Sara Walker says “what is life?” is the wrong question

Sara Walker, an astrobiologist and theoretical physicist at Arizona State University and the Santa Fe Institute, tells Lex Fridman that the historical question—how did life start on Earth—is a decent place to begin, even if it is not the question she most wants to answer. The leading popular account is the RNA world: in the 1980s, researchers noticed that RNA can both carry genetic information and catalyze reactions, the two jobs usually split between DNA and proteins. That insight, she says, later collapsed into a cruder story: a molecule popped up on the early Earth, underwent Darwinian evolution, and life began. Packed into that story are assumptions she wants to unpack.

The rival tradition is metabolism-first. It sits closer to early-Earth geochemistry and asks how catalytic cycles of molecules could reproduce themselves, self-organize, and only later acquire evolution. Fridman summarizes the split as energy versus genetic code, or energy versus information, and Walker agrees. She also thinks the split is mostly disciplinary. People who start from genetics tend to come from biology, worry about where sequence information comes from, and mean “information” in a narrow sense. People who start from metabolism tend to be physicists or geochemists, asking what energy sources can buy what kinds of organization.

Asked which camp she prefers, she says she likes neither. The most popular setting for a metabolic origin is hydrothermal vents: hot, chemically active regions on the seafloor that would have supplied energy, organics, and roughly the right conditions, and that still exist today. We do not actually know much about early Earth, she cautions—when oceans formed, when origin happened, what the planet was like then. Even so, hydrothermal systems are also what researchers look for when they hunt life elsewhere. Finding life, in that framing, is finding a kind of condition, not a kind of object.

On the genetic-code side, the pitch for RNA as “first life” is that one molecule can both persist information and do work. Treat the sequence like a string in a computer: it can be copied, so information propagates. Walker’s example is ordinary inheritance—her eyes are brown because her mother’s were. Copying is what evolution needs. Catalysis, meanwhile, means the molecule is not inert: it can interact with its surroundings, perhaps mediate a metabolism that then fuels its own reproduction. One molecule, in this picture, is both storage and execution, the jobs computers split between disk and processor.

She is not talking RNA down. It is one of her favorite molecules; she thinks it is beautiful. It is just not step one. The RNA world, she insists, is not a single hypothesis but a set. The hardest version—an RNA molecule spontaneously emerging on early Earth and starting to evolve—is the one she cited, and she gets “a little animated” about it because it seems “blatantly wrong.” The softer version she accepts: RNA was the first genetic material in the lineage that led to us, before DNA evolved. If life were a greatest-hits album, RNA would be on it, and early. That does not make it the first thing that happened.

Did origin happen once, twice, or many times on Earth? Walker calls that an important question wrapped in other questions. The usual picture is a singular event nearly four billion years ago, after which all known life is a descendant. Nothing rules out origin as a continuous process that simply cannot get going because life already occupies the niches. The more unsettling alternative is a shadow biosphere: other life on Earth so alien we do not recognize it. Fridman parks that for later and turns to panspermia—life arriving on an asteroid, planetesimal, or, as Wikipedia has it, “space dust.”

She thinks panspermia is entirely possible. Microbes can survive space travel in rock; experiments suggest the Mars–Earth hop can work. Once a technological civilization exists, spreading through the cosmos is even less mysterious. What it does not do is help her problem. Move origin off Earth and you know even less. If Mars life turned out biochemically identical to ours, we would not have learned what other original-life events are possible. Fridman offers a wilder image: a Simpsons-style makeup gun that sticks wherever it hits, a “life gun” that lowers the time and luck needed to start from basic chemistry. Walker takes the implication seriously. Whether seeded life has to be geochemically compatible with its planet is an open question. We could shoot life at Mars and watch it live there happily—or not.

Her own thinking, which she says changes weekly, is that life is more a planetary phenomenon. Origin grew out of geochemistry, then coupled to and controlled geochemistry; we say a planet has life when life starts influencing the planet’s properties. If that is right, going to Mars is not a trivial transplant. “Let’s send our civilization to Mars” really means “let’s reproduce our planet on Mars.” Information from Earth has to arrive and make Mars more Earth-like, the way a cell reproduces to push information into the future. Planets that want to reproduce life in the universe, she argues, have to reproduce their conditions, including geochemical ones. Radical, she admits—and, for long-term sustainability, essential.

