Sara Imari Walker on Assembly Theory: Life as Lineages of Propagating Information
The Long Now Foundation
What is life, and why do we still lack a language for it? In this Long Now Talk, astrobiologist and theoretical physicist Sara Imari Walker argues that we understand life about as well as humanity understood gravity before there was a theory of gravity. She calls the work she presents "a work in progress." Its centerpiece is assembly theory, developed with chemist Lee Cronin and other collaborators. The theory proposes that complex objects can only be produced by evolution and selection, that this complexity can be measured, and that objects should be understood as the histories that built them. After the talk, host Benjamin Bratton, director of the Antikythera program at the Berggruen Institute, questioned Walker about technology, minds, falsifiability, and whether AI could count as life.
A Host's Map of "Walkerism"
Bratton opened with what he called the key concepts of "Walkerism," offered as a guide to what was coming. He said astrobiology is not only about Martians or life in outer space. It is also about us, since we are an astrobiology ourselves. He said selection begins before biology and runs much deeper in time and process. He said "life" and "being alive" are different things. He described life as scaffolds building on scaffolds, becoming more complex over time, so that the newest thing is also the oldest: the complexity in an object reflects how much time it took to evolve. On that view, the most complex things in the universe, such as our technosphere, are the oldest.
Bratton added that, in this framework, technologies are a form of life, and a form we think through. They let us see the world differently and then remake it. The basic unit of life is not the cell but something closer to the whole planet. Finally, he suggested that discovering life, which in Walker's view has not yet happened, will be more like discovering gravity than finding an organism.
We Don't Yet Understand What We Are
Walker framed her research as an attempt to build new languages for understanding ourselves. Her analogy was gravity. Before there was a theory, people could not see that planetary motion in the night sky and the force holding them to the ground were the same phenomenon. In her view, we are in that position with life.
She started from the question "Are we alone in the universe?" People usually approach it through the size of space. The Hubble Deep Field covers a tiny patch of sky yet contains millions of galaxies, each with billions of stars. Walker said she had been told throughout her career that life must be common, given all that territory. But the only planet we know to be inhabited is this one. She described answering the question as a problem on the scale of centuries or millennia, and said she hopes her generation can move it forward.
Models of the Universe and Their Concepts of Time
Walker then looked at how we model the universe. Today we collect telescope data and build simulations of it. She observed that a spatial map of galaxies resembles a cellular automaton, a simple computational model popular in complex systems research. Her example was Conway's Game of Life, where people have been discovering patterns for decades. These models assume the kind of dynamical system Newton introduced: an initial condition plus a fixed rule that governs everything for all time. When researchers study emergent complexity in the Game of Life, they assume the rules never change and that nothing fundamental about the universe changes because of the patterns inside it.
Walker argued that this is not what we experience as living things. Human history shows that ideas and culture change the structure of reality around us. So a paradigm written in theoretical physics 300 years ago no longer fits the reality we actually live in.
She traced a series of models, each tied to the technology of its time. Clocks accurate to the second made it possible to map planetary motion precisely. That led to the discovery of elliptical orbits, Galileo's work on inertia, Newton's gravitation, and a clockwork picture of the universe. Walker's recurring point is that each new physical theory brings its own concept of time. Newton's time is linear, a substance the system passes through rather than an active property of it. Steam engines led to thermodynamics and a picture of the universe as a giant engine, where the second law gives time a direction. Walker noted that directional time sits uneasily with linear time and that this is still debated. In the last century, the popular idea became that the universe is a simulation. She pointed out that two kinds of relativity emerged in that period: Einstein's relativistic time and a relativity in computation, since the same program can run at different times on different computers. She suggested that "time is relative" was an idea circulating across the culture.
In her assessment, all of these models share one paradigm. The universe does not fundamentally change over time, and the rules are fixed.
Fixed Laws Versus Endless Forms
Walker set this against Darwin's theory of evolution. She said Darwin was frustrated that life had no laws as simple as Newton's, yet he still produced elegant explanations. She asked the audience to hold two ideas side by side. One is a predictable, deterministic universe with fixed rules. The other is Darwin's "endless forms." Can a universe as we currently understand it actually produce endless forms? If the real universe can, what would that physics look like? Our planet, she said, seems to produce them.
This is why she approaches "Are we alone?" through a different question: "What are we?" If we do not understand ourselves, or how we shape our own reality, we will not recognize anything like us elsewhere. We also lack a shared language to discuss the phenomenon of life.
She described a moment from her postdoc years. At a chemistry conference, a colleague opened a talk by declaring that "life does not exist." Walker saw this as typical of chemists working on the origin of life, for whom it is easier to assume the phenomenon away than to face it. She also raised the familiar paradox that no atom in your body is alive, yet you are. At origin-of-life meetings, she said, she felt existentially perplexed, because almost nobody discussed the origin of life. They discussed how to make parts of living things prebiotically, such as amino acids or proteins. Looking back, Walker described this as deconstructing 4 billion years of evolution and replacing it with intelligent design in the form of the experimenters' own choices. The question she thinks we should ask is different: how does the universe generate information and complexity when it has none? That, for her, is the origin of life.
Theories as Technologies, and the Science of Measurement
Walker asked the audience to think of theories as technologies. The models she had described shaped how people understood their place in the universe. She noted that some people even use fundamental physics as a basis for their daily philosophy. Ideas from theoretical physics may seem esoteric, she said, but they seep into the base layer of how a culture understands reality. So she believes physicists have a responsibility to cultivate theories that enable the future and offer optimistic views of what we are. Those theories also have to match reality, which she called the hard part of being a physicist. For her, science at its core is not about experiments and hypothesis testing. It is about building explanations of how the world works that last for centuries.
