] complexes descend from a single common origin,” said Blankenship, who spent his career studying the molecular mechanisms of photosynthesis. “But the nature of that very first organism is not very well understood.”

To solve such riddles, biologists often turn to organisms that share many, but not all, of the traits they want to understand. But for years, they believed that nearly all modern cyanobacteria evolved in a single, closely related cluster, offering little variation that might reveal mechanisms of early photosynthesis. The discovery of Gloeobacteria, a group of photosynthetic bacteria that branched off from other cyanobacteria over 2 billion years ago, changed this. Although Gloeobacteria haven’t remained at an evolutionary standstill — no organism has — they seem to have changed little over billions of years, making them a sort of genetic time capsule.

“[Gloeobacteria] tell us a little bit about what the earliest cyanobacteria might have looked like,” said Christen Grettenberger, a geochemist and microbiologist at the University of California, Davis. “It’s not some weird one-off species. It has a real pattern of retaining these tools.”

The most recently identified Gloeobacteria species, Anthocerotibacter panamensis, harvests light using a different set of proteins than modern cyanobacteria — but converts sunlight into chemical energy within protein complexes that vary only slightly from those in other Gloeobacteria. These traits add new color to the long, strange evolutionary story of photosynthesis.

A Photon Capture Machine

Before striking a plant leaf, a solar photon travels 93 million miles through empty space. The most dynamic part of this journey happens in the last few billionths of a meter, as a Rube Goldberg machine of proteins and pigments converts the photon’s light energy into chemical energy.

Mark Belan/Quanta Magazine

The leaves of modern land plants are packed with chloroplasts, oblong organelles that are themselves stuffed with stacks of coin-shaped compartments known as thylakoids. Thousands of proteins and pigments stud the thylakoid membrane, creating a sprawling biochemical circuit with a single purpose. A large protein complex there, named photosystem II, hosts light-harvesting “antenna” complexes on its outer ring that maximize the number of photons the plant can snag. Chlorophyll and other pigments embedded within the antennae absorb the energy from captured photons. Then, as in a game of hot potato, chlorophyll and other pigment molecules funnel this excess energy to the reaction center of the photosystem.

Energy is lost every time the photon hops between pigments, but it retains enough to jolt electrons loose from nearby water molecules, releasing oxygen as waste. These liberated electrons then flow through a series of membrane-bound proteins, known as an electron transport chain, where their energy pumps protons and spins molecular turbines. This molecular assembly line generates life’s energy currency, a molecule known as adenosine triphosphate (ATP).

But photosynthesis isn’t complete at that stage. Mostly depleted, the electrons then reach photosystem I, where another burst of sunshine kicks the flow back into high gear. Supercharged again, the electrons drive a separate set of reactions that build sugars from carbon dioxide. These sugars are themselves a form of energy, as the plant (as well as other organisms) can break them down to make more ATP.

Over billions of years, the reaction centers of the two photosystems have remained remarkably unchanged. Evolutionary innovation has occurred, however, in the astonishing array of antenna complexes and accessory pigments seen across photosynthetic life. This juxtaposition of severe conservation with extreme diversity creates a challenge for scientists trying to understand how photosynthesis first evolved.

Even the earliest photosynthetic system must have already been a tiny solar-powered electrical circuit, capturing light and channeling electrons into metabolism, said Christopher Gisriel, a biochemist at the University of Wisconsin, Madison. “At a minimum, we know that it would have all the features that we see in the diversity of reaction centers or photosystems today,” he said. “It would have had the ability to somehow collect that light and perform charge separation and then move those electrons into metabolism.”

There is some photosynthetic diversity in living organisms that can provide insight into how this complex process could have evolved. For example, some bacteria use only photosystem I and avoid oxygen altogether in a form of anoxygenic photosynthesis. Many researchers therefore hypothesize that anoxygenic photosynthesis and its photosystem evolved first. Then, at some point in the distant past, the system’s genes were duplicated. Over evolutionary time, this theory goes, one of those copies mutated and evolved to give rise to a second photosystem that uses oxygen to harvest energy more efficiently.

Not all evolutionary biologists agree that photosystem I came first, but if it did, then how and when did it evolve? Even a primitive, stripped-down version would be an evolutionary marvel. The protein megacomplex boasts multiple subunits in a precise structure, where antenna proteins are decorated with chlorophylls and other pigments and surround a core reaction center where electrons twirl in an elaborate pas de deux.

This is why researchers are flocking to an ancient lineage of single-celled cyanobacteria that could hold a more primal version of this process.

A Strange Antenna

The cells arrived as microscopic stowaways. They traveled on the leaf of a hornwort from Panama to the lab of the plant biologist Fay-Wei Li in Ithaca, New York. There, Li’s laboratory technician Duncan Hauser stripped microbial detritus from the hornwort’s leaves. He thought he had done a thorough job, but a chance glimpse of a small green speck under the microscope revealed he had missed something.

That Hauser had discovered a new species of bacterium wasn’t especially exciting; most microbial species have yet to be identified and described. But as Hauser, Li, and then-postdoc Nasim Rahmatpour peered more intently at the interloper, they realized that its color and shape didn’t resemble those of most other cyanobacteria. An analysis of its DNA revealed their contaminant to be a type of Gloeobacterium, a lineage of photosynthetic microbes that branched off from cyanobacteria over 2 billion years ago. They named it Anthocerotibacter panamensis in honor of its Central American birthplace and published their results in July 2021.

Even among the little-known Gloeobacteria, A. panamensis was unusual: It branched off from the rest of the group 1.4 billion years ago. “This is an entirely different organism from the ones that we know of,” said Li, who is based at the plant-focused Boyce Thompson Institute. “We know very little about them, and they’re completely different from other cyanobacteria.”

Fay-Wei Li and his team characterized a new type of Gloeobacteria, a lineage that split from modern cyanobacteria more than 2 billion years ago, and cultured it in the lab.

Courtesy of Fay-Wei Li

Li’s team used cryo-electron microscopy to examine the bacterium’s subcellular structures and capture detailed images of its inner workings. A. panamensis had both photosystems — but no thylakoids. A sister species, Gloeobacter violaceus, discovered in 1974, also lacked thylakoids, indicating that the now-familiar structures found throughout modern cyanobacteria likely hadn’t yet evolved when Gloeobacteria split from the group. Instead, its photosystems stud the cell’s external plasma membrane rather than the thylakoid membrane found in plants and modern cyanobacteria.

Among the most conspicuous anomalies were the structures A. panamensis used to harvest light. Most modern cyanobacteria have large antenna complexes (or phycobilisomes), built from proteins infused with light-absorbing pigments, that fan out from the thylakoid membrane in a large semicircle. The antenna complex in A. panamensis, however, looked nothing like a fan: It better resembled a canoe paddle. In 2023, experiments showed that this paddle-shaped antenna strongly reduced the rate at which A. panamensis could photosynthesize. Li suspects that the paddle shape doesn’t collect as many photons as a fan shape.

But what about its core photosynthetic machinery? A different team of scientists, including Gisriel, conducted a deep dive into photosystem I of A. panamensis to look for clues to how oxygen-producing photosynthesis might have evolved. The reaction center at the heart of the photosystem, where chlorophyll pigments absorb photons and produce sugar from carbon dioxide, showed only a few small changes when compared to the photosystems of other Gloeobacteria. However, there was far more evolutionary change on the photosystem’s light-harvesting proteins that bind the pigments, such as the ones that make up the paddle-shaped antenna complex.

These results, published in May 2025 in the Proceedings of the National Academy of Sciences, suggest that the forces of natural selection have made limited changes to the photosystem’s core, but that other aspects of the photosynthetic machinery have proved more malleable.

“The photosynthetic system complexes are incredibly similar. There’s not that much variation, even where there’s relatively significant changes [in other places],” said the chemist Gary Brudvig of Yale University, a co-author on the paper. “Once nature came on with the solution, it didn’t change very much. It kept the basic framework.”

The new findings in A. panamensis suggest that Gloeobacteria may indeed reflect a more ancient, basal form of photosynthesis. Over the past few billion years, most cyanobacteria and plants have evolved new features — more complex photosynthetic proteins as well as entirely new structures, such as thylakoids, to make the process more efficient. However, the photosystem I architecture of the Gloeobacteria has changed little over all that time. Now researchers can inspect the core of the photosynthesis reaction center to understand how this ancient molecular machine might have evolved.

“In my opinion, this makes it all that much more unlikely that oxygenic photosynthesis was at the root of the bacterial tree,” said Charles Delwiche at the University of Maryland, who studies the evolution of photosynthesis and was not involved in the study.

That is assuming, however, that structures in Gloeobacteria are nearly identical to what they were billions of years ago. “[It] is not a snapshot in time. It’s not a fossil from 2.5 billion years ago,” said Patrick Shih, a plant synthetic biologist at the University of California, Berkeley who was not involved in the study. “It’s undergone 2.5 billion years of evolution since that most recent common ancestor.” That’s why discovering more sister species to A. panamensis remains important, he said: More examples are needed to make a fully convincing case.

Evolutionary Oddity

The evolutionary discoveries surrounding A. panamensis — its lack of thylakoids, unique antenna, and stable photosystem — have encouraged Grettenberger and Li to embark on a global quest for more Gloeobacteria. Grettenberger wants not only more DNA evidence of this important group, but also species that, like A. panamensis, can grow in the lab.

“It would be really, really nice if we could get even earlier branching [species], so that we could see a stepwise evolution,” she said. Only by enlarging the comparison group will biologists be able to determine whether some of the traits seen in A. panamensis are evolutionary oddities or more broadly representative of early life, she added.

The biochemist Christopher Gisriel (center) and his team found that the core photosynthetic machinery of A. panamensis remained remarkably similar to that of other Gloeobacteria despite 1.4 billion years of evolutionary distance between them.

Paul Escalante

More examples could also help address the mystery of which photosystem evolved first. While the hypothesis that the very first photosynthetic organisms were anoxygenic would be the most logical explanation, in several papers published in 2019 and 2021, the evolutionary biologist Tanai Cardona at Queen Mary University of London argued that oxygenic photosynthesis may well have evolved first. Scientists have so long assumed that anoxygenic photosynthesis — and photosystem I — came first that any new data is immediately interpreted through that lens, he said, even though the data remains incomplete.

Both photosystems “go back to a single ancestor,” said Cardona, who co-authored the 2025 paper on A. panamensis. “But at this stage, we have no sequence evidence to suggest that one gave rise to the other.” The gene duplication event that doubled the photosystems, allowing one copy to evolve a novel process, happened so long ago that any evidence of that first photosystem has been buried by extinction and billions of years of change. Still, clues are likely out there.

For Shih, the implications of this work go far beyond basic science. The reality is that the chemical reactions that comprise photosynthesis and sugar production — the foundation of life’s diet, and thus humanity’s — are highly inefficient. Shih and other scientists are working to engineer improvements in photosynthesis to boost crop production. Before scientists start experimenting, they need to know how the parts of the photosynthetic system fit together.

That system, as we have it today, started some 2 billion years ago, when the emergence of oxygenic photosynthesis left an indelible green stain across Earth’s surface. That was when microbes wrested energy from light and, in the process, breathed new life into the world.

How Terry Tao Became an Evangelist for AI in Math

Kevin Hartnett — Mon, 08 Jun 2026 03:17:08 GMT

Automated proof-checkers such as Lean can provide ironclad assurances that mathematical proofs are valid.

Samuel Velasco/Quanta Magazine. Code courtesy of Alex Kontorovich

The following has been adapted from The Proof in the Code: How a Truth Machine Is Transforming Math and AI by Kevin Hartnett.

Terry Tao has never been afraid of unconventional ideas. In November 2014, he was on a panel of five distinguished mathematicians, all inaugural recipients of the Breakthrough Prize in Mathematics, which came with a $3 million award. The laureates’ conversation ranged from whether mathematics is invented or discovered — most of the mathematicians agreed that, at the very least, it feels like an act of discovery — to an assessment of the odds that we’re living in a digital simulation. “Yeah, I think we’re actually not real,” said Maxim Kontsevich, who did his most important work in the 1990s at the intersection of math and physics.

Yet over the course of the 40-minute discussion, the statements that drew the most incredulity were Tao’s. He predicted that in the future, instead of working alone or in small teams of two or three, mathematicians might work on projects with hundreds of other people at a time. And when these collaborations were over, he said — in his modest, understated way — the results might be checked not by human referees but by computers. “One day we may actually write our papers not in LaTeX, but in some language which some smart software will convert to a formal language, and every so often you’ll get a compilation error — the computer does not understand how you derived this step,” he said.

The statement was greeted by the event moderator and the other laureates as preposterous enough to make the simulation hypothesis seem reasonable by comparison. Even more surprising than the idea of hundreds of mathematicians working together was the fact that such a collaboration would appeal to Tao — because if anyone in the world seemed well suited to going it alone, it was him.

Tao was born in 1975 in Adelaide, Australia, three years after his parents immigrated to the country from Hong Kong. The first signs that their firstborn son was different came early. When Tao was 2 and his family was visiting friends, his parents found him gathered with several 6-year-olds, demonstrating how to count using wooden blocks. Asked how he’d learned to count things, he responded that he had seen it on Sesame Street. Five years later, when Tao was 7, he began learning calculus.

For three weeks in the spring of 1985, Tao’s parents brought him to the United States, where he met with Julian Stanley, director of the Study of Mathematically Precocious Youth, then at Johns Hopkins University. Stanley described Tao as having the greatest mathematical ability he had ever seen. That same year Tao met the acclaimed mathematician Paul Erdős during the latter’s visit to Adelaide. A famous picture shows the grandfatherly Erdős, 72 at the time, reading a document in his lap while Tao, 10 years old with thick black hair, looks on intently, fingers raised thoughtfully to his chin.

Tao’s young legend grew when he entered the International Math Olympiad in 1986. He won a bronze medal that first year, becoming, at the age of 10, the youngest competitor ever to achieve that result. In the two succeeding years he became the youngest-ever silver medalist and finally the youngest person ever to win a gold medal. His formal education proceeded at a similarly accelerated pace. He graduated from the local Flinders University in Adelaide when he was 15 and, in the fall of 1992, boarded a plane with his father for New Jersey, where he started a Ph.D. in math at Princeton University. Erdős had endorsed Tao’s early admission to the program, writing in a letter of recommendation, “I am sure he will develop into a first-rate mathematician and perhaps into a really great one.”

Paul Erdős, 72 at the time, with a 10-year-old Terrence Tao.

Billy Grace Tao

Erdős was right. By the time Tao was 24, he had made enough new discoveries to have his choice of permanent faculty positions; he ultimately decided to settle at the University of California, Los Angeles. Around that time, he met a young English number theorist named Ben Green. The two began collaborating on a proof that certain kinds of patterns called arithmetic progressions — in which the numbers in a set increase by a fixed interval, like 7, 10, 13, 16 — inevitably appear in large collections of prime numbers, despite the fact that primes appear to be scattered randomly along the number line. Their proof would become the signature result of Tao’s early career, contributing to his winning the Fields Medal in 2006, and propelling him to the upper echelons of mathematics.

Tao could have built a successful career without collaborating with anyone, but that’s not the way he liked to work. He viewed working with other researchers as a primary way to discover new ideas — take what you know, pair it with what I know, and see what happens.

This approach led Tao’s mathematical research to range over an unusually broad set of topics, from analytic number theory, including the Green-Tao theorem about prime numbers, to analysis, where he studied properties of the Navier-Stokes equations that describe the behavior of fluids, to algorithms for constructing MRI images from digital data. (The MRI collaboration developed during conversations Tao had with Emmanuel Candès, a statistician then at the California Institute of Technology, while they were both dropping off their kids at preschool.) This thirst for collaborative discovery also led Tao to do a lot of his work in public. In 2007, he started a blog, where he began publishing regular updates about his research. By that point, Tao was one of the most famous mathematicians not only in his field but in the world. His posts received a lot of attention and sometimes led to long exchanges in the comments section, where Tao enthusiastically participated. He did it because he found it fun, and because he hoped the conversation might generate new ideas.

Around that time, another early math blogger had a similar thought. Like Tao, Timothy Gowers was a prominent research mathematician with a taste for public exchange. But rather than trusting serendipity to strike in his blog’s comment section, Gowers wanted to channel public energy in a focused way. In January 2009, he published a blog post announcing his desire to facilitate a new kind of “massively collaborative mathematics.” He would propose a problem in an open online forum, and “anybody who had anything whatsoever to say about the problem could chip in.” He named it the Polymath Project.

Tao jumped in. Like Gowers, he understood that some math problems were more amenable than others to being solved through large-scale collaboration. The key, as Tao wrote in a comment on Gowers’ initial post, was to find problems that could “generate a number of simpler sub-problems … which can largely be worked on in parallel.” By breaking big problems into individual cases, different teams or individuals could work on their own and then assemble their results as pieces of a bigger whole. At the same time, Tao knew that perhaps the biggest challenge with the Polymath model would be organizing: moderating contributions and checking to make sure that all the contributions were correct.

For the first Polymath project, Gowers proposed improving a result called the Hales-Jewett theorem, which was about patterns that appear when you shade cells in a grid with one of two different colors. After a few months of work, coordinated through thousands of comments by dozens of mathematicians, the group had proved a more exact statement about how those coloring patterns emerge. That fall, they released the work as a first-of-its-kind math paper with the pseudonymous byline “D.H.J. Polymath.” Gowers’ experiment had been a success. It allowed many mathematicians — professional and amateur alike — to work together and yielded a proof in the end.

Over the next decade, there were 15 more Polymath projects, some of which Tao led, and the initiative attracted mainstream attention. On October 29, 2011, The Wall Street Journal ran an article called “The New Einsteins Will Be Scientists Who Share” and reported that the Polymath Project had “pioneered a new approach to problem-solving.”

Samuel Velasco/Quanta Magazine

Yet in other ways, the Polymath Project was an idea before its time. Tao found it exhilarating to be at the center of a frenzy of mathematical activity, but he recognized that the comments section of a blog was a limited platform for collaboration. Massive open collaboration increased the likelihood of a certain kind of serendipitous discovery, but at the same time it heightened the odds that any one of the many participants would contribute a mistake. The only way to guard against error was for a moderator to carefully check all the work. But that kind of moderation bottleneck undermined the Polymath vision.

What Tao was really after was an efficient new form of scientific discovery. And after a while, he came to understand that the Polymath model was not it. To make it real, he thought, some kind of computer verification would be needed — a way to check contributions automatically, rather than by hand. But given the state of technology in the 2010s, he might as well have wished for passenger service to Mars.

Tao had been aware of computer-verified mathematics for years. He knew about a few success stories, but he also knew that formal math was still impractical, requiring far more effort than it was worth in most cases. Nevertheless, Tao was intrigued by its potential. Almost uniquely among the world’s elite mathematicians, he saw the potential in new methods for doing mathematics.

In July 2022, in part to satisfy his curiosity, he began to organize a workshop on all the different ways computers were assisting mathematical research. He brought on a team of co-organizers, including Kevin Buzzard, who was at the time the world’s most visible evangelist for formal mathematics.

Going into the conference, Tao regarded Lean, software that allows mathematical proofs to be written and checked as computer code, as a complicated program that would take months to learn. Buzzard convinced him to give it a try. Along with that encouragement, Tao felt a strong responsibility to lead by example — if he was going to continue promoting machine-assisted proof, he needed to start trying it himself.

On October 9, 2023, Tao posted on social media, “I have decided to finally get acquainted with the #Lean4 interactive proof system (using AI assistance as necessary to help me use it).”

On MathOverflow, a popular online discussion forum for mathematicians, Tao found a question about something called Maclaurin’s inequality. He decided to answer it as an experiment in formalization. First, he wrote up the proof as a typical math paper. It was short, only 10 pages long. Then he turned his attention to his real goal: seeing if he could formalize the simple proof in Lean.

Initially, Tao thought he might be able to do it in a week, but he was quickly confronted by the differences between writing math by hand and typing it in Lean. Tao observed that the hard parts of the proof were easy to formalize in Lean, while the simple parts took a surprising amount of work.

In the regular paper, Tao spent no time at all asserting that if you have three numbers, all of which are greater than 1, their sum is necessarily at least 3. But Lean doesn’t abide assertions, and Tao had to spend time digging up a lemma in Mathlib — a digital library of already-formalized mathematics that Lean users draw on when writing proofs — that proved the self-evident relationship. Similarly, in informal math, it’s not necessary to always specify which number system you’re working in. The number 3, for example, is simultaneously an integer, a natural number, and a real number. In his original paper, Tao could simply write “3” without specifying what kind of 3 he had in mind. However, in Lean he had to spell it out. Tao found his proof kept failing to compile because he had neglected to specify the correct type at different points in the formalization. It wasn’t until nearly a month later, on November 6, that Tao posted in a comment on his blog, “Just a remark that I have managed to formalize the results of this paper in Lean4.” The result was minor, and the Lean code he had written to formalize it was terrible. Yet Tao was now officially a contributing member of the Lean community.

At the same time he was learning Lean, Tao continued work on a host of other research projects. These included one with his longtime collaborators Ben Green and Tim Gowers, as well as Freddie Manners, a former student of Green’s and now a professor at the University of California, San Diego. It was an elite set of collaborators — Gowers, like Tao, had won a Fields Medal, while Green was among the most decorated number theorists in the field.

The group had a particular problem in their sights, one that revolved around a mathematical object called a sumset. If you have a collection of numbers, it can be used to form another, related collection: its sumset. The sumset is made by taking the sum of every unique pair of numbers in the first set. All those sums together form the sumset of the original set.

If the original set is full of random numbers, then its sumset will be comparatively large. A set of 10 random numbers has a sumset of about 50 numbers (and a set of 1,000 numbers has a sumset of about 500,000 numbers). But if, instead of containing random numbers, the original set follows some kind of pattern, its sumset will be much smaller because many sums will appear multiple times (and you only include each sum once in the sumset). The set of the numbers 1 to 10 is one example — its sumset only contains 17 numbers (not 50, as you’d expect if it were a random collection of 10 numbers), because many of the sums repeat (1 + 6, 2 + 5, and 3 + 4 all equal 7, and you only enter 7 once in the sumset).