If life is planetary, Fridman says, we still have to ask what life is. Walker crosses it off the list. Science gets stuck, she argues, not for lack of answers but for asking the wrong questions. She is being a little cheeky: what life is sits at the core of her existence. The better frame is what it is about our universe that allows the features we associate with life. Definitions are not scarce. A popular one is a self-reproducing chemical system capable of Darwinian evolution; another is something that metabolizes and eats. That is not how she thinks. She thinks there are principles, not yet understood, about how information interacts with the physical world.

She drops a water bottle instead of a cup, “which is why I’m a theorist, not an experimentalist.” Why did it fall? There was a long stretch of human thought when nobody knew what gravity was, or that the heavens and the ground obeyed the same laws. Newton unified terrestrial and celestial motion and the explanations got more powerful. Life, she says, sits in that same historical slot. Gravity is universal, yet we study it in galaxies, black holes, and planets. If we want to understand information in the physical world, we should study living and intelligent systems, because they are that physics made visible. We cannot see the structure yet because we have not thought deeply enough.

Wearing her physicist hat, she prefers simple, elegant explanations. She has also, in some ways, rejected her training: the elegant explanations we have so far do not include us. She cannot help thinking there is something special about what we are. Asked whether the words “information” and “computation” would appear in a finished answer, she says no. She hates that about her work. The words are historically loaded; they describe something people have already seen, and she is trying to describe something they have not. Swap in binary and it is still wrong. “Physics of causation” is worse, because causation is even more loaded. The poverty of vocabulary is a signal about the problem, not just a communications failure. At the deepest level, she says, life is the physics of existence: what gets to exist, and why. For elementary particles the interactions are simple. For people and civilizations, what comes next depends on what came before. Why did cups evolve, and not some other object that delivers water without spilling?

Fridman asks whether it helps to throw away biology, chemistry, even textbook physics, and treat the universe as computation—cellular automata and other simple local rules that grow complexity. Walker likes the toys. She uses them mainly to poke holes in the current conceptual framework. As a proof of principle that complexity can emerge from simple rules, they are great, and beautiful. As a model of dynamical systems, even of the universe’s evolution, they are useful. As an explanatory framework for life they have the same problem the laws of physics have, and she can say it clearly.

Cellular automata sit in a Newtonian picture: a fixed law of motion, handed down by the great programmer in the sky, which you then just run. A good law does not contain information about the patterns it generates. You would not want her cup, her water bottle, or Walker sitting in the chair written into the laws of physics; that would not be a simple explanation of anything. Automata and physics both have that virtue. The cost is that you must specify an initial condition, and everything that happens is a consequence of it. She does not think that is the right framework for biology. Genomes encode information that is read out to specify cell architecture, and the architecture includes the genome: self-reference, the same territory as Turing and Gödel. In more physics-y language, in biology the rules depend on the state. That is ordinary in computer science—the update rule depends on the machine’s current state—and a very special subclass of computation in physics, where the rule never changes.

In biology, state and law seem to change together in time. Physics has no such paradigm. Even laws that vary in time without depending on state are radical; the usual hope is to remove time from the equations. Walker and collaborators go the other way. Time may be fundamental. Objects in the universe are unique because they require a particular history to exist, which means events have a real order—not the felt flow of time, but causation. If she had to bet on one horse: causation. Space is emergent. Everything is emergent except time, or causation. Why, then, do the laws look so simple and universal? The idea, which she credits to collaborator Lee Cronin, is that physical laws are low-memory laws. They do not take much information to specify; the universe implements them cheaply. Walker herself needs a four-billion-year history. She comes with historical baggage, a set of causes that are part of what she is, so some local rules apply only to objects with her kind of history. She does not believe in Boltzmann brains, or vacuum fluctuations that cough up the ornament on your desk. Those things require a particular causal chain.