She then turned to metrology, the science of measurement, a field she encountered only recently. She said she never took a course on what a measurement is, even as a physicist, and that the subject is very complex. Her current view is that measurement is how we map abstract ideas onto the physical world, using devices that probe beyond our own minds for structural regularities. If an abstraction cannot be embedded in a measurement, it cannot be tested.
Lee Cronin's Problem: How Would You Measure Life?
Walker said she began working on the origin of life as a PhD student, reluctantly at first, because she had wanted to be a cosmologist. She came to love the problem because nobody had a conceptual framing for it. At the time, she thought theories of it might never be testable.
Then she met Lee Cronin, a chemist approaching the same problem from another direction. Cronin disliked how much design went into origin-of-life chemistry. Researchers searched for molecular structures found on Earth today and assumed they were relevant to the origin of life as a general process in the universe. They did not ask how unconstrained chemical systems generate complexity without anyone designing it. Cronin set out to build robots that explore chemistry "as messy as possible," hoping to see life emerge in his lab. That raised a question: if life did emerge there, how would he measure it? Walker described this question as the foundation of assembly theory.
The theory's conjecture is that life is the only mechanism the universe has for generating complexity. Walker drew out what she sees as the consequences. There are no Boltzmann brains and no spontaneously fluctuating objects. You exist nowhere else in the universe, because 4 billion years were needed to build you on this planet. She acknowledged this runs against current physics, which allows that anything could exist anywhere with low probability, that there might be another you somewhere, or that you might be a brain that fluctuated into existence. In her framework, complexity arises through evolution and selection. The space of possibilities is so vast compared with the universe that existence is special.
How Big Combinatorial Space Is: The Taxol Example
To make this concrete, Walker introduced combinatorial space, the space of all configurations of something. That could be LEGO, language, technologies, or molecules. She borrowed an example from Cronin: Taxol, an anti-cancer drug that is not especially unusual as molecules go. If you made one copy of every three-dimensional structure sharing Taxol's molecular formula, she said, the result would fill 1.5 universes of volume. Cheminformaticians cannot compute how many possible molecules exist, so the size of chemical space is unknown.
This leads to her central question. The universe lacks the time and resources to make every molecule with Taxol's formula, or DNA's, or every possible configuration of the chairs in the room. So why does this Taxol exist and not the alternatives? Walker's conjecture is that observed objects exist because evolution and selection have narrowed the space of possibilities. There is information in the history that produced one object rather than another. This is where she starts building an informational theory of life.
LEGO Castles and the Assembly Index
Walker asked the audience to imagine a pile of LEGO and to picture what they would build from it. She then showed a LEGO castle, one most people recognized. She asked how many had imagined something far simpler, and many had. Even imagining a complex object is hard, she said, because it takes many steps.
That is the idea behind the assembly index. Break an object into its parts, then rebuild it by joining parts, including pieces you have already built. The shortest such sequence is the object's shortest causal history, and its number of steps is the assembly index. Walker described it as a measure of how hard it is for the universe to produce the object. Castles, she said, do not happen spontaneously. The one on the slide required human cultural evolution, the history of castle building, someone writing a hugely popular book about a boy and magic, and billions of LEGO bricks on the planet.
For molecules, chemical bonds replace LEGO bricks. Walker gave the example of ATP, a critical molecule in biology, which has an assembly index of 21 by this procedure. The index can be computed from the molecule's graph. But Walker credited Cronin with the key insight. A mass spectrometer breaks molecules apart and measures the mass-to-charge ratios of the molecule and its fragments. If that process could reveal how complex a structure is, the index could be measured instead of only calculated. Walker said assembly theory began with thought experiments about mass spectrometry. According to her, Cronin's lab can now measure the same property with three techniques: mass spectrometry, NMR, and infrared spectroscopy.
A Threshold for Life and the "Life Meter"
Why does it matter that this can be measured? Walker returned to the size of chemical space. There are many ways to add one bond, so each assembly step moves into an exponentially larger space. What the assembly index captures is selection compressed into the objects that exist, against a background that keeps expanding. Because of how that space is structured, the conjecture predicts a threshold. Beyond it, we should never observe an object unless it was made by life, meaning by a selective, information-processing system.
Walker described a 2021 paper from Cronin's lab that built a "life meter" on this principle. The team measured abiotic samples, biological samples, and dead biological material. According to Walker, the only molecules found above an assembly index of 15 came from life. She said NASA was interested and sent blinded samples designed to trick the lab, including material from the Murchison meteorite, which she described as one of the messiest abiotic samples. The method still separated biological from non-biological samples. Walker said the results matched the theory's prediction of an abrupt complexity threshold. Abiotic material such as meteorites is what chemists call "tar," a random mixture so combinatorially complex that individual molecules cannot even be distinguished. Life, by contrast, selects specific complex structures and produces them in high abundance.
She pointed to an application: recognizing alien life by its constructed complexity, even when its molecules are unknown, which she called "life as no one knows it." But she said her main interest lies elsewhere, in explaining what we are and finding the fundamental physics of life.
Assembly Space as a Physical Space
Walker said the ideas that follow may seem to blend computational and material language in odd ways, and that this is intentional. She sees assembly theory as unifying information, usually treated as abstract, with physical properties. She noted that past unifications in physics changed how people thought, and she hopes this one will too.
She treats assembly space as a physical space. People created coordinate space by inventing rulers and measuring physical geometry. In the same way, assembly theory makes measurable a space that has seemed abstract: how causally deep objects are. Showing the assembly space of adenosine, a molecule important in genetic systems, she described it as both a material property, because it can be measured, and an informational one, because it captures how much information and selection went into the object's history. She described it as physical, measurable, and predictive.