In addition to having a small sumset, the numbers 1 through 10 are an example of an arithmetic progression because they increase by a constant interval. A conjecture with roots in the 1960s by the computer scientist Katalin Marton asserts that this isn’t a coincidence. She predicted that sets that produce small sumsets must also include long arithmetic progressions. Gowers, Green, and Tao had made headway on a refined version of this problem called the polynomial Freiman-Ruzsa conjecture in the early 2000s but eventually got stuck. Then, in 2023, Tao, Green, and Manners picked it up again with an eye toward introducing techniques from probability theory that Manners had developed.

They realized that by combining those techniques with Gowers’ earlier ideas, they might be able to solve the whole thing. They brought Gowers into the collaboration, and the quartet made steady progress through the summer of 2023. By late fall, they had it. On November 9 — just three days after Tao uploaded his first formal Lean proof to GitHub — they uploaded their proof to arxiv.org.

With Lean on his mind, Tao suggested to his three co-authors that they could try formalizing their paper. The work seemed like a good candidate for formalization both because it was an important result and because it relied on relatively simple techniques. They wouldn’t have to spend months adding prerequisite material to Mathlib — most of the necessary definitions were already there.

However, Green, Gowers, and Manners weren’t especially interested in taking the time to learn Lean. So Tao set off on his own — though he knew he likely wouldn’t be alone for long. Any project he led was likely to draw attention.

On November 13, Tao kicked off a new channel in a Lean-focused chat group. “Hi everyone. I am thinking of starting a project to formalize in Lean4 the recent proof of Timothy Gowers, Ben Green, Freddie Manners, and myself of the polynomial Freiman-Ruzsa (PFR) conjecture,” he wrote. He would use the channel to coordinate activity on the project and would be “happy to accept volunteers to contribute to this project in whatever capacity they feel able.” It was a reboot of the Polymath Project, only this time they were formalizing an existing result rather than trying to prove a new one — and all the work would be verified by Lean, meaning Tao wouldn’t have to check it himself.

Within a day, Yaël Dillies, a Ph.D. student at Stockholm University, had set up a rough blueprint for the project that divided the proof into 13 sections. Within each section, Tao identified the sequence of lemmas and definitions that needed to be formalized. In a typical math paper, lemmas — simpler results that help build toward the proof of a larger theorem — might be about 20 lines long, but for the PFR formalization, Tao broke the proof down into five-line lemmas. His goal was to make the proof as modular as possible, allowing many people to make small contributions.

For the first week, most of the activity on the thread was about formalizing basic concepts from probability theory that the proof required but were not yet in Mathlib. In particular, they had to formalize Shannon entropy — a measure of the uncertainty or disorder in a data source, like a set of numbers. But along with formalizing math, Tao and the others spent that first week figuring out how to work together. Initially, the conversation was free-form, with Tao posting about what he thought needed to be done and others chiming in with ideas about how to do it, much as the Polymath projects had unfolded in blog comments.

On November 22, Tao posted a list of 22 outstanding lemmas and wrote, “If you want to claim one or more of these lemmas, please do so by replying in this thread.” The replies flooded in: “I’d like to claim the entropy of a uniform random variable :),” wrote Paul Lezeau, a Ph.D. student at the London School of Geometry and Number Theory. “I’m gonna take a stab at the general fibring identity,” replied Aaron Anderson, a Ph.D. student at UCLA.

Drawn by word of mouth, more and more mathematicians joined the effort. By the end of November, Tao, like a harried volunteer coordinator, was writing little Lean code himself, instead focusing on finding tasks for others to do. On November 28, he wrote, “Given that we may temporarily have a surplus of volunteers for the PFR project as it nears completion, I thought of one additional small task that someone might be willing to work on.” Forty-six minutes later, Kim Morrison replied that they had completed the task. “Wow, that was quick! Thanks!” Tao answered.

Even before the formalization was complete, the Lean community began discussing what it meant. In particular, they debated whether the efficiency of the project signaled a new era of fast formalization, or whether it reflected the singular influence of Terry Tao. In a wrap-up post in the group thread, Tao reflected that he had not written much of the code himself. “This is actually quite encouraging to me, as it suggests to me that it will be possible for mathematicians to lead Lean formalization projects without requiring extensive Lean programming skills (though one may need at least enough expertise to be able to state lemmas, if not prove them).” Eight minutes later, Johan Commelin, a mathematician at Utrecht University and the director of the Mathlib initiative, replied, “I don’t want to immediately hijack this thread,” before going on to question whether the lessons Tao had learned during the project were broadly applicable. “Of course you got a lot of help with this project because of its high-profile nature,” he wrote.

Commelin also noted that while projects like PFR were fun and exciting to take part in, contributing to them was not the kind of thing young mathematicians were recognized for when they applied for academic jobs. “At the moment, it is still not clear how formalizers (for lack of a better job description) will be credited by the mathematical community, and how these activities will be valued on the job market.” Tao replied, “For what it’s worth, I’m more than happy to mention contributions to this project in letters of reference as appropriate.”

By 2024, Tao had become the most prominent public voice touting the potential of machine-assisted mathematics. He was three years into his tenure on President Joe Biden’s President’s Council of Advisors on Science and Technology and had become co-chair of a working group on generative AI. In a pair of high-profile speeches in 2024, he expressed his vision for a new kind of mathematical collaboration: one that combined human insight, the creativity of large language models, and the guarantees of correctness provided by formal verification systems.

He came to this view in part because he saw clear limits on what current AI tools could do. They excelled at solving straightforward problems or tasks with plenty of prior data, but at the frontier of mathematics — where there were few published results and little data to train on — AI faltered. In his early experiments with LLMs, he observed that they behaved like overconfident undergraduates, offering suggestions without the expertise to tell the difference between good and bad ideas.

Yet Tao had a way forward in mind. He didn’t think AI would replace human mathematicians anytime soon, but he did consider it particularly well suited to helping solve certain types of complex mathematical problems: ones that could be broken into thousands of small, manageable subproblems — essentially the same class of problems that worked well for Polymath projects. At that scale, mathematicians could employ AI to solve large swaths of the easiest subproblems, with its results outputted as formal proofs that Lean could check, and step in to handle the most difficult remaining questions themselves. In 2024, Tao was promoting this vision to anyone who would listen, and following the PFR project, he had realized that if he really believed in the work, he needed to step up and lead it himself. He also knew right away which problem he would start with.

It was a question that he had stumbled upon a year earlier. In July 2023, a user on MathOverflow posed a seemingly simple puzzle. Consider an operation like addition, the user wrote. It might follow certain fundamental algebraic laws like the commutative law, which says x + y = y + x, or the associative law, which states that (x + y) + z = x + (y + z). In many cases, there’s no relationship between one law and another — the commutative law doesn’t imply the associative law, for example.

The MathOverflow question concerned the relationship between two particular laws, and another user answered it quickly.

But the question of how laws relate to each other in general caught Tao’s curiosity. Rather than solving puzzles one by one, Tao began sketching out a rough diagram that showed how different possible algebraic laws relate to one another. It became clear the picture might be quite complicated.

He saw that if he restricted his study to algebraic laws involving operations applied exactly four times, there were about 4,694 laws he had to account for. Each law could potentially imply or fail to imply any other law, creating 22 million logical implications to check. Once he had checked them all — either by proving they held or by finding a counterexample in which they failed — he would have a complete picture of how all 4,694 of those laws relate to each other. It felt like exactly the right scale for the new style of mathematics he was proposing.

Tao called his new endeavor “Equational Theories” and announced its formation in a post on his personal blog on September 25, 2024. He opened by ticking through the main reasons large-scale public math collaborations had been hard in the past and then wrote, “Proof assistant languages, such as Lean, provide a potential way to overcome these obstacles.”

To start, Tao and the growing number of volunteers who joined him tested the more than 4,000 laws against simple mathematical structures known as magmas. Magmas are stripped-down versions of arithmetic that made for a useful starting point because any law that failed to hold for magmas couldn’t possibly imply other, more complex laws. The participants quickly tested millions of these simplified systems using basic Python scripts and within days had resolved more than 99% of the 22 million potential implications. Tao posted on day 2 — September 27 — that he was astonished at how rapidly the project was advancing: “This project has moved far, far quicker, and scaled up much much faster, than I had expected — only 48 hours in and already a large fraction of the implications are likely to be resolved soon! I thought the 3-week PFR project was fast, but this is an insane additional level of speed.”

Once the simplest implications were resolved, the Equational Theories volunteers moved in a decentralized way to automated theorem provers that could search for solutions to problems all on their own, without interactive help. These provers, along with old-fashioned human ingenuity, knocked down the open questions one at a time.

Like a scientist watching his own creation come to life, Tao admired the work as it unfolded. “The project seems to be successfully decentralizing; in particular, there is now a lot of activity going on now that I am not fully aware of,” he wrote.

To many mathematicians, Tao’s project was intriguing but odd. Buzzard followed along, fascinated by the social experiment though bored by the mathematical content of Equational Theories. He thought it was both elementary and weird, though he admired Tao’s inventiveness. John Baez, another prominent mathematician, was more blunt. He remarked, “This seems like a colossal waste of time to me,” before acknowledging that he felt the same way about college football, and plenty of people liked that, too.

Within a month, the Equational Theories group had narrowed 22 million questions down to 238. By late November, they were down to 138. As they chipped away at the remaining cases, progress slowed. When the new year began, about 30 implications were left unresolved, and the rate of progress slowed even further. By the end of March, they had been stuck for several weeks on just four. Contributors tried their hand at the remainder, but with so few implications left, many people drifted away; Tao’s updates slackened from their near-daily cadence to once every few weeks.

But settling every single one of the 22 million implications had never really been the goal of Equational Theories. Out of sheer curiosity Tao had wanted a map of the complete landscape, and now he had that, minus a few details. More importantly, he viewed Equational Theories as a pilot project for a fundamentally new way of doing mathematics — and in that regard it was an unqualified success.

Equational Theories, in Tao’s view, was the opening act of what he hoped would become a new era of “experimental” mathematics. He had in mind the kind of transformation that had already come to fields like physics. Physics had once been a largely theoretical discipline where solitary thinkers or small groups of collaborators tackled one or two problems at a time — in other words, it used to look a lot like math still did. But with technological advancements came a new, experimental branch of the field — massive collaborations at laboratories like CERN’s Large Hadron Collider, where hundreds or even thousands of researchers with specialized skills worked together and generated huge volumes of data. These experiments didn’t replace theory but complemented it, with new results washing between the two modes of investigation.

Tao imagined a similar evolution happening in mathematics. He trusted that novel forms of inquiry would inevitably lead to novel insights, as they always had before. As the Equational Theories team methodically crossed implications off their immense table, they stumbled onto genuinely new mathematical constructions — like “magma cohomology,” an alien extension of the concept of group cohomology, a deep and well-studied field describing when groups can or cannot be enlarged in certain ways. Tao reached out to John Baez — the Equational Theories naysayer and an expert in cohomology — to ask if this construction had been seen before. Baez admitted he had never encountered it.

To Tao, that was exactly the point. The project had shown that mathematics could be conducted differently, experimentally — and in doing so, it had turned up something genuinely new. Tao had never expected Equational Theories to unearth a revelation; he wanted it to demonstrate the efficacy of a new kind of mathematical machine. And in that sense it worked. Terry Tao had found a new way of doing mathematics, and he showed no signs of going back.

Excerpted from The Proof in the Code: How a Truth Machine Is Transforming Math and AI by Kevin Hartnett. Copyright © 2026 by Kevin Hartnett. To be published on June 9, 2026, by Quanta Books in partnership with Farrar, Straus and Giroux. All rights reserved.

Are Memories Transferable — or Edible?

Claire L. Evans — Fri, 05 Jun 2026 02:21:19 GMT

Lisett Ledón for Quanta Magaizine

t was the dead of winter in Boston. The surface of the Charles River was frozen solid. But Zachary Kelso braved the biting cold to finally put to rest a mystery that has haunted neuroscience labs for over half a century.

To do that, Kelso, a research assistant in the Harvard lab of the neuroscientist Sam Gershman, needed some worms. Specifically, planarians: arrow-headed flatworms, which are among the simplest creatures to possess a brain and a nervous system with bilateral symmetry like ours. Normally, labs order these widely used model organisms from biological supply companies. But the mail-order worms weren’t up to snuff. So Gershman had dispatched Kelso to the Charles’ icy banks to catch some wild ones. “I thought, ‘I’m going to look crazy because I’m using a hammer to beat through the ice,’” Kelso recalled. “So I wore the more business end of business casual.”

It wouldn’t be the last time Kelso found himself in this situation. The Charles River planarians, it turned out, didn’t cut it either. Neither did the worms he sourced while stream-hopping around Eugene, Oregon, in March 2025. Nor did the ones he fished from Michigan lakes that June — this time in thigh-high waders — while picnicking families gawked from shore. Kelso diligently turned over rocks, angled with bits of meat tied to a string, and even followed maps from a vintage guidebook called The Fresh-Water Triclads of Michigan. But his adventure was fruitless. Sure, he caught plenty of planarians. But back in Gershman’s lab, none of them would do what they were supposed to do.

In the 1960s, an eccentric behavioral psychologist named James McConnell convinced the scientific establishment that planarian worms, like Pavlov’s dogs, could be classically conditioned — and that memories of this training could be transferred from worm to worm through cannibalism. These bizarre findings were replicated by other scientists, and worm training became a staple of high school science fairs. Now, 60 years later, the worms have stopped learning, and nobody knows why.

I first learned about this scientific mystery while reporting another piece for this magazine about what a cell can remember. As I dug into the historical literature on memory research, I kept coming across McConnell’s strange worm experiments, which captivated a generation of scientists before disappearing entirely. Planarian memory had itself been forgotten. I was content to dismiss it as a fluke of history until Gershman mentioned, in passing during an interview, that in addition to their work with the unicellular ciliate Stentor coeruleus, his lab was attempting to reproduce some wacky worm experiments from the 1960s. Had I heard of them?

Planarian worms have remarkable regenerative capacity. A fragment that’s 1/279 of the original worm can regrow into a normal adult in weeks.

Ernest Cooper

Gershman, I learned, was keen to pick up where McConnell had left off. As part of a growing cohort of cognitive scientists looking beyond the brain for clues to the origins and basis of memory, he’s fascinated by any creature that seems to remember without the benefit of neural, synaptic networks. Little Stentor coeruleus, for example, can modify its behavior based on previous experience — quite a feat for a single-celled creature that can’t possibly have a neuron. Planarian worms, if McConnell’s findings were to be believed, might be the next great model organism for memory research.

The trouble was, it wasn’t going well. In fact, no matter how hard Gershman tried to train them, none of his planarians would learn a thing.

Can a worm learn? When McConnell posed the question in the early 1950s, the notion that memory had something to do with synaptic associations between neurons in the brain was just beginning to gain currency. McConnell, then a graduate student in psychology at the University of Texas, reasoned that planarians — among the simplest creatures with true neurons — should therefore be able to learn.

His early worm experiments were not particularly novel. He simply substituted worms for rats in what were, at the time, standard classical conditioning studies: repeatedly shocking the worms while exposing them to a bright light. After a period of this training, the worms came to associate the light with the shock and scrunched their bodies in anticipation whenever the light flashed. Voilà: worm learning!

Planarians have stranger features to offer for experiments. If a planarian is chopped in half, both halves will regrow into a new worm — the tail will grow a new head, and the head will grow a new tail. A fragment as small as 1/279 of the original worm can regrow into a completely normal adult worm in a matter of weeks, a regenerative capacity so powerful that, as one early naturalist put it, planarians are effectively “immortal under the edge of the knife.” For McConnell, this ability begged the question: When you chop a worm in half, do both halves remember?

This is where the real worm torture began.

In the ’60s, McConnell, by then a young professor at the University of Michigan, started beheading his trained planarians. The worms that grew back from the severed heads behaved as the originals had, associating the light with the shock — a result he expected, given the preservation of their primitive brains. What surprised McConnell was that the worms that regenerated from headless tails remembered, too. This meant that whatever form the worms’ memories took, they weren’t the exclusive purview of the brain. “It appeared that the memories were laid down throughout the animal’s body,” McConnell later reflected.

James McConnell stands next to a printed logo of The Worm Runner’s Digest, the half-satirical scientific journal he published from his lab at the University of Michigan.

University of Michigan News and Information Services

Thrilled, McConnell pushed his experiments further. He cut the worms into smaller and smaller pieces; each time, the regenerated segments retained the memory. He stitched the heads of trained worms onto untrained tails, but they kept falling off. He pureed trained worms and injected them into naïve recipients, a delicate process that the historian Larry Stern has compared to “impal[ing] a prune with a javelin.” Finally, remembering that some planarians are cannibals, he fed trained-worm puree to their brethren. In subsequent trials, the “cannibal” worms picked up the light response right away, as though they were remembering, rather than learning, what to do.

If McConnell’s experiments appear gruesome, his line of inquiry was of his time. The discovery of the DNA helix in the 1950s had revealed just how much information is packed into proteins and nucleic acids. The notion that the physical traces of memories, or “engrams,” might have some chemical basis seemed plausible enough to many scientists. Could McConnell’s cannibal worms have eaten an engram? McConnell certainly thought so. He was convinced that their memories were encoded in the structure of their RNA — and could be transferred from worm to worm.

“In the jargon of computer engineering, information is always ‘fed’ into a computer,” the journalist Arthur Koestler later wrote in an appreciative survey of McConnell’s work. “Here the metaphor became flesh.”

These were sensational findings, and McConnell took every advantage of the media attention they generated. Before becoming a scientist, he had had a brief career in radio, and he knew how to repackage nuanced ideas as pithy soundbites. In magazines such as Time and Esquire, he spoke grandly of a future of memory consumption — of “piano lesson pills” and “professor burgers.” He brought his trained worms onto The Steve Allen Show, and, belying his clean-cut hair and horn-rimmed glasses, dubbed himself “McCannibal.”

A planarian worm swims in a drop of water, viewed through a microscope.

Luke Groskin

Students began writing to McConnell’s lab at the University of Michigan to ask for worm-training tips for their school science fairs, and McConnell shared advice. Science, he believed, should be for the people; he saw himself as a latter-day David pitching stones at institutional Goliaths. This made him one of the most famous public scientists of his era, but it did not endear him to more serious peers. It also didn’t help that he published all his research in The Worm Runner’s Digest, a countercultural journal he distributed from his lab.

The Worm Runner’s Digest was “sort of Mad Magazine meets a serious scientific journal,” Gershman told me recently. At its peak, it had some 2,500 subscribers around the world. The hand-drawn shield on its cover featured a two-headed planarian and the Latin motto ignotum per ignotius, which roughly translates to “the unknown explained through the even more unknown.” Its 1959 inaugural issue consisted of only 14 mimeographed pages about the care and feeding of planarians, but it quickly grew. In addition to publishing dozens of memory transfer papers and related scholarship, McConnell welcomed humor and printed science fiction stories, rousing editorials, student-drawn planarian cartoons, spoof articles, and poems.

While the Digest is now something of a cult classic, the mix proved confusing for many readers. McConnell eventually cut the publication in half, not unlike a planarian worm, and renamed the serious half The Journal of Biological Psychology (no relation to the current peer-reviewed journal Biological Psychology, founded in 1973). But McConnell’s reputation as a heretic and prankster was well established.

The wheels started to come off in the mid-1960s. Although McConnell enjoyed a period of fame and funding — including an accelerated path to tenure at the University of Michigan — attempts to replicate his memory transfers yielded inconsistent results. While many apparently succeeded, the failures were more visible. In 1965, the Nobel Prize–winning biochemist Melvin Calvin tried to replicate McConnell’s worm experiments and failed, even with the help of some of McConnell’s former assistants and using the same device. His high-profile publication of the results sparked an acrimonious debate about, among other things, proper worm handling.

By the 1970s, the planarian memory fad had come and gone. Scientists had moved on to rats, cats, goldfish, and even praying mantises. Researchers showing successful memory transfers in rats — by injecting brain RNA from one animal to another — had published their findings in prestigious journals such as Nature and Science, making the planarian model seem unimpressive in comparison. But when further experiments proved inconclusive, interest in the question of memory transfer petered out. As the science historians Harry Collins and Trevor Pinch put it, “memory transfer was never quite disproved; it just ceased to occupy the scientific imagination.”

In addition to worm-related scholarship, The Worm Runner’s Digest, a scientific journal published by James McConnell, featured cartoons, poetry, editorials, spoof articles, and other humorous or satirical items.

Courtesy of Sam Gershman and Zachary Kelso

McConnell closed his laboratory in 1971, and his long period of subsequent obscurity was broken only once, in 1985, when he became a victim of the Unabomber. (He lost his hearing temporarily after the blast.) He died in 1990. If a younger generation of scientists is familiar with his cannibal planarians, it’s as “a cautionary tale that neuroscientists tell to their students at bedtime to scare them away from ill-fated projects,” Gershman said.

Still, McConnell’s unconventional work and contrarian attitude has lingered in neuroscience lore, and the idea of memory transfer remains a subject of private fascination. What if McConnell really did manage to feed a memory to a worm? For Gershman, who is searching for a way to study memory at a molecular level and connect it to observable behavior, the question was an itch that had to be scratched. He decided to settle the matter once and for all.

It all seemed straightforward enough. In the spring of 2025, Gershman and Maddie Snyder, one of his postdocs, set out to reproduce the worm-training protocol of one of McConnell’s students, Alan Jacobson. Jacobson’s papers were the most rigorous of the planarian memory transfer era, and Gershman and Snyder followed them to the letter. “We wanted a behavioral basis to be able to study the circuits that are driving memory in these extremely unstable animals,” Snyder said. “Are those circuits at all being used for memory consolidation or storage? Because if you lose your head and all those circuits are gone, then what is the mechanism of storing memory?”

Despite their best efforts, however, they couldn’t do what Jacobson, McConnell, and so many others had done back in the 1960s: condition the worms to scrunch their bodies in response to light. (They reported the results on biorxiv.org in April 2026.) “I was really scratching my head about this,” Gershman said. He’d assumed that the memory transfer would be the dodgy part of the experiment — not getting the worms to form a memory in the first place.

They talked to other planarian labs. They tried different stimuli. They ran the planarian footage through a machine learning pipeline. In desperation, Kelso and Snyder even visited the Harvard Science Museum to examine a vintage “inductorium,” an electric-shock contraption used in midcentury worm science. But it offered no clues.

Kelso searched for any of McConnell’s former collaborators who might still be alive, and through a stroke of luck he found contact information for Daniel Kimble and his wife, Reeva, who had both worked in McConnell’s lab. Now in their 90s, they live in Eugene, Oregon. When Kelso called them, he discovered that not only did they run most of McConnell’s experiments in the 1960s, but they’d also kept the entire print archive of The Worm Runner’s Digest in a box in their basement.