If a desk ornament needs a whole causal history, how is anyone supposed to engineer life in a lab? Walker’s answer returns to the RNA-world critique, sharper now. When those scenarios are tested, a core problem is how much information the experimenters put in: specified flasks, pure reagents, mixing, transferring, changing pH and UV, until a molecule appears that may not even be RNA, just a base. People rarely notice that they are agents in the universe making that experiment, and therefore putting a little life into it. The constraints would not have existed on early Earth without us as a boundary condition. Even with no identifiable biology added, the design is the input. Think of the protocol as a program. The origin problem becomes: minimize the information we put into physics so origin can happen spontaneously.

Recreating Earth’s exact historical sequence is the wrong bet. In computer science, if you already know the output it is easier to write a program that produces it; verifying a prime is easy, finding the next one is hard. Historical origin-of-life work knows the answer and retrodicts a procedure while pouring information in. The question to ask is how the universe’s structure permits life, which is the same as asking what the phenomenon is. You need experiments that let chemistry’s vast complexity run with as little constraint as possible, and then watch when information emerges, how, and what it is. Gravity exists everywhere, but you ignore it inside an atomic nucleus. Walker’s conjecture is that the physics of information works the same way: it becomes relevant when you need information to specify why this object exists. That scale is chemistry. You cannot make every protein of length 200 amino acids; the universe lacks the resources. For this protein to exist in abundance, a system must know about it and be able to build it. Existence becomes a problem at the chemical level.

What would a Nobel for this actually look like? Not a bacterium crawling out of a test tube, and not a historical reconstruction of early Earth, any more than the Higgs search was a hunt for the historical origin of mass. You need theory that says which features to look for, then an experiment outside the current paradigm. One prize experiment: prove that information originated inside the bounds of the system, was not injected by you, and began to govern and specify the subsequent dynamics. Another, which Fridman poses as injecting sugar and getting information out: every physical object is already information, a set of causal histories and possible futures. With Cronin, and with Michael Lachmann and Chris Kempes at Santa Fe, she has discussed “seeding assembly”—place a high-complexity object into a lower-history system and watch complexity appear. Walker describes herself in the same language. She is not her atoms. She is an informational pattern that keeps re-patterning those atoms into Sara, plus a space of possible things she can help bring into existence because of the information in her history.

Fridman turns to the entity she calls Sara, and to consciousness. The problems are absolutely related, she says, because she is interested in what we are, and the most interesting feature of that is minds interacting with other minds. “Minds are the most beautiful thing that exists in the universe.” “We,” for now, means humans, because she is one, or thinks she is; she is not a human-centric thinker. She is fine being described as a bundle of information temporarily aggregated into an individual—she takes it as a compliment—and that bundle claims to be conscious. She thinks she is conscious in the conversation. She might not be. He would not know.

That is the hard problem: subjective experience looks impenetrable from outside. She likes pairing it with the hard problem of matter. You do not know an electron’s intrinsic properties without interacting with it; all properties of anything in the universe are defined by interaction, because observation is interaction. Consciousness is hard for science because it asks how it feels to exist from inside. The question she has found that might be tractable is whether there are things that can happen in the universe only because there are physical systems with subjective experience. Does experience have consequences that would not occur in its absence? She cannot ask whether Fridman is experiencing the conversation. She can ask whether that experience changes how he interacts with the world. If it does, consciousness is a real physical thing. She is a physicist; she has tried to drop that bias and failed. Information is physical too. Virtual reality, simulations, programs—all physical, just not in the way they are represented in our minds.

A tweet of hers compared the search for neural correlates of consciousness with chemical correlates of life, and noted that neuroscience and astrobiology treat those correlates differently. Analyzing her own tweet on the podcast, she jokes, is itself an assembly experiment: bringing a past thought into the present. Neuroscience largely accepts that the subjective side is out of reach, so it studies correlates—verbal report is the obvious one: she can tell you when she is conscious, and we assume sleep is not. Theories of consciousness try to build experimental bridges to those proxies while admitting a correlate may not be the thing. Astrobiology has no concept of chemical correlates. Find molecules involved in biology, and you have found life. Her point was that life has abstract properties imprinted on material substrates. The substrates are correlates. They are not what we are looking for. We are looking for the physics that organizes the system, not the particular molecules—just as consciousness is instantiated in a brain without being the brain. “Chemical correlates of life” is a phrase she made up; she may write it into a paper someday.