She then laid out nested layers. Starting from what is actually observed ("assembly observed"), you can build an outer layer called the "assembly universe": every imaginable structure, including ones that break the rules. In LEGO terms, this means gluing bricks together any way you like. Inside that is "assembly possible," the structures allowed by the actual rules. For LEGO, that means real brick connections. For chemistry, it means real bonds and thermodynamic stability. Standard physics can handle this layer. Assembly theory adds another layer inside it, "assembly contingent": the structures the universe can actually build, because they are built along historically contingent paths.
Because evolution must build on what came before, and the space is too large to explore fully, Walker argued that things like us are deeply contingent. In her view, rewinding history even 100 years would produce a radically different present, and the vastness of the space makes it unpredictable. She added that our ability to reach other possible histories by examining our own causal structure may be one of the ways we create novelty. We can learn the rules of our own history and use them to build the future.
The Assembly Equation and the Origin of Life as a Transition
Walker described the "assembly equation," which measures how deep a system sits in assembly space. She said it depends on two things. It grows exponentially with the number of steps, since the space expands exponentially. And it counts how many copies of an object exist. A one-off fluctuation does not imply a lineage that can rebuild the object. What matters is that a planet can produce many structurally similar things. Humans are not exact copies of one another, but evolution reliably produces human-like structures.
With this, she defined the origin of life as the point where objects become too complex to arise without a particular trajectory building on itself. Beyond that point, history and memory must accumulate, selection must operate, and an information-processing system must feed back on itself recursively. Without that, you get only a combinatorial mess that tries every option, and for complex objects the universe cannot do that. It "has to make choices," which Walker acknowledged is anthropocentric language. "We are our history," she said. "That is the physical thing that is us."
Her definition of life is "lineages of propagating information": causal histories accumulated over many generations and embodied in physical objects. Known physics describes the region below the transition. Walker said the transition point itself is not yet known, but she thinks assembly theory is bringing us closer. On her account, life first arose in chemistry, the first combinatorial space a planet produces. But the origin of life is also an ongoing process that happens whenever a new combinatorial space appears, including in language and technology. It is a general process of generating novelty and complexity from combinatorial spaces, and that process is the physics she wants to find.
A Paper That Went Viral, and a New Language
The theoretical paper in Nature was, Walker said, her first scientific paper to go viral. She showed a favorite meme: a confused person next to the paper's abstract. She thinks part of the confusion came from language. She recalled intense arguments with Cronin and said the collaborators went through about 150 drafts. The language was internally consistent, she said, but used words in unfamiliar ways, because it was trying to bring together many conceptual threads into a new vocabulary. She hopes that makes it a new "existential technology."
From there she presented several concepts. First, time is material. Assembly theory's concept of time is that time is a property of objects; in that sense, we are 4 billion years old. To the objection that one cannot simply invent new materials, she cited Roger Penrose, who she said dislikes the word "material" because it implies we know what material is, and Madonna, who "lives in a material world." Walker said she is confident social reality is as real as elementary particles. Looking at the history of physics, she argued, what we call matter is whatever we have learned to measure and embed in theory. Mass and acceleration became the terms for motion because they could be measured reliably and turned out to be the right measurements, and finding them took a long time. So inventing new material realities fits that history.
Second, the fundamental unit of life is the lineage, not the cell. Her friend Michael Lachmann answers the question of his age by saying he is 3.8 billion years old, roughly when life is thought to have begun. Walker argued that reducing us to atoms strips out all that time and causation, which is where we actually exist. She described evolved objects as "bigger in time than space," packing 4 billion years of constructed complexity into something the size of a brain. She suggested reality seems increasingly strange because more of the objects around us are deep in time, while our senses evolved to handle space, not time. We did not grasp the curvature of spacetime for most of history, she said, and we do not yet grasp how deep the causal structures around us are. Information appears to jump between materials because it is structure embedded deep in time.
Third, on this view, the largest known causal structure in the universe is the technosphere. Measured by space, our planet is small; measured by causal time, it is enormous. Walker said she believes the human brain currently has the highest density of causal volume per unit mass on the planet, and that this explains our capacity for abstraction. Technology has not reached that density yet, because its causal structure is spread more thinly per unit mass.
Fourth, the universe does not predict the future; it constructs it. What it predicts is the past. Because the volume of things that could be created far exceeds what exists now, the present does not contain enough information to determine the future. Walker said we live in an indeterministic universe and that she considers this fully consistent with quantum mechanics, though she called that a separate conversation.
Fifth, history and knowledge determine what is possible. Launching satellites and making elements with high atomic numbers are both consistent with the laws of physics. But they happen here because of knowledge accumulated over billions of years and because intelligent beings evolved. Walker noted we see neither on any other planet.
First Contact in the Lab
Walker ended with the idea she said she likes most: "first contact in the lab." She described AI and aliens as two concepts we currently do not understand. Our technologies seem to be coming alive, but we cannot see the historical contingency that produced them. We can trace information through genomes but not from genomic biology into technology, even though, she said, information does pass between those substrates, since technology is part of our lineage. She called this "quantifying the ghost in the machine." To understand how we evolve alongside our technologies, she argued, we need to see that underlying structure.
So she expects that, if the effort succeeds, we will discover alien life first on Earth, not on another planet, through experiments that search chemical space for new forms of life. Showing a single network representing the chemistry of Earth's biosphere, she said we occupy a tiny corner of chemical space and that other planets may generate other kinds of living chemistry. The goal is to create entirely new lineages in the lab without human design, to learn "how the universe designs itself" and to isolate the physics of the origin-of-life transition. She said she does not think there is any description of the universe from the outside, only a universe constructing itself whose rules we want to learn. First contact, for her, is both an event and an existential technology.