Kelso and Snyder bought tickets to Eugene. Over the course of two days, while consuming plates of homemade cookies and countless cups of tea, they absorbed everything they could from the Kimbles. On breaks from hand-scanning back issues of the Digest, they went stream-hopping nearby and filled casserole dishes with silty fresh water. Back at the Kimbles’ kitchen table, the two generations of scientists watched the silt settle to see what worms might emerge.

Planarian worms have been subjected to extensive experimentation exploring their regenerative abilities, performed by professional researchers and high school students alike. Here, a worm has been induced to grow an additional head where its tail should be.

Taisaku Nogi, Dan Zhang, John D. Chan, Jonathan S. Marchant

This, after all, is what McConnell did. Rather than using laboratory strains of planarians, he sourced his from a lake near the University of Michigan. And so, not wanting to leave any stone unturned, Kelso made one final trip to McConnell’s former fishing grounds in Michigan. He returned with plastic tubes full of worms, but not a single one was a learner. “At some point, we had like 12 different strains of planaria, none of which showed any learning,” Gershman said.

As Snyder tells it, the Kimbles were absolutely convinced that their conditioning experiments in the 1960s had worked. The worms learned — they were sure of it. The literature of the era seems to support their certainty; at least 36 labs reported similar results. So why is it that when the exact same experiments are done today, using the same laboratory protocol and even the same worms, fished from the same Michigan waters, planarian worms are utterly uneducable?

One explanation, Snyder said, is that McConnell, the Kimbles, and all the other “worm runners” were inconsistent in how they scored the planarians’ behavior and may have misinterpreted more anodyne worm “turns” for the definitive “scrunch” of the light reaction. Every scientist is a product of their time, after all, influenced in myriad, often invisible ways by sociological conditions, funding pressures, and, in this case, a highly charismatic leader. This interpretation made Snyder hyperaware of her own potential biases. “Throughout this entire project,” she told me, “I was like, ‘What are the things that I am taking for granted now in our models of neuroscience, and our assumptions of what is known and unknown, that I should really notice?’”

A more remote possibility is that planarian worms themselves have somehow changed over the last six decades — falling victim to pollution or genetic drift. Gershman finds this scenario unlikely. “What are the chances that a bunch of researchers just happened to do these studies at the particular time when this phenomenon happened?” he asked, incredulous. “They just got extremely lucky, in the millions of years of planarian evolution? And then our luck ran out?”

No matter the reason, in 2026, despite their nervous system and simple brains, planarians don’t learn. From an evolutionary perspective, this might actually make sense. “The reason that we learn associations, to some degree, is so that we can predict danger and avoid it,” Snyder said. But planarians have a different relationship to danger. Their regenerative physiology, so key to McConnell’s experiments, protects them from blunt trauma. Bitten in half, they simply grow back. What use is memory to such a creature? “That’s a whole other philosophical conundrum,” she said.

It’s a great time to be wrestling with such conundrums. Planarian learning may be a dead end, but memory transfer experiments with other organisms are back in scientific vogue — and those experiments appear to be working. In 2018, the neuroscientist David Glanzman of the University of California, Los Angeles performed a memory transplant on the sea slug Aplysia californica, a darling model organism for memory research owing to its relatively simple nervous system and gigantic neurons. After training the slugs to respond to a shock to their tails, Glanzman was able to transfer the sensitization from one slug to another via a direct injection of genetic material. This suggested that some aspect of the memory was stored in RNA, which was McConnell’s contention.

Then, in 2021, the Princeton University geneticist Coleen Murphy found that Caenorhabditis elegans worms — microscopic roundworms with 302 neurons to their name — could learn to avoid a pathogenic bacterium by eating, or even just swimming around in, pureed worms who had learned the hard way. Murphy’s group identified a retrotransposon, a jumping segment of genetic material, called Cer1, that appears to “carry a memory” between individuals, she said. A few years later, a group at the Indian Institute of Science published a paper suggesting that trained C. elegans worms release extracellular vesicles — small lipid particles containing genetic information — that can impart their training to their naïve counterparts.

None of these researchers are half as flashy as McConnell was, but their work indicates that he may have been right about worm memory after all. He just bet on the wrong kind of worm — and doubled down on it, despite inconsistent evidence. In the end, he lost his reputation, but his sheer gusto sparked the curiosity of another unconventional scientist. Fortunately, this one’s following the evidence.

At Harvard, Gershman is shifting his focus from the inscrutable planarian to the more legible C. elegans. It may not regenerate when it’s cut in half, but C. elegans is a long-standing model organism in neuroscience — and it’s been consistently shown to learn. With new experiments underway, Gershman is cautiously optimistic. “I just hope we’re not going down another rabbit hole,” he told me. A wormhole, on the other hand — that’s for certain.

More Conversations, Complex Questions, and Bold Ideas in Season Five of ‘The Joy of Why’

Simon Frantz — Thu, 04 Jun 2026 02:12:35 GMT

Chanelle Nibbelink for Quanta Magazine

What is the future of gene editing with CRISPR? Has AI changed mathematics forever? Will we find other civilizations in the universe? What if we’ve been wrong about dark energy all along? These are just a few of the big, bold questions we’ll be exploring in the new season of The Joy of Why.

Mathematician Steven Strogatz and physicist Janna Levin are back as your hosts for these and other conversations that explore the frontiers of basic science and mathematics. Each episode features an in-depth conversation in which Steven or Janna sits down with a leading scientist or mathematician to unpack one big idea or area of research. The two hosts also chat together throughout each episode, sharing their own thoughts, reactions, and questions.

New episodes drop every other Thursday, kicking off on June 11 with biochemist and Nobel Laureate Jennifer Doudna, who helped revolutionize gene editing and biology as co-discoverer of the CRISPR-Cas9 system. The wide-ranging conversation explores the discovery and sudden rise of CRISPR as a tool that can modify genes in a highly precise manner, the successes and issues the work raised, and what comes next.

Check out the lineup of this season’s guests below — and stay tuned for a surprise or two. All 12 episodes of Season 5 will be available to stream or read here, and on Apple Podcasts, Spotify, TuneIn, or wherever you listen to podcasts.

Scientists and mathematicians appearing in the latest season of The Joy of Why include (left to right, top to bottom) biochemist Jennifer Doudna, astrophysicist Adam Reiss, mathematician Lauren Williams, neuroscientist Ishmail Abdus-Saboor, computer scientist Melanie Mitchell, astronomer David Kipping, evolutionary biologist Toby Kiers, physicist Albert-László Barabási, and mathematician Maggie Miller.

Credits (left to right, top to bottom): Christopher Michel; Courtesy of Adam Reiss; Lucy Lu; Courtesy of the Columbia Zuckerman Institute; Kate Joyce/Sante Fe Institute; Courtesy of David Kipping; Courtesy of Toby Kiers; Istvan Labady; Stanford University

Entanglement Builds Space-Time. Now “Magic” Gives It Gravity.

Charlie Wood — Wed, 03 Jun 2026 02:34:29 GMT

Irene Pérez for Quanta Magazine

In 1973, John Archibald Wheeler described the relationship between space and matter in two sentences: “Space acts on matter, telling it how to move. In turn, matter reacts back on space, telling it how to curve.” Wheeler’s words serve as a pithy encapsulation of general relativity, Albert Einstein’s theory of gravity.

Wheeler’s sentences also lay out a challenge that theorists face today: When they build a model of the universe — at least one that works at the quantum level — it’s been difficult to get space and matter to interact in the way that they must.

Einstein cast gravity not as a force but as the geometric bending of space and time. In a popular analogy, the fabric of space-time is like the flat expanse of a mattress, and a massive object like a star is like a bowling ball sitting on top. The weight of the bowling ball compresses the mattress, forming a dimple — matter tells space-time how to curve.

In this analogy, a planet is like a smaller ball. If it rolls close enough to the bowling ball, its path will be altered by the dimple in the mattress — space-time tells matter how to move.

But general relativity has a fatal flaw. When a star dies and collapses, its mass is concentrated into an unimaginably dense point. The dimple in the mattress stretches into a deep depression, one that essentially rips all the way through. Physicists call this arrangement a black hole. If a ball reaches such a rip, it’s no longer guided by the fabric, and the analogy breaks down; scientists need a new theory to understand this and other, similarly extreme situations.

In the late 1990s, physicists had a stroke of luck. They learned that if they imagined space-time as a collection of purely quantum particles, they could in principle describe a black hole — rip and all — in an entirely new way.

Theorists have spent the last few decades trying to understand exactly how a space-time constructed from such quantum particles could work. And they’ve made progress: They’ve found that entanglement between particles gives space-time its structure, building an environment where matter can move — and satisfying the conditions of Wheeler’s first statement. But the origin of Wheeler’s second statement remained mysterious; in their models, matter didn’t tell space how to curve. The bowling ball sat atop the mattress without making a dent.

Until now. Physicists including Charles Cao at Virginia Tech have recently determined how quantum particles could give space-time its bendiness. In a handful of recent works, multiple teams have identified a feature of quantum mechanics that Cao calls “the fabric softener of space.” It’s a measure of quantumness called “magic.”

“Without magic, things are a little too simple,” said John Preskill, a physicist at the California Institute of Technology who contributed to Cao’s newest paper. “And, you know, quantum space-time isn’t quite that simple.”

How To Code a Universe

Perspective shifts abound in physics. For instance, there’s more than one way to look at the motion of a pendulum. You might specify its location using the height and the horizontal displacement of the weight hanging at the end of the string. Or you might use the length of the string and its angle instead. The perspectives are equivalent; simple trigonometric equations take you from one perspective to the other.

Mark Belan/Quanta Magazine

For 50 years, theorists have been chasing a far more profound perspective shift: a new way, beyond Einstein’s curved space-time, to look at the universe.

In the early 1970s, Jacob Bekenstein and Stephen Hawking took the first step in that direction when they discovered that you could reinterpret a black hole (and anything that had fallen into it) as a spherical collection of particles. In the late 1990s, Juan Maldacena, Edward Witten, and others extended this insight to a whole universe; they described an exotic, static world as a throng of interacting particles, also arranged in a sphere.

In both cases, you could replace the 3D region of space-time with particles on the region’s surface. You could consider the surface to be 2D, like a globe flattened into a paper map. Physicists call this dual nature of space-time the holographic principle, since it resembles the way a holographic sticker can cram a whole 3D scene onto a flat surface without losing data.

Over the last couple of decades, theorists have explored what gives the 3D fabric of space its shape. Entanglement, a quantum property that links particles to one another, seems to serve as space’s connective tissue. Take, for instance, a wormhole, a theoretical bridge connecting two distant regions of space. Holographically, a 3D wormhole is equivalent to two entangled sets of particles. Start snipping the “threads” of entanglement that link one set with the other, and the tunnel connecting the regions gets thinner and thinner. Cut the final thread, and the connection dissolves entirely.

Cao learned about the link between entanglement and space as a graduate student at Caltech in 2016, most notably through a paper by Daniel Harlow, a physicist now at the Massachusetts Institute of Technology. “Charles spent a month understanding the paper,” said Jason Pollack, then a fellow graduate student, now a physicist at Syracuse University.

Harlow, building in part on the work of Preskill and others, had identified the type of math required to shift perspectives from 2D to 3D. He needed to encode a space and its matter — stars and planets and electrons — into a bunch of quantum particles. So why not use a quantum error-correcting code?

Quantum error-correcting codes are crucial to quantum computing because quantum computers work by manipulating “qubits,” quantum versions of bits that can exist in superpositions of 0s and 1s. Qubits are extremely delicate, frequently losing their superposition and therefore their extra information. And so physicists have worked out ways to protect this delicate information through redundancy. By spreading out one qubit’s information among many qubits, they can preserve it even if some of the qubits are lost.

The same type of redundancy shows up in holography. “When you design codes for quantum computing, you’re doing the same kind of thing that [holography] already did for you,” said Bartek Czech, a physicist at Tsinghua University in China. A single holographic location — a region of space and the matter in it — is not encoded in just one set of quantum particles; rather, it is spread across many sets, due to their entanglement. Harlow and collaborators detailed how this works in a code in 2014, and he further fleshed out the relationship in the 2016 paper that impressed Cao.

But these codes, known as “stabilizer codes,” had a shortcoming. They divided the entanglement of the particles into two types: one responsible for space and another responsible for matter. And the divide was unbridgeable. Such a perfect split is a virtue in quantum computing, since you want your encrypted data to stay perfectly isolated from the corrupting influence of the outside world. But in holography, that perfection left no room for the two to interact. “We knew how to build a space-time,” Czech said, but “this space-time was inert. It didn’t do anything.”

To get space and matter to interact, Cao knew he needed a more sophisticated code. “It was clear that something else beyond entanglement had to be there,” said Ning Bao, a physicist at Northeastern University.

The Magic Ingredient

Cao started by playing around with existing error-correcting codes. In 2020, he and a collaborator, Brad Lackey, tweaked one such code and found that it allowed space to change — just not in response to matter. It wasn’t gravity, but it was progress. Except that Cao and Lackey didn’t fully understand why the tweak worked.

M.C. Escher’s 1959 woodcut Circle Limit III has the geometry of a holographic world: A whole universe fits inside a spherical surface. In holography, you can learn about what’s happening in the interior by studying the surface itself.

M.C. Escher

The next year, Pollack and his collaborators realized that if you actually tried to create a quantum program that executed the tweaked code on a quantum computer, you’d need to use a particular operation known as a T gate, which rotates a qubit.

Cao took notice. He had just attended a quantum computing conference where researchers were buzzing about gates like these, in part because they are the key to making quantum computers more powerful than classical computers.

Researchers had previously thought the key was entanglement. They had worked out a way of running software on a classical computer that would mimic a quantum task. When that quantum task involved entangling qubits, quantum computers had an advantage over classical computers, as the classical program took ages to run. But then physicists discovered a way of speeding things up; it turned out that certain classical algorithms could mimic certain entangling operations even on a laptop.

In 2004, Alexei Kitaev and Sergey Bravyi, both then at Caltech, brought researchers’ attention to quantum operations known as non-Clifford gates, which include the T gate. When a quantum program uses these operations, the equivalent classical program takes much, much longer to run. Kitaev and Bravyi described the complexity that these operations introduce as “magic.” The more non-Clifford gates you need to produce a quantum state, the more magical that state is.

After Cao learned about magic and non-Clifford gates, he joined forces with Brian Swingle and Christopher White, both researchers at the University of Maryland. In 2020, they studied collections of particles equivalent to an exotic universe called an anti-de Sitter space. The group found that the particles were highly magical. What would the role of this magic be, they wondered, for the anti-de Sitter space the particles represented?

Cao — in partnership with Alioscia Hamma and others and building on work from Xi Dong, now at the University of California, Santa Barbara — found the answer a few years later. They showed that magic gave space its springiness. Magic, in other words, is connected to space’s ability to bend. And therefore magic is connected to gravity. “If you have one,” Bao said, “you always have the other.”

By early 2026, Cao and his collaborators had all the pieces. They knew that magic made space bend. And they knew that quantum codes got their magic from non-Clifford gates. So Cao, Preskill, and others created a next-generation code to succeed the stabilizer codes Harlow and others had focused on a decade before, when they split encoded space from encoded matter. This new code used lots of non-Clifford gates. The gates made the code magical, letting the entanglement for space and the entanglement for matter affect each other.

“This is pretty cool, because in quantum gravity, we don’t expect the background is fixed,” said Cynthia Keeler, a physicist at Arizona State University who was not involved in the work. “It should fluctuate.”

The essential nature of magic especially intrigues physicists like Swingle, who hope to use it on a quantum computer to simulate how gravity behaves in situations where general relativity fails. “If we need high magic, then we intrinsically need a quantum computer,” Swingle said, “because there’s no other way, in general, to get at that kind of question.”

Gravity From Quantumness

In principle, entanglement and magic could be enough for future physicists to simulate space on a quantum computer. But Cao’s new code still needs a lot of work.

During a talk about it at the American Physical Society’s annual summit in Denver, Cao joked that he was the only speaker who wasn’t actually studying quantum gravity. That’s because his code is still extremely general. It doesn’t describe the kind of space in which we live, doesn’t capture the particular reactions Einstein described, and doesn’t include the ticking of time.

The code is more of a proof of concept of the general shape that a theory of quantum gravity should take. If you want your space to bend, use a magical code. “This gets you a precursor of gravity,” Cao said. “You satisfy one of the necessary conditions. Right now, we are at step 0.5 of 5.”

But even at this early stage, the research program highlights some surprising features that any theory of quantum gravity should have.

Einstein and Wheeler thought of space-time as a large, featureless fabric existing with fixed bends and folds — a typical classical object. But now physicists are learning that the two defining features of quantum mechanics, entanglement and magic, correspond to the two defining features of space, its shape and its flexibility. This suggests that space itself is one of the most quantum things imaginable. “All the familiar aspects of gravity are actually a very direct manifestation of something quantum,” Swingle said.

It also suggests that gravity results from imperfect quantum encoding. Non-magical codes produce inert, gravity-free spaces because they protect their encoded information perfectly. Cao and collaborators have shown that gravity comes from the mixing of the encoded information. So by necessity, the encoding must be approximate, and therefore some aspects of what’s going on in the space-time can’t be perfectly recovered by measuring a subset of the quantum particles in the usual way. This approximation, which would indicate a poorly written code for a quantum computer, is “the reason Newton’s apple fell on him,” Czech said.

Cao, for his part, finds the feature appealing. Quantum error correction and quantum computing are human pursuits, he said. He sees no reason that gravity should accommodate our prejudice for perfection.

Correction: June 3, 2026
Charles Cao was a graduate student at Caltech, and he later held a postdoctoral position at the University of Maryland.

The article previously called the gates that introduce magic Toffoli gates and conflated them with a simpler gate called the T gate. Toffoli gates and T gates are both a part of a wider category called non-Clifford gates, all of which introduce magic.

The Dirt That Refused To Die

Siddhant Pusdekar — Mon, 01 Jun 2026 02:44:57 GMT

What appear to be biochemical processes may instead be a natural feature of geology.

Samuel Velasco/Quanta Magazine

For 15 years, Sébastien Fontaine has been trying to kill dirt. The biochemist, who runs a lab at the French National Institute for Agriculture, Food, and Environment, wanted to know how much carbon is released by soil — just dirt alone, completely devoid of life. His team sealed dirt into jars and blasted them with sterilizing gamma radiation. Then they waited for the carbon dioxide released by the soil — a sign of ongoing microbial respiration — to drop.

They waited, and waited, and waited some more: weeks, then months. Under a microscope, the irradiated soil showed no signs of life, but it continued to emit carbon dioxide. The soil wouldn’t stop breathing.

Fontaine’s lab repeated the experiments and produced the same results. Finally, convinced that they weren’t dealing with an artifact of the experimental setup, they set out to find the source of breath in dead soil.

Now, Fontaine and his colleagues have reported that their soil samples continued to consume oxygen and spew carbon dioxide for six years. In a 2025 paper in Science Advances, they proposed that a metabolic process that powers much of life is also possible outside living cells. Their experiments point to how it could work in dirt, absent the living proteins that would typically organize it. If they’re right, some biochemical reactions, such as those that release the energy of carbon-rich sugar molecules, may not be unique to living things. Such reactions — known as metabolism when performed by cells — could even predate life on Earth, Fontaine said.

The experiments show “what happens to biomolecules when they’re left to their own devices,” said Joseph Moran, an organic chemist at the University of Ottawa who was not involved with the research. They’re finding that the chemistry of life is not exclusive to life, he added. “It’s the chemistry of geology.”

The Living Dead

When he made this accidental discovery, Fontaine was trying to establish a baseline for carbon in lifeless soil. Using a sterile syringe, the researchers periodically sampled the air in a hermetically sealed jar containing soil and measured its carbon content using a mass spectrometer. After radiation wiped out the soil microbes, the carbon emission rate declined quickly but didn’t disappear. It remained stable for over 100 days.

When he shared the results with other researchers, they advised him to treat it as an experimental artifact — a source of error not worth ferreting out — and move on. But he couldn’t. He needed to understand whether a metabolic process only known to occur in biological cells — a precisely orchestrated sequence of chemical reactions, requiring several molecules and enzymes — was unfolding in sterile soil. To see what was happening, his team added a dash of enzymes extracted from yeast cultures. Immediately, the soil’s carbon emissions spiked. This, they speculated, was because the enzymes had ramped up a reaction that was already happening.

Convincing the scientific community, however, was an uphill battle. When Fontaine submitted the manuscript to journals for publication, some reviewers “were highly positive, and others were really suspicious, especially concerning the sterility of the soil,” he recalled. In 2013 the results were published in the journal Biogeosciences. Still, Fontaine could not rest. Bruised by the harsh reviews, he decided to definitively prove that his irradiated soil samples remained free of life. Over the following decade, his lab would, in fits and starts, chip away at their obsession.

They considered the possibility that the soil wasn’t really dead, and tried to kill it harder with more radiation, pressure, and heat. Still, the soil continued to emit carbon for months.

Through an electron microscope, Benoit Kéraval, then a graduate student in Fontaine’s lab, found cells in the irradiated soil. But staining showed no RNA or DNA molecules, indicating that the cells were definitely dead. When they experimentally added microbes to simulate contamination, the cells rapidly recolonized the soil microcosm and released much more carbon dioxide. So what they were observing in the sterilized sample likely wasn’t a result of inadequate antiseptic measures.

By 2018, when Clémentin Bouquet joined the lab, the team was confident in its findings and ready to dig into the underlying mechanisms.

Dirty Electrons

For six years, Bouquet and Kéraval studied two sets of sealed, irradiated soil samples — one of normal soil, and one that was supplemented with glucose. For 142 days, they took regular air samples and saw the daily rate of carbon dioxide emissions decline but not disappear, just as they had before. Then the samples sat in an incubator for over 1,000 days, as the researchers focused on their other experiments into how microbes process and store carbon in soil.

When they measured the samples again, at days 1,606 and 2,442, the emissions had slowed further, but the soil was still breathing. The glucose-augmented samples showed higher emission rates, which strengthened Fontaine’s suspicion that nonbiological catalysts in soil can induce reactions that resemble the metabolic breakdown of sugar.