Can consciousness be engineered? “Yes,” she says, so fast she notices. Fridman jokes that she has no free will; she denies it. Fast answers, she thinks, often mean she has not really thought, which is itself information: consciousness and subconscious can conflict, and intuition leads a lot of her science. Later they return to interaction. Consciousness, in the picture she likes, is created between things, not only between two humans. Philosophers get angry when the illusion of consciousness is treated as consciousness. She likes engineering systems that fool you into thinking they are conscious, because the magic is the experience in the interaction. Fridman talks about an ocean of loneliness, love as a brief relief, and building a submarine that can truly listen. Walker has a colleague who suggests consciousness might be contagious. She wants to know whether his motive is to make more of the universe appreciate those interactions, or to enjoy the human pleasure in them. He says the joy is in the creation of consciousness itself.

A lowercase tweet of hers linked determinism and randomness to understanding life. She is not sure whether the universe is random or deterministic; an old intuition still lingers. If a cup, a desk ornament, or a phone could spontaneously fluctuate into existence on Mars, the universe would not need life to create those objects. They do not look like flukes. They need causes that accrue, and a history. In that sense you can think of life as the most deterministic part of physics—where the causes are precise, or at least the most stable and reliable for physics’ tools. Fridman presses the apparent contradiction: causation looks fundamental, which pulls toward determinism, and she believes she has free will. She rejects the premise. She does not accept the picture in which an initial condition plus a fixed law of motion describes everything. Drop that assertion and the incompatibility disappears.

Free will, physically: she as a physical system has causal control over some things, not all. She is a nexus of particular histories and of futures specified in part by her configuration as her. It may not be free will in the traditional sense; she is not sure people know what they mean by the phrase. Does Sara have a choice about what happens next? She thinks so. Where the choice comes from may be the physics of consciousness. She admits this might be hope talking, because she wants free will, but she also thinks nobody can rule it out yet. Imagination, she insists, is physical. It exists in the universe. Humans imagined flying machines and rockets for centuries, on paper, before building them; the existence of rockets is partly caused by the fact that we could imagine them. Some things have never physically existed, yet we can imagine them and then cause them to exist, individually or collectively. Biological intelligence is predicated on past history. Multicellular neural architectures crossed a line: they can imagine events not on their timeline and, if the events are consistent with physics, make them happen. Life, condensed: it changes the probability distribution over what can exist. Consciousness is a layered property that scrambles that distribution further. The description may be probabilistic; she does not necessarily think the underlying physics is.

Walker used to doubt that a theory of everything—unifying quantum field theory and general relativity—would say anything about life. She is increasingly convinced it will. Physics is full of deep holes that become especially clear when people try to patch quantum mechanics to general relativity, and some of those patches start to resemble the physics of life. At a workshop at Arizona State’s Beyond Center, Lee Smolin argued that the day quantum gravity is solved, that theory will be the same one that gives rise to life. She agrees in part. Causal-set approaches treat spacetime as emergent from a causal set, mathematically kin to Stephen Wolfram’s physics project and to assembly theory, the framework she and collaborators are developing: complex objects as assemblies of parts that encode possible histories. David Deutsch and Chiara Marletto’s constructor theory comes from another angle toward similar ideas.

A sneakier way to approach “what is life?” is to ask what we should look for in aliens. A paper with Cronin’s lab, she says, argues for high-assembly objects. Take a molecule, break it into elementary building blocks, enumerate the ways of putting them back together, and find the shortest path. That length is the assembly number. Higher means a longer causal history, or more information needed to specify the object. The intuition is old; as a physically measurable property of molecules it is new. Cronin’s lab has shown a threshold: non-biological samples do not produce high-assembly-number objects. A protein will not fluctuate onto the Martian surface. Generalize, and cups and desk ornaments live in an abstract assembly space you can probe with mass spectrometry. Fridman cannot stop thinking about Lego castles and shortest paths.