She closed with an image of the layers society is built from. She joked that she loves fashion and theoretical physics, the top and bottom layers. Her serious point was that if we neglect the base layer, our fundamental understanding of reality, the whole structure becomes incoherent. That is why she thinks new fundamental physics is urgently needed to understand our moment, what life is, and where we are going.
Q&A: Causal Depth, Agency, and the Technosphere
Bratton began by asking whether the causation accumulated in the technosphere over billions of years corresponds to the causal agency it has now, its ability to do things that were not possible before.
Walker said the two are closely related: the things with the most causal power are those deep in time. But she said the causes you can act from are mostly behind you, while what lies ahead is a noisy horizon. She does not think the technosphere will have causal power of its own until a larger structure builds on top of it, which she described as a kind of post-selection. When Bratton noted that she had spoken of co-evolution, she agreed that the whole system, including us, is being co-constructed, and said agency is distributed in time as well as among agents.
She used free will as an example. You cannot be in Arizona right now if you are in San Francisco, but you can be there tomorrow if you planned ahead. Free will, in her view, is distributed over time and depends on the act in question, and the same holds for causal structures. These structures are recursively deep and layered, with many interacting at once. She joked that she should have been an artist, because it would be easier to communicate the pictures in her head.
Q&A: The Perceptual Horizon and AI as a Microscope
Bratton asked about her idea of a "perceptual horizon." Is the discovery of life, which she compares to the discovery of gravity, possible only once that horizon has grown wide enough?
Walker explained with atoms. The word comes from ancient Greek, and chemical elements were named atoms because they were thought to be the bottom level of matter. Later technology revealed structure inside atoms. She said string theory, currently a leading candidate for a fundamental theory, is not accepted mainly because we cannot yet measure at that depth, which she called a technological hurdle. Gravitational waves had passed through the planet for billions of years, but detecting them required Einstein's general relativity and then LIGO. The line of what counts as "fundamental," she argued, moves with our technology.
From this she concluded that fundamental things are not at the horizon itself but in the constructed complexity within it. We expand what we can know by building new instruments that probe the edges of our "causal bubble," and we must be a large enough causal structure to see things smaller than ourselves. She suggested that the current moment, when technology is animating more and more things, will help us understand what life is. She added that we cannot yet see patterns in our own history and culture because human history has not lasted long enough to show regularities.
Asked what technologies might make that possible, Walker named three: theories as existential technologies, including assembly theory as a joint effort; instruments in the lab; and, to some degree, AI as a measurement tool. Bratton noted that AI is often discussed as a scientific collaborator, while she seemed to treat it more like a microscope. Walker agreed. Microscopes show tiny structures and telescopes show distant ones. AI algorithms show large patterns, but only the surface layer at the edges of very large causal structures. In her view, assembly theory describes the underlying architecture that produces the data layer we now live in, and AI is the technology for seeing that layer.
Q&A: Minds, Abstraction, and Rejecting Platonism
On the place of mind in assembly theory, Walker said minds are very deep in assembly space. Her example was the perfect circle. It cannot exist as a physical object, since that would need infinite precision, but it exists as an idea. She thinks much of the combinatorial space exists in virtual causal structures that could not stand as physical objects alone. Because minds are so deep in time and can reach those abstract layers, they can manipulate more of their environment and bring more creativity into the world. For Walker, our capacity for abstraction is evidence of how deep in time we are.
When Bratton asked whether this was Platonism, Walker said she has an "allergic reaction" to it, intellectually. She described herself as a materialist, adding that she is unsure whether it is the Madonna version or the Penrose version, since both coexist in her mind. She sees no reason to believe anything exists beyond what exists here. If ideas and abstractions are treated as physical features of this reality, not residents of some imagined realm, she argued, we learn much more about what is happening. She wants to know what abstractions are as physical architectures. Mathematics is supposed to be a universal language, but she sees it as our minds' ability to shape the future through the patterns in our own architecture. She views mathematics, and science generally, as constructive rather than predictive: part of how the planet generates possibilities and moves forward. She finds this more useful than assuming an inaccessible reality that governs us, because it gives us more agency and more understanding of ourselves. She took the same stance on information. It shapes everyone's life, she said, so it makes no sense to treat it as non-physical.
Q&A: Falsifiability and Measuring Other Substrates
Asked what falsifiable predictions assembly theory makes, Walker gave two. First, the theory would be contradicted if objects of arbitrarily high complexity appeared without an evolutionary process, for example a cell phone spontaneously appearing on Mars, and that has not been observed. Second, she and Cronin have discussed testing whether every object has a minimum construction time. She called Cronin a brilliant experimentalist but said designing an experiment to show that something cannot fluctuate into existence and requires a history is a hard problem.
She added that tests are not everything. Gravity has many tests, but it is also a language for a regularity our ancestors observed long ago, like why we stay in our chairs. She sees assembly theory as a language for a similarly deep regularity and admitted she does not know how to test an explanation that seems obvious. The mathematical theory's features can be tested, but she believes the deepest work of science is building better explanations, and that science is more varied than it is usually portrayed. When Bratton brought up Paul Feyerabend's epistemological anarchism, she said she likes the idea.
An audience question followed: mass spectrometry may work for molecules, but how would one measure a technology or a culture? Walker said the theory is general but extending it to new substrates is hard. So far it has been done rigorously for molecules and for crystalline materials such as minerals and silicon chips, where it can precisely tell natural minerals from engineered silicon. The difficulty is that the assembly index has to be embedded in a measurement scheme. You must reconstruct how the universe actually produces the structure, including which causal constraints and physical laws are involved, not just your model of the causation. That is a substrate-specific task. Walker said this is where assembly theory differs fundamentally from computation. Computational models can be built for anything, but each selective mechanism has one assembly space, each with its own laws, constraints, and contingency, and assembly spaces can be nested inside one another. She said the goal is to demonstrate the theory in enough materials to show that life is a general phenomenon.