During metabolism, sugar is broken down into smaller carbon molecules, which feed the Krebs cycle — a series of reactions in which high-energy electrons are stripped from carbon-rich molecules. Electrons liberated by the Krebs cycle then pass through another set of reactions that consume oxygen. For some researchers, it was a stretch to suggest that this process could unfold outside a cell. Fontaine would need to show that soil can play the same role.

He devised a fuel cell that could detect electrons zipping through soil in the form of a current. His team added soil that had been irradiated almost five years earlier, and then closed the circuit. A current passed through the soil that was several times higher than in a control setup involving a saltwater solution. According to Fontaine, the experiment demonstrated that sterile soil supports a flow of electrons indicative of processes that resemble the oxygen-dependent metabolism of the Krebs cycle.

When Clémentin Bouquet joined the lab in 2018, he co-led a six-year experiment that suggests that lifelike biochemistry takes place even in sterile soil.

Öykü Ataytür

It was once thought that the Krebs cycle cannot occur outside the controlled confines of a cell, which teems with enzymes that keep everything ticking along and increases the chances that biomolecules will bump into each other. In a 2025 preprint on biorxiv.org, Fontaine and colleagues reported observing four of the eight intermediate molecules known to be part of the Krebs cycle in 6-month-old sterile soil samples. Many of these molecules formed after the irradiation.

According to the authors, their results suggest that clods of earth can indeed catalyze these reactions without the presence of life.

An Origin of Life?

For Joshua Schimel, a soil ecologist at the University of California, Santa Barbara, Fontaine’s findings were not too surprising. “Glucose naturally, in the process of being oxidized, is going to form some of these Krebs-cycle intermediates,” he said. Many soils are rich in iron oxides and aluminum oxides, which can catalyze this conversion, he added.

The idea that metals can catalyze biochemical reactions is central to a theory about the origins of life that has emerged over the last decade. Metals such as iron and zinc sit at the core of many of the most ancient enzymes found across life forms. Some researchers, including Moran, believe they might have catalyzed these reactions before life emerged. Studies, including his, suggest that the chemical reactions that break down and construct glucose derivatives, which are normally associated with life, might have existed before the enzymes and genes that enable them in living cells.

“There’s a handful of researchers like myself that think, actually, we should organize our thoughts about life in a different way — that we actually should put metabolism at the base of what life is doing, and then genes are a way of controlling that at a higher level,” Moran said.

Cell-free metabolic reactions could be more common than previously thought and don’t need special conditions to get started, said Markus Ralser, a biochemist at Charité University Hospital in Berlin, who found some of the first enzyme-free metabolic reactions.

“This fits a bit into my thinking about how metabolism started in evolution,” he said of the new work. “If it would be very hard to do, then the planet would not be full of life now.” This idea is complicated, however, by the low-oxygen conditions in which life arose.

Another explanation for the observed results could be that enzymes, loosed from the irradiated cells, might be hanging around in the soil and continuing their biochemical jobs. Even when degraded, enzymes have stable backbones that might be capable of catalyzing reactions, said Sudha Rajamani, an astrobiologist at the Indian Institute of Science Education and Research, Pune who wasn’t involved in the study.

Ralser agrees with her. “My gut feeling is they still have a lot of enzymes there [in Fontaine’s irradiated soil], even after six years,” he said. To know whether metals and minerals in soil could spontaneously carry out these reactions, the researchers would have to eliminate enzymes from the mixture. But that’s really hard: They would have to get the soil so hot that it would damage the soil structure itself.

However, the activity of such enzymes diminishes “exponentially” after they spill out of cells, Bouquet said. Plus, no enzyme is known to last six years, Fontaine added. He doesn’t doubt that enzymes released by living and recently dead cells contribute to carbon emissions in real-world soils, but the long-term experimental results make it “very unlikely that the respiration we observed is due to enzymes,” he said.

For Bouquet, chasing this years-long obsession has highlighted that “even in a context as close and familiar to us as terrestrial soil, we are not always able to distinguish or recognize processes that indicate the presence or absence of living organisms.” Now a researcher at the Collège de France and the National Museum of Natural History in Paris, he is looking for prebiotic origins of other biochemical cascades.

“I find it particularly interesting to imagine the survival of processes that may predate life itself,” Bouquet said, “right there under our feet.”

Key Chemistry Question Answered, No Quantum Computer Required

Kevin Hartnett — Fri, 29 May 2026 01:54:11 GMT

Nash Weerasekera for Quanta Magazine

What Garnet Chan cares most about is basic science. He entered chemistry decades ago to understand some of the most consequential biochemical processes on Earth.

But since then, he’s become a central figure in a different arena: the debate over whether quantum computers will have a decisive advantage over ordinary “classical” ones. Over the past decade, many quantum computing researchers have identified the very reactions Chan studies as an area in which quantum computers should excel. Chan, however, has long doubted that powerful quantum computers — which are still years away — will be necessary.

“My main interest is in solving chemical problems. If classical computers are the right tool to do it, we should,” he said. While he believes quantum computers will eventually play an important role in the field, “I don’t see why we should wait for a fault-tolerant quantum computer to be built.”

Now he has a result that strengthens his case.

In early January, Chan and five other quantum chemists based out of the California Institute of Technology reached a key milestone in understanding the enzyme nitrogenase, which converts atmospheric nitrogen into ammonia and makes life on our planet possible. It was a major triumph for theoretical chemists, the outcome of decades of effort.

But for years, nitrogenase had also served as a proof-of-concept target in the realm of quantum computing. To understand the enzyme, researchers must follow the behavior of many electrons that are all linked together via quantum entanglement. The number of possible configurations grows explosively large. Researchers hypothesized that they would likely only be able to decipher the system via a machine that could manipulate quantum states.

But Chan and his colleagues used purely classical methods. That makes their result a pivotal statement not only about the chemistry that supports life, but also about whether quantum computers are needed to understand it.

“I think it’s important to clarify that this is not an impossible task where you have to first build a quantum computer to say anything about the problem,” Chan said.

While Garnet Chan is excited for the day when quantum computers will help solve important problems in chemistry, he sees no need to wait: Contrary to popular belief, he argues, quantum computers aren’t needed to answer some of the field’s biggest questions.

Jerry Camarillo Photography

Not everyone agrees. Some researchers cite the many years it took to obtain the result classically. Even if one chemistry problem has ultimately proved tractable with classical methods, they say, quantum computers are still needed to make these kinds of discoveries at scale.

“If we pick any optimization problem and you put 20 years into it, you can figure out that one system,” said James Whitfield, a quantum computing theorist at Dartmouth College. “But whether that solution is transferable? Questions like that won’t be answered by solving one instance of one molecular system.”

Solving this particular problem about nitrogenase may not settle the debate over quantum computers just yet, but each step toward understanding the enzyme’s full chemistry makes the debate less hypothetical.

Nature’s Ammonia Factory

Alongside photosynthesis, nitrogen fixation is one of the most essential chemical processes for life on Earth. Nitrogenase is what makes it possible.

Before nitrogenase evolved, living things were limited by the amount of nitrogen available to be incorporated into organic matter. It was an ironic obstacle, given that the planet was in fact suffused with nitrogen: The element accounts for about 80% of the atmosphere. But atmospheric nitrogen exists as the diatomic molecule N₂, which is inert and therefore unusable in biological processes. Only rare high-energy events could break the molecule into nitrates that life could use.

“Organisms were literally waiting for lightning to strike. That’s how you’d get nitrogen to be available for biomass,” said Daniel Suess, a chemist at the Massachusetts Institute of Technology who studies nitrogenase.

But 3 billion years ago, the nitrogen floodgates opened when nitrogenase evolved in early prokaryotes. The enzyme accomplished what no other biological process could do: It broke the triple bond holding N₂ together and converted the inert gas into biologically useful ammonia.

The enzyme was effective but extraordinarily complicated. That was beside the point to the early microbes that benefited from it, but it would come to matter enormously to the humans who, billions of years later, wanted to replicate its trick in order to make fertilizer.

Part of what makes nitrogenase so chemically difficult is its “active site” — a cluster of iron and molybdenum atoms called FeMo-co. Each iron atom carries four or five unpaired electrons whose behavior depends on that of the others. In fact, FeMo-co is one of the most correlated systems in all of biology, and a prime example of what’s known as the electron correlation problem: Because its electrons can’t be treated independently, it’s extremely hard to determine properties of the overall system, such as its true electronic structure and energy.

Mark Belan/Quanta Magazine

For most of human history, the pressing question wasn’t how nitrogenase worked — it was how to get enough of what it produced. As late as the 19th century, the most reliable source of usable nitrogen was guano harvested from islands off the coast of Peru, a resource so valuable and rare that nations went to war over it. Then the German chemists Fritz Haber and Carl Bosch cracked industrial nitrogen fixation in 1909, and the practical significance of the problem receded.

The scientific one — understanding how nitrogenase, tucked inside an ordinary soil bacterium, accomplishes what the Haber-Bosch process requires an industrial furnace to do — remained open.

It was an important question in its own right — and one that would achieve new prominence as people debated the best way to solve it.

An Unlikely Test

A classical computer processes information as bits, which take one of two values: either 0 or 1. A quantum computer instead uses qubits, which can exist in a superposition of 0 and 1 simultaneously and can become entangled with one another in ways that have no classical analogue. That means that when (or if) a large-scale quantum computer exists, it will be able to explore many possible solutions to a problem at once, rather than grinding through them in sequence.

For certain kinds of problems with the right mathematical structure, this promises an exponential speedup over anything a classical machine could achieve. The question, ever since quantum computing took off as a subject of theoretical study in the 1990s, has been which problems qualify. One of the most promising domains seems to be simulating chemical interactions: The electron interactions that govern how molecules behave are quantum mechanical at their core, which suggests that a quantum computer might be uniquely suited to modeling them.

The status of nitrogenase as an informal quantum computing benchmark traces back to a 2011 meeting Microsoft organized to explore applications for its nascent quantum computing group. Chan, who’d already been studying nitrogenase for more than a decade at the time, gave a talk on the enzyme.

He doesn’t know to what extent that talk influenced later events, but in 2017, Microsoft researchers published a paper in the Proceedings of the National Academy of Sciences arguing that the entangled complexity of nitrogenase made it a compelling test for quantum computers.

In Chan’s view, it was a strange fit from the start. He disputed the claim, continuing to believe that it was possible to model nitrogenase using classical methods like the ones he’d spent his career developing.

Over the next decade, he would get to work proving it.

Ground-State Debates

Chan and other researchers didn’t set out to explain how nitrogenase works end to end. Rather, they turned to a widely used computational model of FeMo-co and asked a more preliminary question: What is its ground-state energy?

The ground state — FeMo-co’s lowest-energy electronic configuration — is the starting point for the whole reaction. But FeMo-co contains a cluster of seven iron atoms, each with four or five unpaired electrons whose quantum “spins” can point up or down, whose orbitals can shift, and whose behavior depends on what the electrons around them are doing.

This makes measuring FeMo-co’s ground-state energy extraordinarily complex. There are more than 78,000 plausible configurations the electrons might be in; the ground state is a superposition, or a sort of weighted combination, of all these configurations. In principle, the Schrödinger equation tells you how all these different configurations contribute to the ground state and what its overall energy should be. But in practice, solving that equation directly and exactly for a system with as many interacting electrons as FeMo-co has is often impossible.

This is true for both quantum and classical computers. In both cases, you have to start with a simpler approximation of the ground state’s basic structure — an educated guess, often reached only after years of research, about which configurations are contributing the most to the ground state.

Fritz Haber (right) in his laboratory at the Kaiser Wilhelm Institute for Physical Chemistry in Berlin, alongside the chemical engineer Ladislaus Farkas. Haber developed industrial processes for mass-producing both ammonia fertilizer and chemical weapons.

Sueddeutsche Zeitung Photo/Alamy

Then, if you’re using a classical computer, you can try to progressively account for other configurations and show that you can safely ignore the huge number of remaining configurations because they don’t add much to the ground-state energy.

On the other hand, in theory, a quantum computer won’t require you to leave configurations out of your final estimate. Instead, the computer can represent your initial guess directly as a quantum state, and then evolve that state forward in time until it naturally reaches the right ground-state structure — allowing you to calculate the energy precisely.

Many researchers think quantum computers are at an advantage here, because the process of classically ruling out insignificant configurations can get prohibitively difficult. Chan and others, however, disagree. For one thing, they argue, quantum computers still encounter the same bottleneck of needing that reasonable initial guess, and there’s no obvious reason why quantum methods should have any advantage at clearing that bottleneck. Moreover, classical techniques have been rapidly maturing.

But for Chan, asserting that quantum computers might not be needed after all was “like trying to resist the ocean tide,” he said.

Sifting Out the Solution

Since receiving his doctorate from the University of Cambridge in 2000, Chan had been developing and refining ways to compress complicated quantum states by focusing only on their most important configurations. He and his team now hoped to apply these approaches to FeMo-co.

They used two different techniques to winnow down the configurations they needed to look at. Using one method, they started with their guess and incrementally adjusted the behavior of small numbers of electrons. They then showed that adjusting larger numbers of electrons didn’t lead to significant energy changes, giving them a clear recipe for which configurations they could ignore and which they couldn’t.

Their second method was the one that Chan had spent his career working on. It involved breaking their initial state into pieces and allowing only a limited amount of information to flow between those pieces. They then showed that they only needed to consider changes in that information flow up to a particular limit. “Realizing that the description could be achieved by ‘simpler’ methods and pushing these methods extremely hard (as the problem is still computationally challenging) was the key,” Chan wrote in an email.

Both methods produced the same energy estimate for FeMo-co’s ground state (and matched what scientists had observed experimentally), giving the researchers confidence that they had found the true ground state.

The Debate Shifts

Chan hopes that the technical breakthroughs his team made can now be extended to model the full nitrogenase enzyme and its reaction. “My hope is that all these people advocating ‘We need to build a quantum computer to solve the nitrogenase problem’ will join this mission now that we have a route to doing it,” he said.

But getting from the ground state to a full mathematical description of the reaction will be far more difficult, involving calculating energies for a whole sequence of intermediate chemical states. “We’re not even close to achieving the holy grail of this,” Suess said. “We’ve still just described the resting state. But the method is promising in that it suggests we can proceed with some confidence.”

It’s also unclear what the result might mean for researchers’ hopes for quantum computing. Whitfield argues that calculating a single ground-state energy value was never where quantum computers were expected to best their classical counterparts. Their likely advantage, he said, instead lies in that next question on the table: modeling how the system evolves over time. That’s likely to showcase how inefficient classical methods can get — and how much more powerful quantum computers can be.

After years of friendly sparring with the quantum computing community, Chan does not expect the new result to change many minds. After all, he said, quantum chemistry simulation via quantum computers still holds great promise: If a quantum computer were to become available tomorrow, he would gladly use it. But he hopes his team’s new result will help correct the misconception that the hardest chemical problems are simply out of reach until quantum hardware arrives.

“Science is self-correcting,” he wrote in an email, “but quite often, the corrections do not receive the same attention as the initial claim, because the field has moved on to other claims.”

How We See the Beautiful, Violent Sun

Simon Frantz — Thu, 28 May 2026 02:16:52 GMT

A solar flare captured by NASA’s Solar Dynamics Observatory in October 2014.

NASA’s Goddard Space Flight Center

The sun is one of the most studied objects in the history of science. The ancient Babylonians and Chinese tracked sunspots and solar eclipses, etching their observations into clay tablets; these records would outlast their civilizations. When the telescope arrived in the early 1600s, astronomers such as Galileo Galilei, Christoph Scheiner, and Johannes Fabricius turned these instruments toward this nearest star, projected the image onto paper, and saw dark blemishes drifting slowly across the solar surface.

In the 1800s, our ability to understand the sun’s composition launched a new era of solar science. Spectroscopy could split light emitted from objects into a kind of barcode that characterized elemental makeup. Armed with this method, Pierre Janssen and Norman Lockyer independently found lines in the sun’s spectrum that didn’t match any known element on Earth. Lockyer named it helium, after Helios, the Greek god of the sun. It would be another 27 years before Sir William Ramsay isolated and identified that element on our planet.

In the early 1900s, the pioneering American astrophysicist George Ellery Hale discovered that the sunspots that Galileo and others had traced weren’t blemishes but magnetic storms, regions of intense activity that waxed and waned on the 11-year solar cycle. The French astronomer Bernard Lyot built a coronagraph in 1930: a telescope with a disc at its center that blocked the sun’s blinding light, mimicking an eclipse on demand. For the first time, scientists could study the corona — the sun’s ghostly outer atmosphere — without waiting for the moon to cooperate.

From the 1950s on, the space age allowed scientists to create instruments that could escape the observational barriers of Earth. Satellites and probes began directly measuring the solar wind — the constant stream of charged particles the sun throws off in all directions — along with the violent phenomenon of coronal mass ejections, plasma founts that are some of the most energetic events in our solar neighborhood. Since 1995, the Solar and Heliospheric Observatory, a collaboration between NASA and the European Space Agency, has been on constant surveillance, and NASA’s Solar Dynamics Observatory joined the fold in 2010. The Parker Solar Probe first flew through the corona itself in 2021. Its pass in 2024 was the closest any human-made object has ever come to a star.

Observations and questions have continued to accumulate. Why is the corona hundreds of times hotter than the surface below it? What drives the solar cycle? How do the electromagnetic radiation bursts known as flares decide to erupt? The instruments keep improving, and the secrets they uncover continue to fascinate.

Published in 1613, Galileo’s Letters on Sunspots (Istoria e dimostrazioni intorno alle macchie solari) featured his observations of dark spots on the face of the sun, which he thought resembled clouds.

Public Domain

Around the same time, the Jesuit mathematician Christoph Scheiner developed a method for safely observing sunspots by projecting the sun’s image through a telescope onto a screen

Houghton Library, Harvard University

Scheiner argued that the sunspots he observed and recorded were satellites of the sun. Galileo disagreed, arguing that sunspots must reside on the sun.

Public Domain

On September 1, 1859, the English astronomer Richard Carrington spotted an unusual and sudden brightening on the solar surface, which he mapped out in this drawing. Seventeen hours later, the northern lights were visible as far south as Cuba, and telegraph systems across the Western world failed and even caught fire. The Carrington Event, as it became known, was the first documented case of a geomagnetic storm associated with a solar flare.

Public Domain

The American astronomer Samuel Pierpont Langley created this drawing of a sunspot in 1873. It has become an iconic image in solar science. Note the scale as indicated by the inset of the Americas at the upper left.

AIP Emilio Segrè Visual Archives

In 1919, the English astronomer Sir Frank Dyson organized expeditions to the West African island of Principe and the Brazilian town of Sobral to observe a total solar eclipse. The results showed that starlight bent around the sun, confirming a key prediction of Albert Einstein’s general theory of relativity. Einstein became world famous overnight.

W.B. Robinson

Another solar eclipse meant another expedition, this time on September 10, 1923. The Yerkes Observatory sent a team to Santa Catalina Island, California.

B.W. Harris Yerkes Observatory, University of Chicago

Launched in 1995 and still in operation, the Solar and Heliospheric Observatory orbits around a point on the direct line between the sun and the Earth, giving it an uninterrupted view. It revolutionized our ability to forecast space weather and provides detailed views of large solar flares such as this one.

ESA/NASA

The Earth-orbiting Solar Dynamics Observatory (SDO), launched in 2010, provided more high-resolution images of solar activity. This 2014 video shows a solar flare in a blend of two wavelengths of extreme ultraviolet light: 304 angstroms (red) and 171 angstroms (yellow).

NASA’s Goddard Space Flight Center

This time-lapse video captured a partial solar eclipse when the moon passed between SDO and the sun. (Watch for January 30.)

NASA’s Goddard Space Flight-Center/SDO

Solar Orbiter, a joint mission of the European Space Agency and NASA that launched in 2020, captured these two images in February 2021 and October 2023. As the sun approached its solar maximum (a year after the second image), observations revealed more explosions, dark sunspots, and swirls of super-hot gas.

ESA & NASA/Solar Orbiter/EUI-Team

A very large sunspot group transited the solar disk in October 2014, as captured by the SDO. This spot was part of NOAA 12192, the largest active region on the sun for almost a quarter of a century.

NASA’s Goddard Space Flight Center

In July 2025, NASA’s Parker Solar Probe — which, like the Solar Orbiter, orbits the sun and not Earth — took the closest-ever images of the sun, just 3.8 million miles from the solar surface, within the outer corona. This image shows the solar wind racing out from the corona.

NASA/Johns Hopkins APL/Naval Research Laboratory

All previous views of the sun have been oriented toward its equator, taken from the plane on which Earth orbits. Solar Orbiter has provided the first look at the sun’s south pole. In this image, we see the sun’s polar magnetic field in motion. The magnetic network on the solar surface leaves imprints in the chromosphere, between the sun’s surface and the corona. Over eight days of observations, Solar Orbiter measured the tracks of these imprints, which were elongated by the sun’s rotation.

ESA & NASA/Solar Orbiter/EUI-Team

When Quiet Undersea Volcanoes Turn Disruptive

Evan Howell — Tue, 26 May 2026 03:29:26 GMT

Rock formations emerge from the waves along the shore of Iceland.

Jonas Preine

Jonas Preine, a recently minted Ph.D. from the University of Hamburg, squinted at a computer screen in the lab of a ship as it bobbed in the North Atlantic near Iceland. The image before him just didn’t make sense.

It was June 2024, and Preine was among a crew of scientists who had set off from Reykjavik under slate-colored skies, trading their regular lives — family, friends, and the typical office environment — for cramped quarters and nausea on board the Meteor, a research vessel chartered for Expedition M201. They’d been lucky so far, enjoying relatively calm seas as they motored toward their destination, an unexplored deep-water basin dotted with volcanic shapes. The researchers carried reams of equipment: geophysical tools to collect seismic profiles of the Earth’s interior, cameras to image the ocean floor, and coring and dredging equipment to sample rocks and verify what might appear in grainy, partially processed computer images.

That first evening, about 100 kilometers from port, the team paused to test their geophysical tools in shallow waters. The seismic imagery they collected of the seafloor’s layered interior would lead them to an unexpected discovery, one that would complicate what we know about the usually sluggish volcanic fissures that lace the bottom of the ocean. Their findings could also be connected to mysterious islands from the recesses of history that witnesses said appeared suddenly, only to disappear later beneath the waves.

The Meteor transported the scientists on Expedition M201 to their research site off the coast of Iceland.