NASA and other agencies look for chemical correlates. Assembly does not care what the molecule is. It cares how much history and information it took. Observe something on another planet and you are reverse-engineering the shortest path to make it. Walker is blunt about her field. People see oxygen in Earth’s atmosphere or amino acids in a cell, then go looking for oxygen on exoplanets, amino acids on Mars, phosphine on Venus. A molecule is not life. Life is a system that patterns structure into matter; it does not care which molecules. Looking for a molecule tests no hypothesis about the nature of life. Oxygen on an exoplanet tells you nothing about what kind of life is there. “Are we alone?” is a deep question and deserves a deep answer, not a molecule and a wow. Life is also much less obvious than we think. Children can tell a puppy from a plant from a chair, so we assume we will know it when we see it. We also cannot see that the table is atoms, or that she is sitting in a gravitational potential well. Alien life might not even share a communication channel: not a language barrier, but an informational lineage that cannot copy into ours. Wildly speculative, she says, and consistent with the physics. Two probes follow: measure assembly structure across Earth and look for high-assembly things outside our biological lineage—aliens at home—or treat it as computer vision, asking whether high-assembly objects have stable statistics when projected onto a 2D plane. AI, for her, is fascinating less as a threat than as the first system that might help us see ourselves as four-billion-year lineages, which is what we are.

The internet, Fridman warns, would ask two questions of an alien: can I eat it, and can I have sex with it. Walker laughs and says those are serious existential questions; the internet has a mature grip on basal cognition. In her framework they cut to how divergent two histories are. We can eat one another and exchange genetic information because we share history, all the way to LUCA, the last universal common ancestor, the nexus of Earth’s causal map. Meaningful interaction with aliens requires finding the shared stretch of that map. Each lineage might have something like an Erdős number for common ancestry: closer means more chance of mixing. Technology-space is huge; chemistry-space looks lower-dimensional. Dealing with aliens may mean dropping back to those basal layers to find a common map—culture and language, harm, whether the parents approve, whether the common ancestor approves.

How many alien civilizations are out there? She has no intuition, and finds that interesting. Some questions you feel around and cannot see, even if they are sitting there, a bit like the quantum-to-classical transition: two different kinds of physics. Her mentor Paul Davies used to say that whether aliens are common or rare follows fashion and cultural mood. Observationally we have us, and one origin event. People argue that because it happened fast, and because Earth-like planets are many, origin should be common. The inference does not hold. Our observing the event is contingent on the event, a selection effect Brandon Carter articulated with anthropic arguments decades ago. The event could have been wildly improbable or very common; the data do not distinguish. Studying life is harder than other physics because we are studying ourselves. Her honest answer is that she does not know, because she cannot yet see how information is generated spontaneously, and inventing a one-in-ten chance for Earth-like exoplanets is not a mechanism.

Fridman notes that she always says “we,” not “humans,” putting herself inside the system. Physics, she says, has always treated us as external observers. Life demands the opposite: minds inside the universe, at a particular moment in history, looking out and trying to understand what they are. We do not have fundamental descriptions that put us inside. That was also her problem with cellular automata: you are always an external observer. What does a cellular automaton look like from the inside? She is an eternal optimist on purpose. If explanations have causal power—if theories enable technologies and physically transform the world—then optimistic theories are part of the physics of what comes next. Life is not only the physics of existence but of more things existing. The best way to maximize the number of things that can physically exist would be to evolve intelligent, technological systems that can explore that space. A shadow biosphere—independent origin events, aliens under our noses, as Davies liked to say, even in our noses—is worth looking for because it is cheap compared with spacecraft, and a second origin on one planet would imply origin is not rare. Her own work pulls the other way. Represent a cell’s chemistry as a network and the scaling matches ecosystems and the planet. Mix in alien chemistry and that scaling may break. Life is not a scale-dependent phenomenon. A “shadow” that must stay isolated is hard to keep. Geographic isolation is a weaker boundary than chemistry; life travels. Chemical distinctness is the stronger candidate, including wild possibilities like life in Io’s volcanoes or a second living layer deep in Earth’s crust.