On culture and language, she said some languages are presumably more complex than others. The hard part is choosing the substrate for building a language's assembly space: speech, writing, or what is stored in computers. Viewed from outside, language is patterns some entities on this planet utter to others, and understanding it as a physical causal structure is highly non-trivial. She said she thinks it is possible but did not want to give easy answers.
Q&A: Substrates for Minds and When AI Becomes Life
The last questions asked whether other substrates can hold minds and when AI becomes life. Walker said the interesting substrates are those with enough combinatorial richness to be open-ended, perhaps uncomputable, like chemical space, which the whole planet lacks the resources to generate in a model. Those are the substrates capable of open-ended complexity.
She added atmospheres to molecules and minerals as a third substrate she is fairly confident about, because of the interest in detecting life on exoplanets. Atmospheres are not combinatorially rich; she said about 16,000 volatile molecules could plausibly appear in one, a small space. As a result, the selection constraints in an exoplanet's atmosphere would be weak. Living and non-living worlds would sit on a continuum, though she thinks a transition could still be detected. Substrates with deep combinatorial structure can enter an open-ended cascade, and those are what she would really call living. The very deep structures within such spaces can have minds.
On AI, Walker said it is shallow for now. The integrated structure of technology on the planet is getting deeper, but she considers any individual model a shallow system. She questioned whether prediction algorithms only seem causally deep because we, who interact with them, are deep, while their physical embodiment remains shallow. She said the real question is not about the output on a screen but what it is like to be a silicon chip running a large language model, and that none of us knows.
Bratton suggested that since humans built these models, they might contain much of our evolutionary time. Walker agreed that they contain a great deal of the evolution of language, but said much of that is only the linguistic component. Word meanings shift constantly; she noted that asking the room "What is life?" would produce as many answers as people. A predictive correlation between one word and others does not capture the conceptual foundation beneath it. In her view, meaning is built mostly from the causal richness of our physical architecture and history, not from predictive associations among words. She tied this to disinformation. People with different causal histories use the same words to mean different things and can cloud one another's perception of reality. At the surface layer, without seeing the structure underneath or someone's individual experience, it is easy to assume bad intent where there is none.
The session ended on the question Walker had left open: what is it like to be a large language model embodied in physical hardware? Bratton thanked her for offering assembly theory as a scaffold for what comes next.
Good evening. Welcome to our Long Now talk with Sara Imari Walker. I'm Rebecca Landow, executive director here at the Long Now Foundation.
I'm excited to now introduce to you our guest host for the evening, Benjamin Bratton, director of the Antikythera program at the Berggruen Institute. Benjamin and Sara are long-time collaborators and they're working at the very edges of their respective disciplines. And we're fortunate to have them here with us tonight as kind of voyagers at these new frontiers of thought. Thank you for being with us and enjoy and over to you, Benjamin Bratton.
Thanks, Rebecca. As you'll see, Sara's talk is going to be quite a wild ride. So, what I sort of come up with is sort of like, let's say, some key concepts of Walkerism that if you can kind of hold on to, this may help give you a little bit of a menu of what you're in store for.
First, astrobiology. We generally think of astrobiology as like the Martians of something out of space. No, it's also a way in which we think about us right here. We are the astrobiologies, life in relationship to this.
Second, selection, what we think of as natural selection, begins before biology. It's actually much deeper in time and deeper in process, which if you think about actually sort of changes everything.
Life, which is the main sort of the topic of Sara's work, is something that is different from being alive. Alive and life are actually different things, which she will explain in some detail. That life as life evolves and accrues, it builds upon itself. It's all about scaffolds and scaffolds building on scaffolds and scaffolds building on scaffolds that becomes increasingly complex.
The newest thing is actually the oldest thing as Sara will explain. So, the amount of complexity in something also speaks to the amount of time it took for this to evolve. The amount of time in something is the amount of complexity in it and so the most complex things in our universe like our technosphere are actually the oldest things.
Technologies with the things that we call technologies are themselves a form of life. They're not separate from life. They are a kind of life and they are also a kind of life with which we think. We think through the technologies which allows us to see the world in a different way which then allows us to remake the world in a different way. So, once again, it's life making life all the way on down.
What is the basic unit of life? Well, for some it was the cell. For Sara, it's the whole planet. It takes a planet to make what we recognize as life. All of the processes working together to come to constitute what we recognize in this form.
And then lastly, to discover life is going to prove something that ultimately I think in Walkerism hasn't actually happened yet. We'll be more like discovering gravity than anything else. So, those are sort of eight key ideas. If you can manage to hold all of these in your head at once to give you a little bit of a starting point to what you're about to experience. With that, I'm very happy to introduce Sara Imari Walker. Thank you.
I'm so happy to see everyone here tonight. Thanks so much for coming out. I think we're going to have a very fun evening. I'm really excited to share some ideas with you. This is all a work in progress and I want you to feel like you're on a part of the journey cuz really the research program is fundamentally about what we are as living things.
And what I hope that we can do with the kind of work that we're doing is actually develop new languages to understand ourselves. And so when I'm thinking fundamentally about the nature of life, I think we don't understand it yet. It's a little bit like what Ben was saying in the introduction. It's like think about how humans talked about gravity before we had an understanding of the theory of gravity. Right? We couldn't understand planetary motion in the night sky as being the same as what's holding us here on Earth. And so I think we don't have the language to understand what life is. We don't understand ourselves.
And so that's actually what we're trying to do. So the title of the talk is an informational theory for life. And I started very early in my career thinking about information, but it's transformed quite a bit. So I'm going to kind of show you the current state of it.