Jonas Preine

In the lab, Preine snapped some screenshots and shared them with the team. Six weeks later, on returning from the expedition, he needed little effort to convince the project’s lead scientist to stop in shallow waters to investigate what they’d found.

A Conveyor Belt of Lava

Iceland is a geological rarity. Here, on the world’s largest volcanic island, you can hike through gorges dividing the North American and Eurasian tectonic plates.

That’s because Iceland sits on a mid-ocean ridge, a vast seam where Earth’s crust tears apart and oceans grow in the expanding space between. The Mid-Atlantic Ridge began forming about 200 million years ago, as the supercontinent Pangaea broke apart at the end of the Triassic Period. Iceland emerged much later when a plume of unusually hot mantle rock arched the ridge up above the gathering waves.

Across the globe, mid-ocean ridges have a nondramatic style — nothing like Washington State’s explosive Mount St. Helens or southern Italy’s Mount Vesuvius, the destroyer of Pompeii and Herculaneum. So the team aboard the Meteor wasn’t expecting anything unusual when they passed over a submerged segment of the Mid-Atlantic Ridge called the Reykjanes Ridge. They just wanted to confirm that their equipment was in working order.

The crew switched the equipment on, pinging back X-ray-like images that revealed layers of the seafloor’s stone interior. “We decided to do two test profiles over the Reykjanes Ridge because it was logistically easy and potentially interesting,” said Preine, now a marine geophysicist at the National Oceanography Center in England.

In Iceland’s Thingvellir National Park, hikers can travel along a path between the North American and Eurasian tectonic plates.

mauritius images GmbH/Alamy

Seismic images of mid-ocean ridges typically show rough and jagged terrain, formed when lava oozes up into the cold ocean along faults or fissures and hardens suddenly into stone. But that’s not what Preine saw. Along the ridge were smooth mounds with steep sides and flat tops, their flanks draped in scattered deposits that looked like debris from an eruption above the sea surface. The formations reminded him of the topic of his doctoral dissertation, a submerged system of notoriously explosive volcanoes near Santorini, Greece.

Why did there seem to be explosive volcanoes along the usually quiet mid-ocean ridge? And why were their tops beveled flat?

A Fleeting Apparition

Oceans, vast and deep, remain largely unexplored. For generations, scientists could do little more to study the depths than dredge the seafloor, dragging buckets for whatever they could find. Only in recent decades have geophysical technology and deep-sea cameras provided glimpses of these mysterious worlds.

There are more volcanic eruptions in the oceans than on land, said Isobel Yeo, a volcanologist at the National Oceanography Center who was not involved in Expedition M201. “We just don’t know nearly as much about them.”

Here’s what we do know: Across most of the globe, the crushing weight of the abyss suppresses explosive eruptions at mid-ocean ridges. Most of the Mid-Atlantic Ridge lies at least 2,500 meters below the sea, where extreme pressure keeps volcanic gases from expanding and limits eruptions to quiet outpourings of lava.

When the expedition returned, more profiles and imaging made it clear that the team had stumbled on the boundary where that restraint lifts: The Mid-Atlantic Ridge changed character at around a depth of 300 meters. That transition could explain an incident from Iceland’s recent history.

Jonas Preine, now a marine geophysicist at the National Oceanography Center in England, is studying the Mid-Atlantic Ridge.

Courtesy of Jonas Preine

Without warning on November 14, 1963, new land emerged from the cold black waters off the coast of Iceland. Over the course of three years, a volcanic island spewed and sputtered as it rose 171 meters above the sea. The government of Iceland named the island Surtsey, after the Icelandic mythic fire god Surtur.

Throughout history, so-called phantom islands — many of them volcanic — have occasionally sparked frantic, even quasi-comedic claims and naval tensions. “There were humorous reports of the Royal Navy stopping to put a flag on them when they breached sea level, only to see them disappear by wave erosion,” wrote Neil Mitchell, a geophysicist at the University of Manchester, via email.

While some of these phantom islands emerged from more volatile volcanic zones, others, including Surtsey, appeared mysteriously along the gently oozing mid-ocean ridge. In total, historical records document at least 14 eruptions on the northern Reykjanes Ridge in particular over the last 1,000 years.

Preine and his colleagues felt they had a unifying explanation for Surtsey and the strange subsea volcanoes they’d observed. They could pinpoint a specific depth at which the pressure eased just enough to allow seawater in contact with lava to flash to steam, powering an explosive eruption that could breach the sea’s surface.

Surtsey, a volcanic island off the southern coast of Iceland, emerged from the sea in the 1960s.

Omikron/Science Source

The deposits that blanketed the volcanoes’ steep flanks were critical clues, Preine said. Water dampens how far debris gets thrown, as anyone who has ever tried throwing an object underwater can confirm. So the material from underwater eruptions settles close to its source. Once a volcano breaches the sea surface, though, it can scatter ash and rock debris much farther. This can go on for days, months, or even years. Then, once the magma chamber is exhausted, the sea takes over again.

In Iceland, the sea is a force. Preine said the region’s underwater flat-topped volcanoes are worn down to a uniform depth of around 40 meters below sea level — no coincidence, he argued, since North Atlantic storm-wave erosion only reaches down that far. “The key parameter here is depth,” said Ross Parnell-Turner, a geophysicist at Scripps Institution of Oceanography at the University of California, San Diego, who was not involved in the work of Expedition M201.

The influence of lower water pressure and the force of the Atlantic’s waves seemed to explain Expedition M201’s observations. But there was another factor to consider.

A Land of Ice

Páll Einarsson vividly recalls the day in 1963 when Surtsey began erupting. His father, a volcano-curious engineer, drove him to the airport, where he had convinced an airline pilot to take an impromptu flight over the new volcano. From the plane, father and son watched the plume of dark smoke emerge from the endless northern seas.

Einarsson went on to study volcanoes, among other things, and is now an emeritus geophysicist at the University of Iceland. When asked about Expedition M201’s findings, he said he was impressed by the team’s efforts but not fully convinced of their theory about how the flat-topped volcanoes formed. That’s because, for different reasons, similar volcanoes appear both in the deep ocean and on land.

Some 20,000 years ago, Iceland was covered with slow-moving glaciers. Today, glaciers remain across approximately 11% of the country, and the areas left behind are home to squat mountains called tuyas, or table mountains. Some scientists think the tuyas formed when rising magma collided with a thick ceiling of ice and melted it, triggering explosions. But like a paving stone above a cluster of mushrooms, the glaciers acted like a lid, preventing the volcanoes from growing too tall.

Tuyas are steep-sided mountains with flat tops found in just a few places in the world, including Iceland. They are sometimes called “table mountains.”

Hansueli Krapf/Wikimedia Commons

During the last ice age, the sea level was far lower than today, and Iceland’s glaciers extended across the exposed Reykjanes Ridge. That glacial advance appears to have reached roughly the area that now lies beneath about 300 meters of water, where Expedition M201 identified the flat-topped volcanoes. The research team initially wondered whether the submerged mounds could be drowned versions of tuyas.

They did find evidence against the glacier theory: The volcanic material appeared to have accumulated on top of abandoned glacial rubble, suggesting that the eruptions happened after the glaciers retreated. Most outside experts interviewed for this story say they’re convinced that these observations rule out the theory, but Einarsson would like to see more evidence.

Scientists know what they need to resolve this tension: rocks, or at least a good look at them from a submersible vehicle.

With explosive submarine eruptions, “it looks like someone dumped a truckful of volcanic sand over everything,” said Robert Sohn, a geophysicist at the Woods Hole Oceanographic Institution in Massachusetts who was not involved in the work. “It’s immediately obvious.” But Preine and company have only limited material from seafloor dredging in the area and not enough direct visuals to substantiate their theory.

It will take a while to gather new evidence, though, as deep-sea exploration plans are formulated years in advance. “You sail into the middle of nowhere based on a point that somebody put on a map, and you hope very much that they’ve put that point in the right place,” Yeo said. “You’re sort of stuck with the decisions past-you made.”

The science team for Expedition M201 deploys seismic equipment off the deck of the Meteor research vessel.

Jonas Preine

Preine is keeping an open mind. He’s not so quick to dismiss the influence of ice, though he thinks it may have played a different role. Like seawater, ice exerts pressure on the Earth’s crust, suppressing volcanic activity. When ice sheets retreat, as they did along the Reykjanes Ridge, that pressure is released, causing volcanic activity to spike. Preine said sampling rocks to determine the ages of the volcanoes would be “extremely interesting” in testing what remains a working hypothesis: that receding glaciers indirectly fueled a new volcanic era.

Eruptions Past and Future

As Surtsey so vividly demonstrated, the behavior of mid-ocean ridges can shift from calm to explosive under the right conditions. Scientists now suspect that this may have been more common in the past — and wonder when it could happen again.

Whatever researchers discover about the Mid-Atlantic Ridge, the pattern may extend far beyond Iceland. Shallow stretches of mid-ocean ridges, from the Azores to the Galápagos to the Red Sea, cross the same depth threshold. In those places, the slow conveyor belt of the deep may occasionally give way to something more volatile, building islands that briefly rise above the surface before waves grind them back again.

There’s a reason to give Iceland special attention, though. Since around 2020, a giant magma chamber has swelled under the Reykjanes Peninsula, the onshore limb of the Reykjanes Ridge, triggering earthquakes and sending lava oozing into the streets. In 2023, the fishing town of Grindavík’s about 3,700 residents evacuated, many perhaps for good.

Now scientists say that pressure is building again. “We are really in the middle of a very remarkable event right now,” Einarsson said.

Offshore, most of that activity remains hidden. But the same forces are at work, Preine said, and “the chances are not low” for another Surtsey to rise again. “A colleague sent me an email saying that there was actually an earthquake swarm on the Reykjanes Ridge this week,” Preine said in January. “Nothing big, but you never know.”

Correction: May 27, 2026
The original article said that Surtsey emerged from the Reykjanes Ridge. In fact, Surtsey is southeast of there. In addition, the article misstated the name of Páll Einarsson’s father.

How Ecotypes Harbor the Genetic Memory of a Species’ Past

Marlowe Starling — Thu, 21 May 2026 02:48:20 GMT

The green ecotype of Cristina’s timema, a species of stick insect, blends in with broad leaves. Other ecotypes of the same species are colored to blend in with narrower leaves. With genomics, scientists are answering century-old questions about how a single species can manifest such distinct traits.

Aaron Comeault

When she was a graduate student in the 1970s, the evolutionary biologist Kerstin Johannesson regularly walked the shores of a Swedish archipelago, scanning the ground for pebbles that moved: marine snails. Her adviser, a taxonomist, had tasked her with describing the species present there by documenting their traits. She noticed that snails with thicker shells stayed on the shore, while those with thinner shells seemed to prefer wave-battered rocks, and in between the two habitats were snails with intermediate shell thickness. While they seemed like distinct species, Johannesson couldn’t help but wonder whether these snails might instead be different types of the same one.

Around 50 years earlier, the botanist Göte Turesson had had a similar revelation in a similar setting. As he walked Sweden’s shores, he noticed that saltbush plants from different stretches of coastline had distinct traits — earlier or later flowering times, or shorter or longer stalks — and between habitats, those traits fell somewhere in the middle. He bred the plants in his home garden and found that these distinct traits had a genetic basis even though they arose from the same species. In 1922, he published his results and coined the term “ecotype” to describe a subpopulation of a species adapted to a hyperlocal habitat.

At that time, the definition of a species was even less clear than it is today. Genes were still theoretical, and the structure of DNA wouldn’t be discovered for another 30 years. Turesson “struggled to be accepted,” said Johannesson, now the director of Tjärnö Marine Laboratory at the University of Gothenberg. How can a species contain multiple distinct phenotypes — or sets of traits — without separating into two species? “He had quite a job to try to convince his colleagues that there were inherited differences and local adaptation within species,” she said.

It wasn’t until the early 2000s, when whole-genome sequencing became accessible to evolutionary biologists, that Turesson’s ideas about ecotypes could be tested. By comparing the DNA sequences of ecotypes across the tree of life — from marine snails to stickleback fish to stick insects and more — scientists can study adaptation and speciation, the process by which new species form, at a molecular level.

Since the 1970s, the evolutionary biologist Kerstin Johannesson has tried to understand how various forms of marine snail could possibly represent the same species.

Bo Johannesson

“It’s by far the most exciting time to be a biologist, ever, in my opinion — maybe with the exception of going right back to Darwin,” said Sean Stankowski, an evolutionary geneticist at University College London. “Even when we understood that organisms were programmed by genetic code, we could really never access that. Now, we’re looking at every single [nucleotide molecule] — A, T, G, and C [adenine, thymine, guanine, and cytosine] — in the genome.”

An analysis of genomic ecotypes by Johannesson, Stankowski, and other researchers explains how some species can maintain the DNA sequences for multiple adaptations, allowing evolutionary processes to effectively select among ecotypes as environmental conditions change — sometimes within only a few generations. The data also suggests that some canonically diverse groups of species, including Darwin’s finches, may not be separate species at all, but rather different ecotypes of the same species.

Ecotypes represent a kind of “genetic memory,” Stankowski said, that reflects a species’ history of survival in different habitats. “Genomics has just taught us everything that’s going on under the hood,” he said. Once scientists looked there, they found an engine for adaptation.

Genetic Memory

In March 1964, the largest earthquake ever documented in North America uprooted the Gulf of Alaska. Within four minutes, some of the gulf’s islands were lifted 38 feet in the air, rivers were closed off from the ocean, and freshwater lakes were created. A decade later, scientists were surprised to find three-spined sticklebacks thriving there.

The fish species typically lives in the salty ocean. Biologists might have expected it to die out in a freshwater lake. Instead, something more interesting happened. The marine sticklebacks, which are armed against ocean predators with bony plates, started to look and behave like their freshwater cousins, which have fewer plates. This change unfolded within decades, too fast for a new species to have formed.

Marine sticklebacks (top) are armored with bony plates (stained red), which provide protection from ocean predators. Freshwater sticklebacks (bottom) have fewer plates and swim faster than the marine ecotype.

Jun Kitano

What could have driven these rapid phenotypic changes? A 50-year study of genomic data, published in 2015, revealed that, tucked into genomes across the population, marine sticklebacks contained the genes necessary to survive in freshwater environments. These alternate genes occur sparingly across roughly 100 regions of the genome, said Catherine Peichel, an evolutionary geneticist at the University of Bern who studies the fish species.

The presence of this kind of genetic diversity — having multiple forms of the same gene that harbor different traits — is known as standing variation. Even in low numbers, those alternate genes have a chance to be expressed, as if natural selection could reach into the past and redeploy those genes when needed. Research from a different lab has shown that transplanted marine sticklebacks can switch to the freshwater ecotype in as few as 20 to 30 years. The emergence of and selection for a novel trait, on the other hand, would likely take far longer than that — if the fish even survived the initial shock of navigating a totally different habitat.

“It’s almost like populations have a genetic memory of their time spent in different environments,” Stankowski said.

Back in Sweden, Johannesson’s marine snails seemed to harbor this genetic memory, too. In 1988, a few years after she completed her doctorate, a rare algal bloom blanketed Scandinavian coastlines with a lime-green slime that killed almost all marine life. “All my snails, they were gone,” she recalled. But she turned the tragedy into an opportunity. Many isolated rock islands were left vacant, so she ran a natural experiment to see if she could trigger her snails to switch ecotypes.

The marine snail Littorina saxatilis has been accidentally described as a new species or subspecies more than 100 times. In fact, the species includes two ecotypes (side-by-side, top) — one larger with thick shells, the other smaller and thinner — adapted to different habitats.

The marine snail Littorina saxatilis has been accidentally described as a new species or subspecies more than 100 times. In fact, the species includes two ecotypes (side-by-side, at left) — one larger with thick shells, the other smaller and thinner — adapted to different habitats.

David Carmelet; Roland Carlsson

Large, thick-shelled snails had armor to protect them from predatory crabs onshore; smaller snails could more easily cling to wave-battered rocks. With the help of her then-3-year-old daughter, Johannesson collected hundreds of large snails and placed them on empty rocks exposed to the sea. As the years turned into decades, and as generations of snails came and went, the entire population became smaller, with thinner shells. In less than 30 years, the wave-battered ecotype had been selected for across the population.

A few years later, she would uncover the genomic features that made the snails’ rapid adaptation possible.

Shuffling Chromosomes

A given ecotype might require the expression of hundreds of genes. So how can selection act on all of them at once? Genomic studies have found explanations in chromosomal architecture.

During egg and sperm formation, genes can be shuffled between chromosomes (highly compact packages of DNA) in a process known as recombination. Some portions of DNA can be deleted or inserted. Chromosomes can break into two, or fuse into one. And entire segments of DNA can be flipped, in what is called an inversion.

Mark Belan/Quanta Magazine

An inversion happens when a portion of DNA from the chromosome breaks off, rotates 180 degrees, and plugs back into the chromosome in the reverse orientation. After inversion, a block of genes sits in one orientation on one chromosome, and in the opposite orientation on the other. This effectively prevents recombination from happening again in that region, and locks that group of genes together in a block. If those genes are somehow related, this can create a supergene, or multiple genes that act as a single unit. The snail traits for thick shells and evasive behavior to hide from crabs, for instance, become linked so that they will be inherited together in subsequent generations.

“It’s like if you had a deck of cards,” said Patrik Nosil, an evolutionary geneticist at the French National Center for Scientific Research who studies speciation and ecotypes of stick insects. “With normal genetics, you shuffle that deck completely — all 52 cards. Whereas with these chromosomal inversions, you have a part of the deck that refuses to shuffle, so you can never change the order of the cards in there. That’s the part that controls the traits that make the ecotypes different.”

This is what Peichel suspects happened in the sticklebacks. Even though they still mate across ecotypes (marine sticklebacks return to freshwater inlets to breed), past inversions ensure that the freshwater genes stay together and that the marine genes do too, she said. This differentiates the two ecotypes while maintaining species-wide gene flow.

Peichel’s lab is getting closer to confirming this hypothesis. Using the gene-editing technology CRISPR, her lab can flip an inversion back to its original orientation. As these sticklebacks reproduce, the genes in this region will be able to shuffle again, perhaps disrupting the suite of traits that form the ecotypes. “This would be some of the first proof of this idea that inversions actually bring together, and hold together, adaptive [genes] that distinguish ecotypes,” Peichel said.

Inversions aren’t always beneficial. They come with risks, including reproductive failure. For example, thousands of inversions have been identified in the human genome, and some can cause pregnancies to fail. But when they successfully hold groups of traits together, the rewards can be high. Indeed, Johannesson and Stankowski’s review surfaced inversions associated with ecotypes across the tree of life, including in plants, birds, fish, mammals, marine invertebrates, and insects.

“No one predicted that [inversions] would be as abundant as they are” in ecotypes, Stankowski said. “In evolutionary genetics, it’s probably one of the biggest realizations of the last two decades.”

And some ecotypes contain many chromosomal inversions. By the mid-2010s, Johannesson and her colleagues had identified nearly 20 different ecotype-related inversions in the marine snail genome. Interestingly, these same groups of traits are found across populations — in Spain and the United Kingdom as well as Sweden — even though these populations are isolated and don’t interbreed. Further analysis showed that some of these inversions are millions of years old, likely having occurred in a common ancestor.

Cristina’s timema has several ecotypes, fit for different leaf types. The stick insects sporting a stripe (top) blend in with narrow leaves, while green ones (bottom) match broad leaves.

Cristina’s timema has several ecotypes, fit for different leaf types. The stick insects sporting a stripe (left) blend in with narrow leaves, while green ones (right) match broad leaves.

Aaron Comeault

“These inversions, which are big chunks of the DNA, were not really obvious until we had a reference genome and a genetic map,” Johannesson said. “When this [result] came, it was like we got to the end of the story.”

To be clear, not all chromosomal inversions form ecotypes, and ecotypes aren’t necessarily formed by inversions. Indeed, another chromosomal recombination process that preserves blocks of genes can be found in California’s dry, shrubby chaparral mountains, where the walking stick insect Cristina’s timema (Timema cristinae) is made for camouflage.

One of its ecotypes has a stripe down its back and feeds on plants with narrow leaves; the other is bright green and stripeless, and feeds on broad leaves. Each ecotype blends in with its preferred food, which helps it avoid predators, Nosil said. In these stick insects, as in the sticklebacks, a chromosomal inversion keeps each ecotype’s group of traits together on the genome. But Nosil’s research has found that the chromosomal changes don’t stop there.

Translocation is another structural change to chromosomal DNA that can happen, by chance, during recombination. Genes from one end of a chromosome are relocated to the other end — an event that can get “really messy,” Nosil said. Some parts of the genome can disappear entirely, while other genes can be inserted into a new place.

There’s another explanation for why structural rearrangements may form ecotypes. Chromosome breaks are rarely tidy, and sometimes they can lead to beneficial new traits. “Then that trait is favored by natural selection,” Nosil said. This idea that structural changes can create new functional mutations and not just prevent recombination is almost always overlooked, he said. But his research into stick insects suggests that it can support the evolution of new ecotypes.

“Over the next years, that’ll become a big point of research,” he said, “to try to understand the relative importance of those two aspects of genomic rearranging.”

On the Origin of Ecotypes

Over the years, evolutionary biologists have debated whether ecotypes are a first step in the evolution of new species. Are thick- and thin-shelled snails, marine and freshwater sticklebacks, or striped and solid stick insects examples of species on their way to splitting into two?

“There’s been growing appreciation now that ecotypes might be a step in the direction of forming new species, but they might never actually get there,” Nosil said.

The evolutionary geneticist Patrik Nosil collects stick insects. His research shows how chromosomes can rearrange to switch between ecotypes.

Courtesy of Patrick Nosil

Speciation isn’t as clean-cut as the branches of a phylogenetic tree would have us believe. Evolutionary biologists increasingly view it as smoothly scrolling across a continuum, rather than a process with discrete steps. There is a gray area between species, and that’s where ecotypes sit.

The distinct marine and freshwater traits in sticklebacks, for example, suggest that they should diverge into two different species. Yet even after millions of years, they haven’t. Instead, when environmental conditions change from salt water to fresh water, evolution seems to select for one ecotype or the other without discarding the alternate set of genes.

Then again, the species concept has been controversial since the days of Darwin. What distinguishes one species from another, or a species from a subspecies, or an ecotype from simple variation, isn’t concrete. Genomics has both clarified and complicated the question.