UFO sightings intrigue Walker more as a cultural phenomenon than as streaks in the sky. They show how we prepare ourselves, historically and now, to understand others. A streak labeled “aliens” tells you nothing unless it comes with a deeper explanation of life. Fridman wants the sightings to challenge a Cold War instinct to hide mysteries as military secrets. She agrees that public engagement matters. A practical problem she had not expected: so many people already believe contact has happened that astrobiologists who actually want to understand life have to unteach those narratives and reteach from evidence. Aliens, she is pretty sure, will be weirder than we can imagine.

Viruses she finds beautiful even in a pandemic, with a wince: tiny objects that can handicap a civilization at global scale, a collision of perceived invincibility and susceptibility across scales. Fridman notes that anything which copies one-to-two-to-four—viruses, AI—can overtake competing forms on a short clock, which is why multiplanetary life gets offered as insurance. Walker is bothered by Darwinian slogans about the fittest replicator winning. Copying, for her, is ancillary to creativity. The drive is creativity; if you want past creativity to persist, it has to be copied forward. In arguments with Lachmann and Cronin about persistence versus existence, she takes existence. Survival of the fittest is easy in retrospect, when you post-select on what survived. You cannot do that going forward. A system that maximizes creativity might get rid of humans. She does not think technology replaces us that way. We have not lost past informational architectures on this planet; they are still here. Planetary-scale transitions matter for new phases. The global internet is one. AGI, in her view, will not be built in a box. It will emerge as a planetary-scale phenomenon from our technology, with us as an integrated component—the whole package, not a distinct successor. We are packets of information, some aggregated as individuals, exchanging and propagating.

Is the packet of information that keeps calling itself Sara afraid of dissipating? She has the ordinary biological fear of death. Rationally she is not afraid for herself, because she will not be aware of being dead. She is afraid for her kids, because her death matters to them. Death is more significant as a collective property than as an individual one. The information does not disappear, she says; it just is not her anymore, and that process is frightening. She is heartened that existence leaves imprints. The physicist’s comfort—that every moment exists and the flow of time is perception—does not actually comfort her about death. We like existing, especially if we love what life is. Fridman wonders whether fear of nonexistence is the fire that gives birth to existence, matter constantly freaking out that the ride ends. She thinks things want to exist, and want more things to exist. Some days she likes existing less; those are surface feelings. Underneath, the drive is deep. The small comfort: her existence matters to future existence.

Advice for young people, or future pockets of existence, is not a career trick for working on impossible problems. It could not have been otherwise. She started at community college. In a physics class she learned about magnetic monopoles: nobody knew if they existed, but the math predicted them and you could go look. That made her want to be a theoretical physicist. It was not actually her driving question. The driving question, she later realized, is the correspondence between our minds and physical reality—what we are. That question is deep enough to cross fields. Without it she could not have done what she has done. Students should find something they really care about. If they do not, they will not be their best, and it will not be worth their time. Passion changes creativity, talent, potential, and the ability to tolerate the hard parts of any life. Fridman adds his version: get captivated by a beautiful phenomenon, then be rigorous about the question underneath it, even if you do not yet know what the life’s work is.

Someone once told her physicists are not allowed to ask why. She does not believe that. The meaning of this planet’s existences, and of the universe, is that we are the creative process: the ability to create more possibilities and more things that exist. Fridman quotes Dostoevsky, that beauty will save the world, and asks whether creation and beauty are connected. Maybe. Why we have such an emotional response to intellectual activity and creativity seems intrinsic to what we are. She chooses questions because they are beautiful, thinks the universe is beautiful, and is deeply fascinated that she exists at all. That feeling of beauty may be part of the physics of creating more things. The conversation ends as it began, with life still going: Fridman thanks her, she hopes they talk again with Cronin, and he leaves Robert Frost’s three-word summary of everything learned about life. It goes on.