It starts with this question, are we alone? I think probably most of us in this room hopefully have thought about this at least once. Not are we alone like as a person, but are we alone in the universe?
And usually when we think about this question, we think about the immensity of space. All of the stars and galaxies out there. So the image in the background here is the Hubble Deep Field. And the Hubble Deep Field is a tiny image in the night sky as far as spatial volume, but it contains millions of galaxies and each galaxy has billions of stars. And I've been told most of my career that surely life must be common in the universe because there's lots of territory for it to emerge. But the facts of the matter are that we only know this planet is inhabited and we don't know if there's any life in any of those billions of galaxies out there.
And I would like to know. I don't know about you guys. Would you guys like to know? Yes, we want to know. Okay, good. All right, I'm in the right crowd. All right, good. The curious lot is here.
All right, fantastic. So how do we come to understand whether any of these stars, any of the planets that we found around stars in our own galaxy actually is inhabited.
I think this is a good question for the Long Now because I don't think it's a year-long problem for us. I don't think it's a decade-long problem. This is at least a century or, you know, millennia-long problem. We've been asking this question for a long time. And I'm hoping that our generation can get us closer. But currently, the way we understand the world is by actually collecting data from our telescopes and building simulations of that data.
All right. So, this is just a spatial map of the galaxies that I had in that picture. And in fact, when we start to model the universe and we try to understand it, we have kind of ways that we think about modeling and paradigms that we use that are centuries old in the way that we think about how the universe works.
So, this picture of the spatial map of the galaxy actually looks a little bit like this picture. This picture is a cellular automata. And so, for those that might not know what a cellular automata is, is a very simple computational model. And people in my field of complex systems, or one of the many fields I have worked in, that study emergent complexity, like to use these simple toy models because they display very complex dynamical patterns.
So, if I took this model and I actually ran it, this one particular one I'm showing is called the Game of Life. You might see a richness of the kind of patterns that are displayed. And this is just a very short clip, but you could go online after the talk if you're super curious and play with the Game of Life for hours. People have been discovering patterns in it for decades. It's actually really kind of an interesting niche culture.
But what happens in this model is it's actually assuming a kind of dynamical system that Newton came up with originally when he was thinking about trying to understand motion in the physical world and mechanical laws of motion. And so, what we do when we try to study the universe is we simulate it with our models. And the kinds of simulations we have currently assume some kind of initial condition and some fixed rule that governs the universe for all time.
And so when we study emergent complexity in the Game of Life, we're assuming the rules never change. We're assuming nothing about the universe fundamentally changes because of the patterns in it. And this is not what we experience as living things. We've lived enough history as humans to know that our ideas, our culture, the things we do actually change the structure of the reality around us. And so this paradigm doesn't fit us anymore. We don't have the right paradigm for understanding the reality we actually live in. It's not the paradigm written 300 years ago in theoretical physics.
And so the question that I'm interested in tonight, but just more fundamentally, since I've been working on the origin of life my entire career, is how does the knowledge of this particular problem transform this kind of a fundamental level of understanding that we have about reality?
And so we can think about the sort of models of the universe we've built over the centuries. So this one is a model of the solar system. And this was derived over many centuries. It started, you know, very early with epicycles and this idea of, you know, trying to build predictive models of planetary motion, which were circles inside of circles. And you'll notice that my image here is not even accurate cuz they're not ellipses. But you imagine some kind of mechanism for the universe.
And the reason we came up with this mechanistic model was because we invented clocks that could accurately keep track of time in seconds. And we used that technology to actually accurately map planetary motion. And then with that accurate map, we were able to realize that they were ellipses. Galileo actually figured out about inertia and Newton came up with laws of gravitation. And so we have this model of the universe that came actually inspired by the idea of mechanical clocks in some sense, of the universe as a mechanism, a clockwork mechanism with the laws of planetary motion.
And interestingly, sort of a theme that you're going to see throughout the talk is that each new theory of physics we develop comes with its own concept of time. And the time we inherited from Newton's generation is that time is linear. It's a substance the universe moves through. If you notice planetary motion, we have laws, the laws remain fixed, and time is something that this system actually passes through. It's not an active property of the system.
Later we came up with the laws of thermodynamics, again trying to look at the real world and understand it. We invented steam engines. We needed theories that actually helped us understand how they worked. This became thermodynamics, and we started to think about the universe as a giant engine. The second law of thermodynamics rules all, and time has a direction. The time has a direction is a little bit antithetical to the time is linear, and there's still discussions ongoing about that, but fundamentally what thermodynamics is telling us is that time is directional.
And in the last century, we had sort of a new paradigm emerge, that the universe is a simulation. So this is sort of the popular zeitgeist that we're living in a computer simulation. And interestingly enough, there were actually two theories of relativity that emerged in the last century. The one that's familiar to most of us is the special theory of relativity and general relativity, which gave us the concept of relative time in physics. But there's a relative time in computation as well, which is the fact that the same program can run at different times on different computers. And so I think this idea of time is relative was just kind of an emergent cultural thing in many of our theories that emerged in the last century.
Now, each of the models of the universe that emerged with the technologies of the time that I just showed you, I think still is in the same paradigm, that the universe doesn't fundamentally change over time, and the rules are fixed.
Interestingly enough, we have a theory that emerged in the 1800s as an explanation for biological forms, the theory of evolution, which Darwin is most famous for. And he actually, you know, was really frustrated by the fact that we didn't have laws as simple as Newton's laws to describe life, but he still came up with very elegant theories and ideas about how to explain it. And the juxtaposition I want you to hold in your mind is this idea of fixed laws, fixed rules that govern our universe. The universe is predictable, deterministic, with this idea of endless forms.