“The problem with defining ecotypes and the issue of defining species — they’re actually quite analogous,” Stankowski said. “It really does boil down to cultural differences” in scientific perspectives, he added, such as how biologists are trained, what questions they ask, and whether they’re inclined to lump species together or split them up.

For instance, many taxonomists consider Darwin’s finches in the Galápagos to be distinct species. However, they can change their beak shape to match the available seeds within generations, and they can all mate with one another. By Stankowski and Johannesson’s criteria, Darwin’s finches “would all be ecotypes of a single species,” he said. The same goes for cichlid fish in east Africa’s Lake Victoria, another famous example of rapid evolutionary radiation. But that doesn’t mean we need to rethink On the Origin of Species.

“Darwin would be blown away at the progress that we’ve made,” Stankowski said. “The ecotype concept really does match quite closely with his view on species.”

Two Researchers Are Rebuilding Mathematics From the Ground Up

Konstantin Kakaes — Wed, 20 May 2026 02:52:19 GMT

Kristina Armitage/Quanta Magazine

Let’s start with what’s probably the most tired, overused joke in math: A topologist is someone who can’t tell a coffee cup from a doughnut. Both, you see, have a hole in them.

Topology is usually described as a sort of “rubber sheet” geometry in which two shapes are considered the same if one can be stretched or compressed into the other without tearing it. But this summary leaves out something essential: How do topologists, and the many other mathematicians who rely on their methods, rigorously account for all this stretching? They don’t look at a doughnut and a coffee cup, squint, and say to themselves, “Sure, I can intuitively see how to squeeze one into the other, so they must be the same.” Rather, they describe a shape in a way that can “forget” about distance while respecting the underlying structure in a flexible way, allowing it to bend and stretch.

When these “topological spaces” were developed over 100 years ago, they played a major part in the revolutions in logic and set theory that marked the boundary between 19th-century and modern mathematics. Their birth was a crucial waypoint on math’s inexorable march from the numbers and shapes that people encounter in everyday life into ever more abstract caverns of thought. Topological spaces have since become the foundation for huge chunks of mathematics. If you think of math as a skyscraper, topological spaces are concrete pilings, driven deep into the bedrock of common sense that all of math ultimately rests on.

But disconcertingly, topological spaces turn out to be extremely poorly suited for a big chunk of modern math: They are an awkward setting in which to do algebra, which is something mathematicians quite like doing.

For years, mathematicians figured they just had to live with the limitations of topological spaces. If you’re working on the 87th story of a skyscraper, fixing the foundations in the subbasement is a scary proposition.

But over the past decade, Peter Scholze of the Max Planck Institute for Mathematics in Bonn and Dustin Clausen of the Institute of Advanced Scientific Studies in France have sought to replace topological spaces. They have defined a new category of mathematical objects called condensed sets, which resemble a sort of infinitely fine dust and retain all the nicest properties of topological spaces without the drawbacks. Dust, it turns out, is a better foundational material than the pebbly, well-understood soil of topological spaces.

“They are solving a problem we didn’t know we had,” said Ravi Vakil, a mathematician at Stanford University and president of the American Mathematical Society, “because we already had what we thought were reasonable solutions.” As a result, “a whole slate of mathematics has become much simpler.”

It’s an ambitious project. The new definitions and concepts that Scholze and Clausen have introduced are powerful but also complicated and hard to learn. Scholze, for his part, is not sure how widely used they will become. On the other hand, he sees them as just the first step in a far bigger program to understand why numbers behave the way they do.

Doing math can be a little like rock climbing: Just as the route you take up a sheer face can incorporate creativity and even elegance in the way it sequences technical maneuvers, so too can a proof. Both traverse existing terrain. Most research — even some of the best research — takes the form of finding new routes to known peaks. But in mathematics, there is a weird relationship between the equipment and the landscape, as though developing a new type of ice axe causes hitherto unknown mountain ranges to emerge. As those new ranges appear on the horizon, older mountains that had seemed forbiddingly steep begin to resemble gentle hills.

Dustin Clausen, along with Scholze, has spent the past decade developing a new mathematical framework. Their “condensed mathematics” is already helping to connect topology, category theory, algebra, and other areas.

Christophe Peus/IHES

Developing these new tools takes a certain revolutionary confidence — especially when it requires setting aside implements that have been used in the community for so long that they seem to be part of the mountains themselves.

Point of Know Return

It’s possible to discover powerful mathematical truths without having a good language to work in.

Which is to say that topology predates topological spaces. As far back as 1735, Leonhard Euler proved that it was impossible to traverse the city of Königsberg by crossing each of its seven bridges only once. This is a recognizably topological result — the size of each of the city’s landmasses doesn’t matter, nor does the length of the bridges between them. Only the pattern of how they connect to each other does.

For nearly 200 years, research in topology proceeded in fits and starts. In the mid-19th century, August Ferdinand Möbius analyzed the strip that bears his name: a ribbon twisted on itself one time before its ends are joined. It is arguably the strangest topological object to have practical utility in the real world — for instance, in one-sided conveyor belts that wear evenly as they drive machines. Around the same time, Möbius began introducing some of the field’s key ideas, such as how to classify shapes with varying numbers of holes by looking at how loops can be drawn on them.

Shortly afterward, Bernhard Riemann, Henri Poincaré, and others made further advances. But they struggled for lack of the right language. As the Australian mathematician John Stillwell wrote in 2009 of Poincaré’s groundbreaking work in topology: “Along with great breakthroughs, there is also confusion.” Poincaré had ideas that he lacked the vocabulary to properly express.

To an outsider, it seems as if the branch of math closest to topology ought to be geometry. It seems as if topology is geometry, just with flexible objects instead of rigid ones. But the resolution to Poincaré’s confusion would come not from geometry, but from a nascent branch of logic called set theory.

At the turn of the 20th century, researchers were trying to wrestle mathematics onto firmer footing. They had only recently realized that their everyday intuition about numbers was completely wrong; now they were fervently debating which axioms, or obvious truths, they should build their theories on. Seemingly small differences in how they stated their most straightforward assumptions had major consequences for what would be possible or impossible to prove.

They used set theory to hash out these debates about the foundations of mathematics. In 1912, Felix Hausdorff, who had recently started teaching at the University of Bonn — where Scholze would end up generations later — set out to write the first comprehensive treatment of set theory. At the time, Hausdorff, then in his mid-40s, was already an accomplished writer: Under the pseudonym Paul Mongré, he had published a collection of poetry, two books of philosophy that tried to reconcile Nietzsche and Kant, and a play that was produced in 40 cities. As a mathematician, he was successful but not yet a superstar.

That changed after the 1914 publication of his book Fundamentals of Set Theory. In it, he gave the first description of topological spaces. A topological space is simply a collection of items that are grouped together into what Hausdorff called neighborhoods — today known as open sets. The open sets give structure to the space.

Open sets must satisfy just two conditions. First, any combination of open sets must also be an open set. (If Brooklyn forms a neighborhood and Queens forms a neighborhood, Brooklyn and Queens together must count as a single, bigger neighborhood.) And second, any finite overlap between open sets must also be an open set.

Mark Belan/Quanta Magazine

Topological spaces can be finite or infinite. They can encode intricate structure, or no structure at all. As Scholze put it, topological spaces are “just everywhere. If you have this intuitive idea of points being close to each other, you have topology.”

Consider a more familiar object: the number line. When we think about the way numbers relate to one another, we’re thinking about just one topological space: the so-called standard topology, in which every possible interval (not including its end points) forms a neighborhood, or open set. All these intervals are open sets:

This topology gives the real numbers the structure we’re used to.

But you can instead take the numbers you are used to dealing with, forget all the everyday intuition you’ve built up about them, and define entirely different topological spaces on them. If you think of each number as a book in a library, this would be like taking all the books off their usual shelves and organizing them in a completely different way.

For instance, you might crumple the number line into a ball and squeeze that ball down to a point, making every number arbitrarily close to every other number, like so:

This is akin to throwing all your books into a disorderly heap. The relationships between the books — historical fiction all being on the same shelf, and so on — are lost, because all the books are now neighbors. This is called the indiscrete topology, and we create it by declaring that there are only two open sets: the empty set and the entire number line.

At the other extreme is the “discrete” topology, in which every point forms its own neighborhood:

In this topology, instead of every point being in contact with every other point, no point touches any other one. It’s like putting every book in your library on its own private island. The relationships are again lost, because each book is isolated.

In a way, the old joke about the doughnut and coffee cup misunderstands the power of topological thinking. It’s not so much that it is possible to change distances by stretching or compressing things. It’s that it’s possible to meaningfully think about structure in spaces where distance simply does not exist.

In this way, topological spaces made it possible to explore novel parts of the mathematical landscape. For instance, they provided a new, distance-free way of understanding concepts like continuity and connectedness, which is a profoundly powerful and counterintuitive capability to have. This has allowed mathematicians to generalize those ideas to a much broader range of settings — and to prove important statements in a wide variety of fields. For example, results like the fundamental theorem of algebra, whose proof tripped up even mathematical giants like Carl Friedrich Gauss, end up with very simple proofs once topological arguments come into play.

The introduction of topological spaces also led mathematicians to ask new questions they might not otherwise have thought to ask. As is true of any good mathematical definition, topological spaces both opened new vistas and made it much easier to traverse known ones.

Hausdorff’s book was arguably the beginning of modern topology. As the mathematical collective known as Bourbaki would later write, his well-chosen definitions imbued “his theory with both the full precision and full generality desired. The chapter in which he develops the consequences of these axioms remains a model of axiomatic theory — abstract, yet anticipatory.”

Just a Drop of Water in an Endless Sea

There are many subdisciplines within modern mathematics. Each has its own vocabulary, grammar, and intellectual flavor. But they are never wholly separate — they interact in strange ways. In part, this is because many mathematical objects, like the real numbers, exist as objects of study in multiple disciplines: They have an algebraic structure, an analytic structure, a combinatorial structure, and a topological structure (among others!). Often, areas of overlap end up becoming distinct areas of study.

In 1945, two American mathematicians, Samuel Eilenberg and Saunders MacLane, published an audacious paper that created an entirely new discipline, now known as category theory. In one fell swoop, category theory created a set of expressways between other, existing areas of math.

Eilenberg and MacLane defined “categories” as collections of objects and the relationships (called morphisms) between them. For instance, a category might consist of sets and functions that relate these sets to each other, or of vector spaces and the linear maps that transform one vector space into another.

But the real power of the pair’s theory rested on the next level of abstraction they introduced: What if, they asked, you mapped one whole category onto another, with something they called a functor? A functor takes objects to objects and morphisms to morphisms, in an orderly way. In other words, it doesn’t just give you a way to translate from one group of objects to another; it also preserves the relationships between them — allowing you to connect different areas of math to each other.

Among other aims, Eilenberg and MacLane wanted to connect topology with the rest of math. Topology was already known not to gel particularly well with one of math’s most important areas: algebra. Although “topology has been this enormous gift to algebra,” said Clark Barwick of the University of Edinburgh, it has also “obstructed progress because topology and algebra don’t play all that nicely with each other, because of the particular way topology was built up.”

Category theory quickly goes from stating things that sound obvious to drawing powerful mathematical conclusions on the basis of what its practitioners typically call, with affection, “abstract nonsense.” Some categories have particular properties that make them more useful to mathematicians than others. In those categories, abstract nonsense becomes powerful — in other categories, it ends up just being nonsense.

In topology, you can define a category consisting of topological spaces and the continuous maps between them. In algebra, meanwhile, an important category is that of abelian groups — groups that have useful symmetries. If you want to produce a grand synthesis of the ways in which algebraic things behave topologically, or, conversely, how algebraic considerations govern topological constructions, the natural thing to do is to form a category out of objects that have both topological and algebraic structure, called topological abelian groups.

But topological abelian groups lack the particular properties that category theorists desire. If category theory unveiled a hitherto hidden network of highways between different mountain ranges of math, topologists who wanted to travel on that highway were stuck driving a janky car that kept breaking down and needing repairs.

That’s the problem that Scholze and Clausen wound up trying to solve. As Scholze told me, “I think topologists don’t actually like topological spaces, because it’s not a convenient category.” What if they could define new objects to replace topological spaces — ones that would retain their power, but also create a better kind of category that would finally allow mathematicians to connect topology to algebra and other fields?

The Nicest Revolutionaries

Sometimes new ideas seem ready to burst into the world.

In 2013, Scholze was 25 years old and already making waves as a deep mathematical thinker, hailed as the youngest full professor in Germany.

He and Bhargav Bhatt, then a researcher at the Institute for Advanced Study in Princeton, New Jersey, coauthored a paper in which they came up with a new definition for a particular type of category. In passing, the pair defined a somewhat abstruse mathematical set, a “sheaf on the pro-étale site of a point.” Scholze didn’t think much of it at the time. Those sets, which they didn’t name, “felt like a weird aspect of the story that I didn’t fully comprehend,” Scholze told me.

Bhargav Bhatt, along with Scholze, defined the objects that would later become known as condensed sets years before anyone realized their enormous theoretical significance.

Simons Foundation

At the time, Clausen was completing his doctorate at the Massachusetts Institute of Technology. After spending five years as a postdoctoral fellow at the University of Copenhagen, he moved to Bonn in 2018 to work with Scholze, who had just been awarded a Fields Medal, math’s highest honor. Clausen had independently come up with the same kind of set for different reasons, and he persuaded Scholze that they should study it more closely. By this time, Scholze recalled, Clausen had glimpsed a potential opportunity to replace topological sets.

Around the same time, Barwick and his then-student, Peter Haine, independently came up with a slightly different definition in order to answer a particular question in category theory that interested them. “We wanted to solve one problem,” Haine said. “We kind of understood that this theory should be useful for quite a lot of stuff, but there is one thing we really wanted to do: prove this kind of specific result which generalized what we had previously done.”

On the other hand, he said of Clausen and Scholze, “I think they had a lot more that they wanted to do.”

Indeed they did. They gave their sets a name — “condensed sets” — and got to work. They didn’t publish their incremental progress. Instead, in April 2019, Scholze began giving lectures on “condensed mathematics” at the University of Bonn; in May, he posted a 77-page set of notes that culminated in a new, elegant proof of an important theorem called coherent duality. As Johan Commelin, who would go on to collaborate with Scholze, remembers, “coherent duality had an extremely roundabout and technical proof before.” Scholze and Clausen’s proof was clean and elegant.

That, Commelin said, “gave people the motivation to say, ‘Let’s organize reading groups and study seminars all over the world to go through these lecture notes and find out what’s happening.’” Commelin organized one such group at the University of Freiburg, but it was challenging. “I don’t think anybody in our group understood all the details,” he said.

Although condensed sets had started as a tool for proving useful results, Scholze recalled, they quickly became something more. “For me personally, condensed sets changed something very basic about how I think about mathematics,” he said.

“Replacing the topological perspective by the condensed one,” he added, “creeps into everything I’m doing.”

Jeffrey Bergfalk, a set theorist at the University of Barcelona, remembers first meeting Scholze and Clausen in 2019 after he and his fellow set theorist Chris Lambie-Hanson of the Czech Academy of Sciences posted a technical paper online. Set theorists form a particularly small and tight-knit community, somewhat removed from the mathematical mainstream. “We were only expecting to hear from people that we knew,” Bergfalk said. But they got an email from Clausen, who had noticed the paper. He and Scholze, the email said, were thinking about similar things in the context of condensed mathematics — a nascent subject that neither Bergfalk nor Lambie-Hanson had heard of.

It took a moment for Bergfalk to realize the scale of Scholze and Clausen’s ambitions for condensed math. “Dustin and Peter were just wonderfully responsive, really cool and communicative,” he said. “They don’t have to be as nice as they are.” And even though Scholze and Clausen were working well outside their areas of expertise, Bergfalk noted, they asked about the sorts of things you wouldn’t expect anybody but a set theorist to care about. “They were asking us to think in the direction we would want to think anyway,” he said. “Which meant we had to learn condensed math.”

For two people who are reinventing a big chunk of 20th-century mathematics, Clausen and Scholze are unassuming. “To a large extent, what I am doing is rephrasing what others have done in my own words,” Scholze told the mathematician Maria Yakerson in a 2021 interview. “I’m not that much interested in theorems or proofs.” What he wanted to do, he said, was to come up with new definitions: “They must make it easy to state interesting theorems, and they must make it easy to prove them.” Scholze doesn’t see himself as creative. He is, he said, just “trying to give names to what is there.”

Clausen, for his part — as he told Yakerson in a separate interview around the same time — avoids publishing papers, because he believes that the scientific publishing industry is fundamentally flawed. He also largely avoids even informally writing up results, leaving that to collaborators. He just wants to focus on the math; like Scholze, he’s constantly looking for the right names, the right language. (At one point, in fact, he considered pursuing a career in literary translation.)

“I was never completely convinced by topological spaces,” Clausen said. They couldn’t give him an understanding of “this world that’s there, this rich world that we’re trying to get at but we don’t have the proper language to talk about.”

But that only motivated him further. “I’m extremely happy not understanding,” he said, “because I’m even happier when I finally do understand.”

Building on a Foundation of Dust

That’s where condensed sets come in. Condensed sets can be seen as a sort of recipe for building continuous objects, such as the real numbers, out of “totally disconnected” spaces — like making a cake out of disparate grains of flour and sugar, in Scholze’s telling.

Take the so-called Cantor set. Start with the line segment containing all the real numbers between 0 and 1, and remove the middle third. Then remove the middle third from the remaining line segments. Repeat this process infinitely many times, and you’ll end up with a “dust” of points. No point is right next to any other. The space of points is totally disconnected.

The Cantor set is the simplest kind of condensed set, as well as a building block for making other condensed sets. You can make more complicated condensed sets, Scholze said, by smashing together clouds of points like the Cantor set in a weird way.

Such dust might seem foreign, but Scholze points out that we use it all the time. When you represent numbers as decimal expansions, for instance, you’re essentially thinking of the numbers as a similar kind of dust. It’s like taking each number in the expansion and cutting out its section of the number line, like so:

In this way, producing a given number actually involves infinitely “dissecting” the number line. As Scholze put it, the “decimal expansion describes a totally disconnected space because with every new digit, you are chopping up your line more and more.” Every number is totally disconnected from every other one.

How, then, can you use such a disconnected set to get a continuous object like the real number line we’re so used to? You have to glue all the disconnected segments back together by equating, say, 0.49999999999999999… with 0.5 (and 0.50999999999999999… with 0.51, and so on).

Scholze and Clausen’s condensed sets work similarly: They’re disconnected, but they can be used to build and study continuous objects, like those you want to understand in topology. And if you start with them instead of topological spaces, you get an additional benefit: It turns out, Scholze explained, that “these totally disconnected pieces are extremely simple to describe algebraically.”

Condensed sets form a special type of category that, according to Juan Esteban Rodríguez Camargo, a collaborator of Scholze’s at the Max Plank Institute for Mathematics, finally makes it possible to mix topology, algebra, and other fields “in a very practical and precise way.”

Scholze and Clausen started by using their condensed sets to re-prove old results that previously relied on topological spaces — like the fundamental theorem of algebra. These new proofs have given mathematicians a novel understanding “suitably oiled and greased up and made supple,” Vakil said. “The more you can fit into your intuition, the better you understand.”

Then the pair decided to push even further.

A Condensed History

Since 2019, Scholze and Clausen have kept building new types of structures out of their condensed sets — and sharing new sets of lecture notes. “The ideas were evolving in Bonn way faster than the rest of the world could consume them,” Commelin said. There were “light” condensed sets, and also solid, then liquid, then gaseous spaces — an entire condensed mathematics.

Neither Clausen nor Scholze thinks of himself as a topologist. If the two were less friendly, or their ideas less effective, there might be some resentment over their attempt to rebuild the foundations of a field they don’t tend to work in. “I wouldn’t want to impose anything,” Scholze said when asked what he thinks the impact of condensed math will be. He and Clausen sound as though they are having fun, trying to come up with ideas that they themselves find useful.

But some mathematicians are likening their work to a similar mathematical revolution that took place in the 1950s and ’60s — when Alexander Grothendieck reimagined the field of algebraic geometry in accordance with category theory, vastly extending its reach and power. Grothendieck’s impact on modern mathematics was profound. And now, according to Dagur Asgeirsson, a postdoctoral researcher at the University of Alberta and Clausen’s former graduate student, “I think it’s fair to compare Peter to Grothendieck in this sense. He is reinventing everything somehow.”

“The real excitement about condensed stuff for me is the possibility of defining new objects of study,” Barwick said. It’s “showing you there is this natural class of objects that you just never looked at before, like an unclimbed mountain. We are just chewing at the corners of this vast territory.”

In one set of lecture notes, Clausen and Scholze quoted a well-known saying by the prominent mathematician David Mumford. Mumford’s field of algebraic geometry, he said, “seems to have acquired the reputation of being esoteric, exclusive, and very abstract, with adherents who are secretly plotting to take over all the rest of mathematics.” Clausen and Scholze went on to note that their plan was to use condensed math — also esoteric, exclusive, and abstract — to continue where this effort had left off. They were not entirely joking.

Regardless, their not-so-secret “takeover of mathematics” is continuing. In the last few years, Clausen and Scholze have defined other novel mathematical objects, such as “analytic stacks” and “gestalten.” Some mathematicians consider these to be even more significant than condensed sets.

Scholze suspects that condensed mathematics might prove useful even in areas distant from his and Clausen’s core interest in number theory. Quantum field theory — a central aspect of modern physics, which has long struggled with its foundations — makes use of very sophisticated algebra and category theory, Scholze noted. “At the same time,” he added, “quantum field theories are by their nature very analytic and topological. Mixing these two worlds is a nontrivial matter, but condensed mathematics gives a possible framework to do so.”

Scholze and Clausen’s body of work shows just how much choosing the right language matters — how reframing concepts makes it possible to traverse known terrain more easily and to explore new mathematical vistas. “Trying to get to the bottom of these phenomena is trying to find a language in which to express them,” Scholze said.

In a memoir Grothendieck published late in life, he described mathematicians as builders, even though he argued that they are most certainly not inventing anything, only finding structures that are already there, waiting to be discovered. He wrote: “The most beautiful house, that in which the love of the builder is most evident, is not that which is larger or higher than the others. Rather, a house is beautiful if it faithfully reflects the structure and beauty hidden in things.”

Correction: May 20, 2026
Peter Scholze is now at the Max Planck Institute for Mathematics in Bonn, not the University of Bonn.
Correction: May 26, 2026
The story has been updated to remove a parenthetical that incorrectly stated that Leonhard Euler lived in Königsberg.

How Alexander Grothendieck Revolutionized 20th-Century Mathematics

Konstantin Kakaes — Wed, 20 May 2026 02:50:37 GMT

Mercedes deBellard for Quanta Magazine

What Albert Einstein was to 20th-century physics, Alexander Grothendieck was to 20th-century mathematics. He is much less well known because math gets technical even more quickly than physics does. But as with Einstein, Grothendieck’s impact came not just from his own results, revolutionary though they were. His work also reoriented his entire discipline in radical new directions.

Grothendieck was intense and ascetic from his early days. Starting in the early 1950s, when he was in his 20s, he produced thousands of pages of formal and informal notes that changed the course of mathematics. Then in 1970, he quit. He left his post at a prestigious research institute just outside of Paris to teach at the provincial university in Montpellier where he studied as an undergraduate. He mostly stopped talking to other mathematicians. In the early 1990s, he moved to a small village in the Pyrenees, where he lived as a hermit.

Mathematicians are still grappling with the innovations he made half a century ago. His work pushed mathematics to a new level of abstraction by focusing on the relationships between objects rather than the objects themselves. “If there is one thing in mathematics which fascinates me more than any other (and undoubtedly always has), it is neither ‘number’ nor ‘size,’ but invariably shape,” he wrote in his memoirs. “And among the thousand and one faces under which shape chooses to reveal itself to us, that which has fascinated me more than any other and continues to do so is the structure hidden in mathematical things.”

His revolutionary mathematics centered around that search for hidden structure.

Revealing Shapes

Grothendieck is most famous for his work in algebraic geometry. The field first developed as the study of shapes defined by polynomial equations — equations that add together variables raised to fixed powers. These can be as simple as a line (x – y = 0) or a circle (x² + y² – 1 = 0). But as you consider more and more variables raised to higher powers and also look for solutions that satisfy sets of many equations instead of just one, things quickly get more complicated — and more abstract.

Grothendieck, seen here in 1954, was fascinated by hidden geometric structure. “If there is one thing in mathematics which fascinates me more than any other (and undoubtedly always has), it is neither ‘number’ nor ‘size,’ but invariably shape,” he wrote.

Paul R. Halmos photograph collection, e_ph_08592_pub, The Dolph Briscoe Center for American History, The University of Texas at Austin

The discipline took flight in the late 19th century, when mathematicians started asking questions about what happens if instead of plugging ordinary numbers into your equations, you plug in numbers from other, more abstract sets.

Before Grothendieck, algebraic geometry was an interesting and vibrant subdiscipline within mathematics. But it was also somewhat in crisis, as the mathematician David Mumford later wrote. “Every researcher used his own definitions and terminology, in which the ‘foundations’ of the subject had been described in at least half a dozen different mathematical ‘languages.’”

Then “Grothendieck came along and turned a confused world of researchers upside down, overwhelming them with [a] new terminology … as well as with a huge production of new and very exciting results.”

Grothendieck is most famous for introducing mathematical constructions that helped him and others prove longstanding conjectures, and that eventually became central objects of study in their own right.

His work also put algebraic geometry in the center of a web of many other areas of math — among them topology, number theory, representation theory, and logic. “Grothendieck never worked directly in number theory,” said Brian Conrad of Stanford University, “but the ideas he introduced into algebraic geometry totally transformed how number theory is done.”

His first major result in algebraic geometry was his 1957 generalization of the Riemann-Roch theorem, a proof from a century earlier that dictates how the shape of a surface limits which functions can be defined on it. As Leila Schneps of the French National Center for Scientific Research wrote, Grothendieck’s proof “propelled him to instant stardom in the world of mathematics.”

Thanks to his techniques, “a whole new wealth of operations becomes available,” Conrad said. “It opens up a whole new way to think about why the theorem is true.”

Then, just as quickly, Grothendieck moved on to the next thing. At the 1958 International Congress of Mathematicians, he announced his intention to remake all of algebraic geometry. He was going to do it with something called a scheme.

A New Scheme of Mathematics

A decade earlier, the mathematician André Weil had conjectured a link between solutions to polynomial equations defined in two very different mathematical settings. The first was finite fields, number systems that operate according to a cyclical form of arithmetic. The second was complex numbers, which take our familiar, everyday numbers and add the square root of -1, called i.

Weil made four conjectures that related polynomials from one setting to those from the other. These conjectures, Conrad said, “sound like communication between parallel universes.”

André Weil posed four conjectures that not only became foundational pillars of algebraic geometry, but also linked the field to other major areas of study, including number theory.

Courtesy of Sylvie Weil

As part of the effort to prove these conjectures, Grothendieck proposed his notion of a scheme. The attempted proofs were “a primary motivation for the theory of schemes,” said Daniel Litt of the University of Toronto, but “what it really bought you was a whole lot more.”

Before Weil, mathematicians only really talked about equations like x² + y² – 1 = 0 by specifying the particular number system they wanted to work in. The solutions to such equations would look quite different if x and y could only be integers, for example, versus if they could be any real number, or any complex number.

After Grothendieck came up with an explanation for why Weil’s conjectures are true, mathematicians came to believe that equations had meaningful structure independent of whether x and y were complex numbers, or elements of a finite field, or bananas. At first, this belief seems to make as little sense as saying that a sentence has meaning regardless of which language you choose its words from. But Grothendieck defined mathematical structures that made it possible to make such statements rigorous and even intuitive to those who mastered his new language.

As Conrad explained, “Grothendieck found the right way to define abstract notions of space, new ways of thinking about spaces.” He recognized that “the way you probe the geometry of a space is not by looking at the points, but by studying other things.”

That’s where Grothendieck’s schemes came into play. It takes some effort to construct even a simple scheme. But if you read on, it’s possible to understand what schemes are and develop an intuition for why they’re useful.

Schemes are geometric spaces that are built out of abstract algebraic ingredients.

Start with an abstract generalization of the integers called a ring. A ring is a set of elements that can be added, subtracted, and multiplied together, but that can’t always be divided. (In the ring of integers, for instance, you can’t divide 2 by 3, because ²/₃ isn’t an integer.)

Grothendieck fathered five children. Here he holds his fourth, Mathieu, who was born in 1965.

Shutterstock

Now look at a subset of your ring that is “closed,” meaning that if you add or subtract two elements of the subset, the result is also in the subset. For example, take all multiples of 5. This subset is not only closed, it has another property: You can multiply any number in the ring by an element in the subset, and the result is inevitably also in the subset. That makes the subset what mathematicians call an ideal.

Moreover, if you multiply any two numbers from the ring and end up in this subset (3 × 5 = 15), then one of the numbers you multiplied (5) must have been in this subset, too, even though the other number (3) isn’t.

This second property makes the subset a prime ideal. (To see why, look at the multiples of 6. These form an ideal, but not a prime ideal, because 2 × 3 is in the ideal, but neither 2 nor 3 is.)

In the case of the integers, the prime ideals are sets of multiples corresponding to each of the prime numbers, along with zero. It’s possible to study the set of all the prime ideals of a ring as a single geometric space. First, represent each prime ideal as a point. Then define a “topology” on those points that puts them into neighborhoods, depending on their shared elements. (Strangely, the zero ideal ends up being “close” to every single prime, illustrating a previously unknown structure hidden behind the integers.)

Grothendieck’s innovation was to add a layer on top of this space — a recently discovered mathematical superstructure called a sheaf, which carries additional algebraic information.

At each point in your space, for instance, this sheaf attaches another set, called a stalk. Let’s return to one of the prime ideals of the integers: the point in our space representing the subset of all multiples of 5. The stalk attached to this point would contain all the fractions whose denominators are not divisible by 5. (The stalk attached to 0 contains all possible fractions.) In this simple example, it’s hard to see what the stalks accomplish, but in more elaborate schemes, computing the contents of stalks and the ways they interact with each other turned out to be a mathematically powerful machine.

Mark Belan/Quanta Magazine

This entire object — the space of prime ideals, with the sheaf (and all its stalks) built on top of it — is called an affine scheme. In general, schemes are constructed by gluing affine schemes together in a precise mathematical way.

So what does all that have to do with an equation like x² + y² – 1 = 0? Well, instead of starting with the ring of integers, you can study a particular ring associated with that polynomial. You can then build the scheme for that ring.

But crucially, the variables x and y can be whatever you want them to be: integers, real numbers, complex numbers, elements of a finite field. By studying the scheme’s properties, you can gain insight about the structure of the equation, independent of any particular number system. It is, impossible though it may sound, a way to study the sentence apart from the language its words are written in.

Grothendieck lived the final decades of his life as a recluse in the French countryside. This 2013 photo was taken a year before his death.

Peter Badge

Broadly speaking, this is why Grothendieck and others could use schemes — and a series of ideas building on them — to re-prove one of the four Weil conjectures and prove two more. (Grothendieck’s student Pierre Deligne would later use other structures that Grothendieck developed to prove the fourth, which is a version of the famous Riemann hypothesis in the setting of finite fields.) Grothendieck continued to come up with even more abstract and powerful concepts, including topoi, stacks, motives, and étale cohomology. All play a major role in algebraic geometry and other areas of math today.

Schemes gave mathematicians a novel, systematic way to study the relationships between objects in algebraic geometry. And because schemes allow you to study rings, which appear all over math, as geometric spaces, they can be used to import geometric techniques into algebra, number theory, and beyond.

Grothendieck died in 2014 after years of solitude, estranged from the mathematical community he had helped create. Nonetheless, mathematicians remember him with reverent affection. As the Harvard mathematician Barry Mazur wrote, “During the early ’60s, his conversations had a secure calmness. He would oﬀer mathematical ideas with a smile that always had an expanse of generosity in it … a sense that ‘nothing could be easier in the world’ than to view things as he did.”

His ideas were complicated, but “most of the arguments are very straightforward once you set things up,” Litt said. “You just keep going and going and going. He found us the highway.”

What Do Gödel’s Incompleteness Theorems Truly Mean?

Natalie Wolchover — Mon, 18 May 2026 03:14:42 GMT

Kurt Gödel Papers, the Shelby White and Leon Levy Achives Center, Institute for Advanced Study; Samuel Velasco and Michael Kanyongolo/Quanta Magazine

In 1931, by turning logic on itself, Kurt Gödel proved a pair of theorems that transformed the landscape of knowledge and truth. These “incompleteness theorems” established that no formal system of mathematics — no finite set of rules, or axioms, from which everything is supposed to follow — can ever be complete. There will always be true mathematical statements that don’t logically follow from those axioms.

I spent the early weeks of the Covid pandemic learning how the 25-year-old Austrian logician and mathematician did such a thing, and then writing a rundown of his proof in fewer than 2,000 words. (My wife, when I reminded her of this period: “Oh yeah, that time you almost went crazy?” A slight exaggeration.)

But even after grasping the steps of Gödel’s proof, I was unsure what to make of his theorems, which are commonly understood as ruling out the possibility of a mathematical “theory of everything.” It’s not just me. In Gödel’s Proof (a classic 1958 book that I heavily relied upon for my account), philosopher Ernest Nagel and mathematician James R. Newman wrote that the meaning of Gödel’s theorems “has not been fully fathomed.”

Maybe not, but six decades have passed since then. Where are we with these ideas today? Recently, I asked logicians, mathematicians, philosophers, and one physicist to discuss the meaning of incompleteness. They had plenty to say about the implications of Gödel’s strange intellectual achievement and how it changed the course of humanity’s unending search for truth.

PANU RAATIKAINEN, philosopher at Tampere University and author of the Stanford Encyclopedia of Philosophy entry on Gödel’s incompleteness theorems

Ever since the ancient Greeks, the axiomatic method has been widely taken as the ideal way of organizing scientific knowledge. The aim is to have a small number of “self-evident” basic propositions — axioms, principles, or laws — such that all truths of the field in question can be logically derived from them.

Gödel’s incompleteness theorems show with mathematical precision that this ideal necessarily fails for large parts of mathematics. The whole of mathematical truth concerning even just positive integers (1, 2, 3 …) is so perplexingly complex that it does not follow from any finite set of axioms.

This means that some mathematical problems are not even in principle solvable by our current mathematical methods. Progress may require creative conceptual innovation. As a result, mathematical truths do not make up a unified whole of equally indubitable truths; instead, their status as knowledge varies gradually from doubtless facts to increasingly uncertain hypotheses.

Raatikainen makes a good point that Gödel’s theorems muddy the waters between where objective truth ends and invented math begins. One historical way people have tried to overcome the limitations of Gödel’s theorems has been to propose additional axioms beyond the commonly accepted ones. Say you want to prove a statement with the traditional axioms, but you find that you can’t — that it is undecidable. If you add a new axiom to your starting set, you may then be able to prove the statement true. Adding a different axiom, however, and you may be able to prove it false. So whether it’s true or false depends on the choice you’ve made. Suddenly, “truth” is more contingent on one’s preferences or assumptions.

REBECCA GOLDSTEIN, philosopher and author of Incompleteness: The Proof and Paradox of Kurt Gödel

Intuitions have always played an important role in mathematics. After all, we can’t prove everything; we need to accept some truths (i.e., the axioms) without proof in order to get our proofs off the ground. But we’ve learned over the centuries that sometimes intuitions prove unreliable — so unreliable as to generate actual paradoxes — meaning we’re driven to assert out-and-out contradictions.

In the early 20th century, Bertrand Russell and Alfred North Whitehead were working on The Principles of Mathematics, which attempted to reduce arithmetic to logic. [The view that math is nothing but logic is known as “logicism.”] The work led Russell to the discovery of what came to be called Russell’s Paradox. It concerns the set of all sets that aren’t members of themselves. The paradox reveals itself when you ask: Is this set a member of itself? The contradiction: If it is, then it isn’t. And if it isn’t, then it is. (Georg Cantor, considered the founder of set theory, had already realized the contradiction back in the 1890s.)

The response of mathematicians — most forcefully David Hilbert, the leading mathematician of that time — was to rid mathematics of iffy intuitions by way of formally axiomatizing mathematics into a consistent and complete set of algorithmic, recursive rules, essentially reducing math to a mechanical game of symbol manipulation. This goal of formalization was christened the Hilbert Program.

What Gödel proved was that the Hilbert Program was unrealizable. His first incompleteness theorem states that in every formal system of mathematics that is rich enough to express arithmetic, there will be propositions that are both true and unprovable. So, although formal systems comprised of mechanical rules of symbol manipulation successfully eliminate all intuitions, they also fail to capture all that we know to be mathematically true — a knowledge enriched by intuitions concerning the infinite structures that we call numbers.

It’s fascinating that our intuitions about numbers might go beyond what we can prove.

Personally, my intuition is silent on the mathematical statement that, in the years after Gödel’s proof, made incompleteness real. It is called the continuum hypothesis, and it asserts that the set of all real numbers (the continuum) is the second-smallest infinite set after the set of natural numbers (1, 2, 3 …). It was found to be undecidable using the standard axioms of mathematics. Extra axioms can be engineered to establish it as true or false, but logicians are divided on which way to go.

A physicist I spoke with warns that the undecidability of the continuum hypothesis has implications for his field: that physicists might need to avoid the continuum altogether.

CLAUS KIEFER, physicist at the University of Cologne, author of a 2024 paper on the relevance of Gödelian incompleteness for fundamental physics

Kurt Gödel’s proof has far-reaching and unexpected consequences for mathematics. Given that physical laws are formulated in mathematical language, is it relevant for physics, too? I think yes.

Among the most important undecidable statements is the continuum hypothesis (CH), proved to be undecidable in the Gödelian sense by Paul Cohen in 1963. The name “continuum” comes from the postulate to identify the points on a line with the real numbers. But how many real numbers are there? There is an uncountable infinity of them, but can this uncountability be specified? The CH states that the real numbers form the next-smallest infinite set after the infinite set of the natural numbers, which are countable.

Now consider that the known fundamental interactions in physics are defined on a space-time continuum. The uncountable number of points associated with this continuum is responsible for various problems in physics. In Einstein’s theory of general relativity, for instance, our modern theory of gravity, it leads to singularities that prohibit the mathematical description of the universe’s origin and the interior of black holes. In the Standard Model of particle physics, described by a quantum field theory, direct calculations yield infinite results for energies and other physical quantities, which must be eliminated by a sophisticated and nonintuitive mathematical procedure.

The situation becomes more severe in the push for a final unified theory of all interactions. A unified theory should be characterized by a consistent and complete mathematical language. But if a unified theory were to describe space-time as a continuum, the CH may render the theory incomplete. Physicists have already shown that the CH leads to undecidable questions in quantum field theory, such as whether certain atomic systems have an “energy gap,” enabling them to settle into stable ground states. This undecidability stems from the fact that the calculation assumes the atoms inhabit a space-time continuum. One may argue that a more fundamental theory (with more complete axioms) could decide the question, but the final theory should not have undecidable statements. So it should not involve a continuum.

In my opinion, this situation of undecidability can only be avoided if the structure of space and time is discrete — that is, characterized by a countable infinity of points only. There are hints for a discreteness in some approaches to quantum gravity, for example string theory or loop quantum gravity, but the situation is far from clear.

It’s worth noting that on top of these troubles with the continuum hypothesis, high-energy physicists have many other reasons to think a continuous space-time is not fundamental to reality, but rather only a long-distance illusion that emerges from other parts.

JOUKO VÄÄNÄNEN, mathematician and logician at the universities of Helsinki and Amsterdam

Incompleteness is an unwelcome but unavoidable fact of life in mathematics, like irrational and transcendental numbers in number theory, or Heisenberg’s uncertainty principle in physics.

There is a kind of “Gödel barrier” that formal language cannot circumvent: The stronger the expressive power of a logic (meaning the more things you can say in the logic), the weaker is its effectiveness (meaning our ability to prove statements true or false in the logic), and the stronger the effectiveness, the weaker is the expressive power.

For example, one of the simplest logical systems is propositional logic, which lets you combine statements with operations such as “and,” “or,” and “not.” It is very effective, but its expressive power is weak. On the other end of the spectrum, there’s second-order logic, which lets you make statements about objects, properties, sets, and relationships. It has tremendous expressive power and very weak effectiveness. It is as if the “product” of effectiveness and expressive power were constant, just as in Heisenberg’s uncertainty principle, which says that there is a limit to the precision with which certain “complementary” pairs of physical properties, such as position and momentum, can be simultaneously known; in other words, the more accurately one property is measured, the less accurately the other property can be known. In logic, in a remarkable analogy, effectiveness and expressiveness are such “complementary” properties. This is the real content of Gödel’s incompleteness theorems.

We stumble forward in mathematics without any certainty of consistency or completeness. This is just how things are.

It is shocking that mathematics, which is the basis of exact sciences, lacks a foundation that can be proved to be consistent and complete. Hilbert can be forgiven for thinking that this cannot be the case. However, it is the case, as certainly as the square root of two is irrational. Mathematics has a puzzling lump of incompleteness which can be pushed from place to place but it will never disappear.

Surprisingly, Gödel himself was a little more optimistic. Here, Rachael Alvir explains that Gödel maintained the dream of a formal logical system that could settle the continuum hypothesis and all other questions about sets, the building blocks of modern mathematics. His incompleteness theorems tell us that any such system, so long as it consists of a finite list of axioms, will give rise to new statements that are undecidable within that system. But he wondered about the possibility of an infinite succession of ever-larger axiomatic systems that could settle every question.

RACHAEL ALVIR, logician and lecturer at the University of Waterloo

We have all been exposed to the general idea that Gödel killed Hilbert’s Program for thorough formalization of math. This is a common interpretation, so I was shocked when I first read Gödel’s original works. In his 1931 paper, in which the incompleteness theorems are first proven, Gödel explicitly states the opposite: “It must be expressly noted that Proposition XI (and the corresponding results for M and A) represent no contradiction of the formalistic standpoint of Hilbert.” In a footnote, he reiterates that the undecidable theorems of the 1931 paper are only undecidable relative to one system. The undecidable statements of any given logical framework can be mathematically proven to be true or false in a larger logical framework.

Gödel had no qualm with the claim that mathematics could prove or disprove every well-posed statement. Rather, Gödel took issue with Hilbert’s restrictive methods. Why should we believe there is a single, finite set of axioms, from which every truth will follow in a finite number of logical steps? Gödel believed that it was possible to redefine what we mean by a formal mathematical framework, or allow for alternative frameworks. He often discussed an infinite sequence of acceptable logical systems, each more powerful than the last. Every well-formulated mathematical question might be answerable within one of them.

Often people will speak as if the CH is the smoking gun that shows sometimes mathematical questions have no answer. But in my opinion, this situation provides very little evidence that there are “absolutely undecidable” mathematical problems, relative to any given permissible framework. It is simply one example of a statement which has not currently been decided, and on its own provides no reason to suspect it could not be decided in the future using new techniques. There are extensive, ongoing debates about this deep in the trenches of mathematics and philosophy.

The strongest point I wish to make is that the mathematical results, on their own, cannot settle the question. It is far from obvious that there are mathematical questions with no solution. For me, Gödel’s theorems do not show that mathematics is limited, but rather that mathematics is much wider and more powerful than Hilbert’s finitistic view.

Alvir further clarified that there are different ways the old dream of mathematical truth might be realized. One approach could be to tack on to the commonly accepted axioms a new one that settles the CH and doesn’t otherwise lead to any contradictions. Another approach is to discover a scheme for an infinitude of axioms that settles the CH and other questions. Or we could switch to a different logical system than the standard one, and in that alt-logic, settle the CH. (“My personal favorite [logical system] is called L-omega-1-omega,” Alvir told me, for anyone who wants to explore that further.) Or maybe the answer is “something totally new,” she said — “a truly novel stroke of creative genius. … We come up with radically new mathematical techniques to solve problems all the time. Why expect we won’t do the same for the CH?”

Of course, proving the CH true or false wouldn’t vanquish all undecidability.

I’m going to let Väänänen’s colleague (and wife) have the last word.

JULIETTE KENNEDY, philosopher of mathematics and mathematical logician at the University of Helsinki, editor of Interpreting Gödel: Critical Essays

It is easy to lose one’s sense of wonder at the fact that such a blindingly obvious set of axioms — the Peano axioms for arithmetic (the set of rules about the natural numbers 0, 1, 2, 3 … closely related to the system that Gödel used in his proof, such as the rule, “Every number has a successor”) — is essentially incomplete and undecidable, meaning that all axiomatizable consistent extensions are incomplete and undecidable. Hold on to that wonder! The incompleteness theorems teach us that when it comes to our attempt to master the conceptual order, whether it be in mathematics or, for that matter, in any other domain, we will always fail — and indeed, in this case more than any other, we should be glad to have failed, for failure was clearly the more interesting, the more profound, outcome.