What does it really mean to have endless forms? Is that really possible in the universe as we understand it now? Is that really possible in the actual physical universe? And if the real universe can do endless forms, what would that kind of physics look like? Our planet seems to have that.
So, we're going to talk about the question, are we alone? But we're going to talk about it as a lens of what are we? Because I think if we don't fundamentally understand ourselves and we don't understand how we're architecting our reality, we don't understand what we would look like from the outside. We're not going to be able to recognize things like us in the universe. We're not going to know what we're talking about cuz we don't have the languages to talk to each other about the phenomena of life.
In fact, in my field, this is really funny. So, sorry. I'm laughing already because I just remember being a postdoc and going to a conference at a chemistry meeting and one of my colleagues stood up and said, "Life does not exist." And this was sort of the opening of their talk and this is the culture of chemists working on the origin of life is it's much easier to assume the phenomena of life doesn't exist than to tackle it head-on. And in fact, there's similar things that, you know, once you look at molecules, it's very clear life is not there and you can just reduce everything to the parts of living things.
But of course, there's also the paradox that no atom in your body is alive, yet you are alive as a living entity. And I certainly, I don't know, how many people in here think they're alive? Okay, cool. All right, good. I didn't see any chairs raise their hands, but I saw most of the people. Okay, I think we're getting the categories moderately right.
I personally felt like deeply existentially perplexed. I would go to these meetings on origins of life and almost no one was talking about the origin of life. They were talking about other problems. How do you make a protein or an amino acid, which is a component of a protein? Like how do you make the parts of living things prebiotically? If we take everything apart about what we are and we try to engineer it in an experiment. To me, as I think about it now, it's not how I understood it then. Basically, they're deconstructing 4 billion years of evolution and they're using intelligent design to replace it. And this is not the question we need to ask. The question we need to ask is how does the universe generate information and complexity when it has none? That is the origin of life.
And I think that question has some really interesting consequences, which we're going to explore tonight. And one thing I want you to think about when I'm talking about the concepts is the idea that theories themselves are technologies. So, one of the reasons I walk through these models of the universe is that each one has framed fundamentally how we think about ourselves and our place in the universe.
So, imagine using fundamental physics to inform your daily life philosophy. Now, it's not something everyone does, but some people do. And the reason I like picking on this example is even if theoretical physics and the ideas that come out of that exercise seem very esoteric to you and part of a small part of a community that has little to do with the daily workings of everyone else on the planet, these ideas percolate and they are part of the sort of base layer of understanding of our reality and they become part of human culture over time. And so, I think we have a deep responsibility to actually cultivate theories as technologies that are actually enabling for the future.
And that our theories themselves need to paint optimistic views about what we are in our place in the universe. They also need to actually correspond to reality, which is the hard part of being a physicist.
For me, science fundamentally, when you really think about science as a human endeavor, it's not about the experiment and the hypothesis testing. It's about building explanations for how the world works that last centuries. That's what we try to do as scientists
at the base level of what we do. And all of the things that individual scientists do, we're trying to fill in that narrative. It's the narrative of reality that we want to tell ourselves.
So, I'm going to start talking now about life and the work that I'm doing with my collaborators. Metrology is a word that I've only run into recently. I like challenging myself by suddenly immersing myself in different fields, and metrology is the science of measurement. And you would think as a physicist I would have understood what a measurement was in my undergrad education. I can tell you I did not ever get a course in what a measurement is. And if you get into the science of measurement, it's incredibly complex.
And so, one of the ways I think about measurement right now is that measurement is the way that we take our abstract ideas and we try to map them actually to the physical world through devices that allow us to probe the world outside of our own mind and look for structural regularities. So, measurement is actually incredibly important because it's how we map our abstraction to physical reality. And if your theory doesn't correspond to measurement or you can't embed your abstraction in a measurement, you have no way of testing it.
And when I started working on the origin of life when I was a PhD student, very reluctantly at first actually, you know, I wanted to be a cosmologist. I thought that was very romantic. And then I realized, you know, origins of life was something that no one had an idea about. No one had any conceptual framing for this problem. And that actually became the love of my life conceptually and intellectually, is this problem.
Okay, so when I was working on it at that early stage, I thought we might be able to come up with theories, but we would never be able to test them. It turns out there was someone else on the planet, Lee Cronin, who's a chemist, who was thinking about the same problem in a very different way than I was.
And Lee's problem was that he didn't like the way that anyone was doing origin of life chemistry, because of this issue of the amount of design we were putting into the experiments. We were looking for things that are on our planet now, molecular structures, and assuming that they were relevant to the origin of life as a general process in the universe, rather than looking for how unconstrained chemical systems generate complexity without any design put in by us or anything else.
And so Lee is trying to build robots in his lab that can explore chemistry. As messy as possible, the messier the better, because he wants to see life emerge in his lab. And he says, "Well, if life's going to emerge in my lab, how will I measure it?" And so this is actually the foundations of the theory that I'm going to introduce to you. It's called assembly theory.
And the conjecture of the theory is that life is the only mechanism the universe has for generating complexity. There are no Boltzmann brains, if you know that concept. We don't get fluctuating existence of objects like in this room. You will exist nowhere else in the universe. There's not a multiverse of possibilities. You only exist here, and you exist here because 4 billion years was necessary to construct you on this planet.
That's the framing that we have. So it's very contrary to current physics, which has this idea that everything can exist everywhere, and there might be another you out there, but it's just a very low probability, or you might just be a brain that fluctuated into existence, and your experience right now is not real. All of those things are not consistent with the paradigm that we're building.