Rubin Tracks Skyscraper-Size Asteroids, Failed Supernovas, and Interstellar Visitors

Jonathan O'Callaghan — Fri, 15 May 2026 01:50:50 GMT

The Vera C. Rubin Observatory sits at the summit of Cerro Pachón, accessible by a 35-kilometer drive on winding mountain roads.

NSF–DOE Rubin Observatory/NOIRLab/SLAC/AURA/A. Pizarro D.

Over the years, anticipation has built for the start of observations at the Vera C. Rubin Observatory in the mountains of the Atacama Desert in Chile. Originally imagined in the mid-1990s as the Dark Matter Telescope, Rubin is designed to study our constantly moving and changing universe in greater detail than ever before. Once every few days for a decade, Rubin will take images of the entire night sky over the Southern Hemisphere, creating the world’s largest time-lapse movie.

In Rubin’s first year alone, scientists expect the observatory to find 1 million undiscovered asteroids — as many as have been documented in the previous 200 years of human history — as well as thousands of comets and billions of stars and galaxies.

“We’ve never had this kind of explosion of discovery within astronomy,” said Sarah Greenstreet, an astronomer at the National Optical-Infrared Astronomy Research Laboratory.

A little over a decade after the first stone was laid to build Rubin’s home on the mountaintop of Cerro Pachón, the observatory is now a reality, outfitted with a telescope with three mirrors, the largest of which measures 8.4 meters across, and a car-size digital camera, the largest on Earth. It has begun collecting preliminary images.

“It almost doesn’t feel real that we’re actually getting data from Rubin,” said Matt Nicholl, an astrophysicist at Queen’s University Belfast in Northern Ireland. “To see stuff being found is a dream come true.”

Astronomers are poring over the initial data, and they’re pleased with what they’re finding: rapidly spinning asteroids; myriad exploding stars; and even a rare glimpse of an object passing by from another solar system. “It’s really living up to expectations,” said Michael Frazer, an astronomer at Curtin University in Australia.

Spinning Asteroids

As the observatory goes through its final tuning, Rubin’s images have not yet reached the sharpness that scientists expect. But some Rubin science is less dependent on image quality, including its searches for asteroids and comets. This means that, even in the images taken so far, astronomers have been able to make discoveries.

In June 2025, Rubin released a set of images taken during its “first light,” including photographs of 1,500 new asteroids. In January, researchers announced that 19 of those asteroids were spinning especially rapidly. The quickest of these “superfast rotators,” an asteroid with a diameter almost twice the height of the Empire State Building, called 2025 MN₄₅, completes a revolution every 1.88 minutes.

While scientists have spotted asteroids that spin faster, they’ve tended to be much smaller — between 10 and a few hundred meters. For asteroids the size of 2025 MN₄₅, about 700 meters (2,300 feet) across on average, “we didn’t expect we would find something [spinning] faster than 10 minutes,” said Dmitrii Vavilov of the University of Washington, a co-author on the discovery paper.

Most asteroids of this size are thought to be piles of rubble, conglomerations of rock loosely held together by gravity. But 2025 MN₄₅ must have a more solid structure; otherwise its own spin would tear it apart. It might be the fragmented chunk of a long-dead planetary core from the early solar system, broken in a collision and left to spin wildly through space for the last 4.5 billion years, said Greenstreet, the paper’s lead author.

Rubin’s vast pool of asteroids could help scientists piece together the history of our solar system. Astronomers think that the planets were much closer together when they first formed but that they have migrated over time to their current orbits. Finding asteroids in certain patterns of motion, such as an orbit in sync with Neptune’s, could help us trace this migration, Greenstreet said.

Scientists also hope that Rubin will supercharge efforts to spot small asteroids, those that are just a few meters in size, before they hit Earth. These asteroids, known as imminent impactors, mostly burn up in our atmosphere, producing brilliant fireballs in the sky.

Recent simulations show that Rubin might find about one of these a year. What’s more, “it should see them a couple of days in advance, instead of a couple of hours,” as current telescopes do, said Frazer, who led the simulation work. That could give astronomers enough time to travel to the location of the impact and watch it unfold, or to look for any meteorites that make it to the ground. “We can send people out and put a whole bunch of sensors down, from cameras to infrasound,” he said.

It will also be possible to alert members of the public to the event so they can watch the flash in the sky. “We can tell people to go look outside, because we know there’s going to be a beautiful fireball,” Frazer said.

Swarms of Supernovas

Rubin will spend its first year creating a baseline map of the night sky, and scientists will compare subsequent images to this template. An automated alert system will ping them when it comes across a change, such as an exploding star or a flying asteroid.

Rubin tested out its alert system for the first time on February 24, 2026. By photographing a patch of sky for which previous surveys had already built up a sufficient template, Rubin was able to ping 800,000 alerts in a single night.

“Everything that changed, appeared, or disappeared was cataloged and triggered an alert,” said Stephen Smartt of the University of Oxford. Smartt serves as scientific lead for Lasair, one of the seven data brokers that will help astronomers sift through Rubin’s vast data haul for discoveries.

The camera, seen here mounted to the telescope, weighs about 3,000 kilograms. You would need hundreds of ultra-high-definition TV screens to display just one of its images.

RubinObs/NSF/DOE/NOIRLab/SLAC/AURA/T. Lange

Once the full survey begins this summer, Rubin is expected to produce 7 million alerts and 20 terabytes of data a night.

For just one example of how this firehose of data is expected to transform our understanding of the cosmos, consider supernovas, the brilliant death throes of exhausted stars.

Back in the late 1990s, two teams of astronomers used observations of under 100 “Type Ia” supernovas to make a revolutionary discovery about our universe: Its expansion is accelerating due to a still-mysterious force called dark energy. Once Rubin is fully up and running, researchers expect to find 250,000 such supernovas in a year.

Scientists hope that Rubin’s supernova data can help resolve the Hubble tension, the observation that the recent universe appears to be expanding faster than predicted when compared to the early universe. “We want to collect huge samples of Type Ia supernovae to probe this acceleration in much greater detail,” Smartt said.

Smartt is also interested in finding failed supernovas, which occur when stars collapse in on themselves rather than exploding outward. They might have their origins, paradoxically, in the most massive stars. In February 2026, scientists pinpointed a possible candidate in the Andromeda galaxy.

Rubin, with the exquisite detail of its images, is well placed to find these types of events, in which stars disappear in explosions that can be too faint for other surveys to see. “It goes down 100 times fainter than other sky surveys,” Smartt said.

Visitors From Afar

Rubin can also be used to track interesting and unusual objects passing through our solar system. It’s traditionally been difficult to catch such speedy travelers, at least without a survey that can pick out very faint objects at a rapid pace. Scientists have only ever observed three such interstellar objects — asteroids and comets that were ejected from other stars and fired into our vicinity — giving us insight into material from other solar systems. Rubin has already proved its ability to spot them.

Scientists announced the observation of an interstellar comet called 3I/ATLAS on July 1, 2025. They detected it without Rubin, via a network of four other telescopes that forms the Asteroid Terrestrial-Impact Last Alert System (ATLAS), which usually finds objects formed nearby.

Other astronomers followed up by looking through Rubin’s initial data and discovered that the observatory had also detected 3I/ATLAS, 10 days earlier. If a similar visitor from afar appears in Rubin’s data during the survey, astronomers will receive an alert.

Scientists don’t know exactly how many more interstellar objects Rubin will find, but they expect it to find at least some. “It could be five to 500,” depending on how often these objects are ejected from their home systems, said Rosemary Dorsey, an astrophysicist at the University of Helsinki in Finland. “I am optimistic there will be some, but if there aren’t, then that is a really interesting problem.”

Going the Distance

One way that astronomers determine the distance to an object in space is by studying its light. As light makes its way toward Earth while traveling through the expanding universe, it shifts toward the red side of the electromagnetic spectrum. The higher the redshift, the more stretched the light is, and the farther its source is from Earth.

Rubin’s preview data allowed scientists to test how well it could measure this light via a technique called photometric redshift, which will let it map galaxies across the universe to probe dark energy and dark matter. “The preview data tells us how accurate those photometric redshifts are going to be,” said Kristen Dage, an astronomer at Curtin University. Rubin performed at least as well as other cutting-edge telescopes, Dage said, but it will measure the redshift of many more galaxies, about 4 billion of the 20 billion galaxies it will find.

Dage expects that this data will also help scientists study fast radio bursts (FRBs), bright, unexplained flashes of radio waves in the sky possibly linked to highly magnetized stars called magnetars. While Rubin cannot detect radio waves, photometric redshift data will help scientists work out the distances to FRBs if they can be sourced to a host galaxy Rubin can measure, which could help scientists determine the processes that trigger them.

All of this is just a smattering of what scientists are hoping to explore when Rubin comes online. With it, a new era of astronomy is set to begin.

“Rubin is going to be putting out so much data, so many alerts every night, that everybody’s going to struggle to keep up with [the] information,” Frazer said — a challenge, but a delightful problem to have.

How the Bird Eye Was Pushed to an Evolutionary Extreme

Yasemin Saplakoglu — Wed, 13 May 2026 02:04:30 GMT

The eye of a red-and-green macaw, with no blood vessels in sight. How can a bird eye work so well without oxygen?

Leonardo Ramos

When an optometrist shines a bright light into your eyes, a vast, branching tree sprouts in your field of vision. This is the shadow of blood vessels. Though we normally can’t perceive them, these vessels always occlude a portion of what we see, and for an important reason. They power the retina, a thin layer of nerve tissue in the back of the eye that communicates light signals to the brain.

The retina is one of the body’s most energetically expensive tissues. Built from complex networks of sometimes more than 100 different types of neurons, retinal tissue consumes two to three times more energy than the same mass of typical brain tissue. That’s why most vertebrate retinas, including our own, are furrowed with dense, branching networks of blood vessels: to deliver oxygen and other ingredients for producing energy.

But there’s a significant exception to this rule. Birds have retinas that mostly lack blood vessels. This may seem especially strange given birds’ exceptional vision. The bird retina is “one of the most metabolically active tissues in the animal kingdom, yet it worked with no apparent blood perfusion,” said Christian Damsgaard, an evolutionary physiologist at Aarhus University. “It was a complete paradox.” For centuries this has puzzled scientists, who figured that the bird retina must obtain oxygen through a unique, undiscovered process.

Damsgaard is the lead author of a study, published in the journal Nature in January 2026, that showed for the first time that bird retinas don’t have some unusual adaptation for acquiring oxygen — they survive without it entirely. Instead, to bring energy to the tissue, they use a process called anaerobic glycolysis that is significantly less efficient than oxygen-powered metabolism but gets the job done.

The evolutionary physiologist Christian Damsgaard measured gas exchange in bird eyes with microsensors. Surprisingly, the inner retina, a highly active tissue, used no oxygen.

Jesper Ekmann

By studying how tissues can survive without oxygen, researchers can potentially develop therapeutics to treat conditions of oxygen deprivation, such as strokes. More fundamentally, they want to understand the limits of evolution.

“What are the extremes of life?” Damsgaard said. “How far can we bend the conditions under which highly metabolically active tissues can actually survive?”

A bird, he learned, can bend them pretty far.

Oxygenated Life

Around 3.4 billion years ago, cyanobacteria invented photosynthesis. Slowly at first, then quickly, their newly evolved method of making energy from sunlight succeeded and spread. The cells pumped so much oxygen, a by-product of photosynthesis, into the atmosphere that it changed the course of life on Earth.

Oxygen molecules make energy production in cells extremely efficient. To extract energy, cells break down a glucose molecule into two pyruvate molecules. This process releases two molecules of ATP (adenosine triphosphate), life’s universal energy currency. A cell lacking oxygen can go only this far. Oxygen, however, enables further biochemical reactions that break down pyruvate and produce another 30 molecules of ATP. In other words, the presence of oxygen makes energy extraction from a single glucose molecule 15 times as efficient, and sometimes more.

Birds, such as this alpine chough (in the crow family), use their exceptional vision to hunt, forage, and migrate. This energetic ability is powered by an inefficient metabolism.

Jean-Paul Wettstein

The energetic advantage of oxygen, through the process of aerobic respiration, was transformative. Once oxygen imbued the atmosphere, evolution selected for organisms that could use it. “We’ve been hooked on 20% [atmospheric] oxygen for millions of years,” said Gary Lewin, a molecular physiologist at the Max Delbrück Center in Berlin. This Great Oxidation Event was followed by mass extinction, as organisms using oxygen outcompeted just about everybody else. While some life forms, such as certain bacteria, are adapted to life without oxygen, all complex, multicellular organisms need that energy advantage to survive.

Humans and most other animals can survive with little or no oxygen for several minutes at most. The mammal with the highest known tolerance for low-oxygen conditions is the naked mole rat, which can survive for up to 18 minutes breathing anoxic air in underground burrows. A few cold-blooded aquatic creatures, including freshwater turtles and goldfish, can persist in low-oxygen conditions at the bottom of a frozen lake for a year or two. But for most animals, a steady supply of oxygen is a must-have.

Without oxygen, a variety of processes shut down — especially in metabolically demanding tissues such as the brain. Without that energy, our cells malfunction and die.

Naked mole rats can survive without oxygen for 18 minutes. To generate energy without oxygen, they use anaerobic glycolysis fueled by fructose.

Javier Ábalos

A Mysterious Structure

All this is why, in 2019, when Damsgaard learned that bird retinas lack blood vessels, he was confused. How could this high-energy tissue survive, let alone perform at the level observed in sharp-sighted bird species, without oxygen?

He pored over the voluminous research on the subject, all of which pointed at a mysterious structure in the bird eye known as the pecten oculi. In the 17th century, anatomists first described the unusual organ: It looked like a radiator, comblike, riveted with blood vessels, and with a large surface area. In the centuries that followed, researchers debated whether it helps deliver oxygen to retinal tissue in bird eyes. Damsgaard read about 30 different theories about the pecten oculi’s function based on anatomy alone.

“Nobody had really done direct physiological measurements on this structure,” he said. “That’s where we came in.”

Mark Belan/Quanta Magazine

In his lab, which studies the exchange of gases such as oxygen and carbon dioxide between vertebrates and their environments, Damsgaard’s team used microsensors to measure oxygen levels in the retinas of zebra finches, pigeons, and chickens. Indeed, in the inner retina, which completely lacks blood vessels, they found no oxygen. (They did measure oxygen in the outer retina, at the back of the eye, which has some blood vessels.)

That was “striking,” Damsgaard said. “Half of the retina lives in a chronic state of anoxia, where there’s no oxygen present at all.”

Using spatial transcriptomics, a method that combines cell imaging with RNA sequencing, the researchers mapped which genes were active in different parts of the retinal tissue. Genes associated with typical aerobic respiration were expressed in the outer retina, where there are blood vessels. In the oxygen-depleted inner retina, only genes associated with anaerobic respiration were active.

To trace the paths of nutrients, Damsgaard and his team worked with cancer scientists who are experts on oxygen-free metabolism (tumor cells often use anaerobic glycolysis to make energy). They found that the inner retina demanded 2.5 times more glucose than other parts of the bird brain.

Then they examined the pecten oculi. Their spatial transcriptomics data showed that the genes for glucose were highly active there. This suggested that the strange structure wasn’t bringing oxygen into the bird’s retina; rather, it was helping to pump glucose in, thereby enabling the less efficient anaerobic process.

As a by-product, anaerobic glycolysis creates lactic acid, which can accumulate and become toxic. The researchers also saw that genes for lactic acid transporters — the molecules that move lactic acid out of tissues — were active in the pecten oculi.

The diversity of bird eyes, lacking blood vessels (left to right). Top: Northern gannet, Eurasian eagle-owl, maguari stork. Center: rooster, rockhopper penguin, parrot (species unknown). Bottom: bald eagle, blue-and-yellow macaw, unknown species.

(Left to right) Top: Chris Hellier, Jiří Dočkal, Annette Lozinski. Center: Mohammed Brzan, Nico Marín, Shyamli Kashyap. Bottom: Ingo Doerrie, David Clode, Hasan Almasi

Their findings provide compelling evidence for the role of the pecten oculi in supporting anaerobic glycolysis, which “has been a mystery for a long time,” said Thomas Baden, a neuroscientist at the University of Sussex who was not involved in the study. “The insight that the retina basically goes oxygen-free, at least in some layers, is surprising. … It really gets properly down to zero.”

This pathway is used by cancer cells and temporarily by our muscles when they’re strained and can’t get enough oxygen — such as when we’re running. But no known vertebrate tissue was known to survive in completely anoxic conditions for a lifetime.

Eyes Like a Hawk

The bird’s retina and its no-oxygen power system are so unusual that they naturally raise questions about how they could have evolved.

This is “a series of beautiful experiments,” said Karthik Shekhar of the University of California, Berkeley, who was not involved in the research. It’s an example of how an animal took the vertebrate eye — a highly conserved structure whose origins go back some 560 million years to a light-sensitive patch on a primitive creature — and tinkered with it to fit its own needs. “Evolution is not really like an inventor; it acts more like a tinkerer,” he said, citing a 1977 essay, “Evolution and Tinkering,” by the French biologist François Jacob. “It takes parts that have existed long before, and it recombines, reinvents, and reshapes.”

The researchers tried to pinpoint when the pecten oculi might have arisen by comparing oxygen levels in the bird retina to those in not-so-distant relatives: two reptile species, Chinese pond turtles and broad-snouted caimans. The reptile retinas had normal oxygen levels and no indication of anaerobic glycolysis. This led Damsgaard’s team to conclude that the oxygen-free tissue likely evolved sometime during the dinosaur era, after the avian lineage had split from crocodiles but hadn’t yet evolved into modern birds. This was around the same time that the retina thickened.

Still, that rough time estimate can’t explain what evolutionary pressure might have selected for the unusual retinal tissue. Researchers can only speculate. “I think the system evolved in theropod dinosaurs in response to selection for sharp vision for tracking prey and identifying mates,” Damsgaard suggested. Then, later, when birds took to the skies, it “served as the physiological basis for maintaining retinal function” during high-altitude flights, when oxygen levels are low, he speculated.

The lack of blood vessels could also offer birds the advantage of better vision. The bird retina is complex and densely packed with more than a hundred cell types that work to render the world in great resolution. Birds use their exceptional visual sense for hunting and foraging — consider an owl tracking a mouse from the sky, an albatross watching for signs of fish on the ocean’s surface, or a hummingbird locating hundreds of flowers every day — as well as for following landmarks across the landscape during migration. Without blood vessels obstructing their view, birds’ retinal cells might be able to take in more visual information.

Could this be an adaptation, or is it a coincidence of evolutionary history? There’s no way to know for sure how birds’ incredible vision evolved. There’s this mystery “that has lingered around us,” Baden said. “What is it about birds that makes their eyes so special?” Their retinal power system seems as if it could explain what makes them so unique. However, Lewin, the physiologist, is cautious about overextending the results and interpretations to every bird, given that the researchers haven’t looked at any migratory species.

The implications stretch well beyond bird adaptations to biomedicine. A common thread in many medical conditions is a drop in oxygen delivery to tissues, which, depending on where it occurs, can lead to scars or brain damage. Human brains can tolerate maybe a minute of total anoxia, Lewin said. That’s what makes strokes, which cut off blood and oxygen supply to parts of the brain, so devastating. By studying low-oxygen conditions in creatures such as naked mole rats and birds, scientists can gain insight into how tissues can tolerate low-oxygen conditions.

“Maybe we can get inspiration for how nature solved these problems by millions of years of natural selection,” Damsgaard said. “There’s so much to be learned from these animals that are able to do something that we cannot do.”

Quanta Magazine

Why the Human Genome’s Tangled Physicality May Confound AI

Introduction

The Old View

Transcription Factors

Enhancers

Loops and Hubs

Chromatin Shape

RNA Interventions

Checks and Balances

“A Highly Sensitive Organ”

Seven Perfect Shuffles Randomize a Deck of Cards. But How Many Sloppy Ones?

Introduction

Mixing Cold Spots

Tracking Labels

To Clumping and Beyond

How Many Elementary Particles Are There, Really?

Introduction

Where Did Earth Get Its Oceans? Maybe It Made Them Itself.

Introduction

A Showdown Between Comets and Asteroids

Hydrogen, Meet Magma

Drowning in a Sea of Possibilities

What’s the Future of Gene Editing?

Introduction

Transcript

An Early Step on the Long, Strange Road to Photosynthesis

Introduction

A Photon Capture Machine

A Strange Antenna

Evolutionary Oddity

How Terry Tao Became an Evangelist for AI in Math

Introduction

Are Memories Transferable — or Edible?

Introduction

More Conversations, Complex Questions, and Bold Ideas in Season Five of ‘The Joy of Why’

Introduction

Entanglement Builds Space-Time. Now “Magic” Gives It Gravity.

How To Code a Universe

The Magic Ingredient

Gravity From Quantumness

The Dirt That Refused To Die

Introduction

The Living Dead

Dirty Electrons

An Origin of Life?

Key Chemistry Question Answered, No Quantum Computer Required

Introduction

Nature’s Ammonia Factory

An Unlikely Test

Ground-State Debates

Sifting Out the Solution

The Debate Shifts

How We See the Beautiful, Violent Sun

When Quiet Undersea Volcanoes Turn Disruptive

Introduction

A Conveyor Belt of Lava

A Fleeting Apparition

A Land of Ice

Eruptions Past and Future

How Ecotypes Harbor the Genetic Memory of a Species’ Past

Introduction

Genetic Memory

Shuffling Chromosomes

On the Origin of Ecotypes

Two Researchers Are Rebuilding Mathematics From the Ground Up

Point of Know Return

Just a Drop of Water in an Endless Sea

The Nicest Revolutionaries

Building on a Foundation of Dust

A Condensed History

How Alexander Grothendieck Revolutionized 20th-Century Mathematics

Introduction

Revealing Shapes

A New Scheme of Mathematics

What Do Gödel’s Incompleteness Theorems Truly Mean?

Rubin Tracks Skyscraper-Size Asteroids, Failed Supernovas, and Interstellar Visitors

Introduction

Spinning Asteroids

Swarms of Supernovas