What we think is that complexity happens because of evolution and selection. And the space is so huge, and the universe is small by comparison to all the things that could create, that existence is really special. If you get to exist, you are like, that's amazing. Okay. So, yes. Woo! What up for existence? All right, cool. I like this enthusiasm. I also like existing. It's great. Better than the alternative. All right.
So, life is the only way to generate complex objects. And so, how did this happen? Okay, so I just want to go through this argument. I just told you that you probably don't exist anywhere else. Great. I'm glad we're here. But what are we actually talking about?
So, I think when I show you the Hubble Deep Field, I think most of us have a sense that the universe is big, right? Because we're told it's huge. It's 13.7 billion light years across, and there's billions of galaxies out there. We don't actually think about how big the space of possibilities a single planet can generate.
Combinatorial space, I'm going to talk about a lot. You can think about it as a space of all configurations of something. So, I had to learn chemistry this way. But you can also think of LEGO, or you can think of language. Think of all possible ideas you could build with human language, or you can think about all possible technologies. You can think about all possible molecules.
Combinatorial space is really big. It's actually uncountably big. I'm going to give you an example of one molecule, and I borrowed this example from Lee, which is Taxol. So, Taxol is a molecule that's been discovered on Earth. It's not a particularly special molecule. It's used as an anti-cancer drug. But if you wanted to make every version of a molecule, a three-dimensional structure that corresponds to the same molecular formula, and you had just one copy of each molecule, that one molecular formula would fill 1.5 universes of volume. One molecule.
So, actually, when cheminformaticians try to compute the number of possible molecules, they can't. We don't know how big chemical space is because we don't know how many possible molecules there are. And so, this gets into an interesting question because it's like, why does Taxol exist on this planet and not the other possibilities?
Clearly, our universe doesn't have enough time and resources to make every single molecule with Taxol's molecular formula. It doesn't have enough resources to make every molecule with the same molecular formula as DNA. It doesn't have all the resources to make all the configurations of what could be the chairs in this room. It can't exhaust all possibilities.
So, this raises a fundamental question about what is the mechanism of what gets to exist. Why are these structures the ones that actually we observe as physical objects, and these other ones we might imagine? We might be able to say something about chemical space because we observe a molecule. So, we have this idea of this counterfactual space that could exist, but we actually don't observe it to exist.
The things that do exist, what our conjecture is, is they exist because evolution and selection have constrained that possibility space. There's information in the history that makes this object and not this object. And this is where we start to get into an actual way of building an informational theory of life.
So, to explain the idea, I'm going to start with LEGO because LEGO's much easier for most of us than chemistry. Probably most of us have had a pile in our home that looks like this if you have children or if you just like having fun. If I was going to shake the LEGO set indefinitely, what kind of structures do we expect to arise out of this? Can all of you hold an image in your mind of something you might build out of these LEGO? Great.
All right, I'm going to also make something out of these LEGO. How many of you made this? Just curious. I've never done this before. Did anyone make this object? No, okay. Oh, we got one. Oh, excellent. Okay, cool.
Probably we already made this object because it's been imprinted on most of our minds. Most of you probably recognize it. It's a castle at least. I hope most people know it's a castle. Okay, good. We do live in the same selected history. Did most of you make objects that were much simpler than this? How many people made something very simple? Okay, so even trying to make something very complex mentally as an exercise is actually quite hard because it requires a lot of steps.
And this is actually the way that we think about formalizing the complexity of an object. So, I'm going to introduce assembly theory now and I want you to think about the LEGO analogy. But the idea is if you take something down to its parts and you take the parts and you put them together and then you keep doing that by putting parts together that you built already, you will have a sequence of steps that's a shortest causal history for making that object, and we call that the assembly index, and this becomes how hard it is for the universe to produce that object. That's all you need to know about it.
Castles are hard. They don't happen spontaneously. They require a lot of evolution, and in this case it requires a lot of human evolution because we had to have cultural evolution. We had to build castles. We had to have somebody write a book about a little boy and magic that became wildly popular, and we also had to have LEGO and billions of them on the planet, and then we get that. Okay, so that's not an easy object for the universe to construct. It requires a lot of history.
Now we want to formalize that idea. How do you formalize this in an objective way? A way that you can actually do physics with. So, this is where we get into the idea that there's a property, this minimal number of steps to produce an object. I'm showing it here now for ATP, which is a critical molecule in biology. For ATP, the minimum number of steps where you take bonds as fundamental units instead of LEGO blocks, and you put the bonds together, and then you put the things you've made already together to get ATP, is 21 steps. Now, I can compute that for a graph of the molecule once I understand the physics.
But here's where what Lee did was really clever. He was in the lab, he was trying to build robots to explore for origins of life, and he said, "I have a mass spec, mass spectrometer. I can do measurements on molecules by breaking them apart and looking at the mass to charge ratio of the molecules and the fragments. If I can use this to probe how complex these structures are in a way that I can measure, then I can do something interesting."
And so, this is actually the genesis of assembly theory, with thought experiments on mass spec and the fragmentation that mass spec does to resolve the causal structure of molecules. And the exercise is that this process of breaking a molecule apart and trying to resolve the minimal steps to rebuild it from those elementary parts actually can be mapped to measurements in the lab.
And actually, Lee's lab has done it now with NMR and infrared and mass spec. So, there's three different ways that you can measure the same property of molecules. And it allows you to say the minimal amount of causation, evolution, selection, information, whatever you want to call it, for this molecule.
Why is this important? Why is it important that we can measure this kind of constructed complexity or this causation? It's important because of this conjecture we had that life is the only thing in the universe that can build complex objects. So, remember how big the space of Taxol is.
Now, if you're in chemical space and you add a single bond, there's many ways you can do it. And so, every time you add a single bond in those assembly steps, you're actually increasing the complexity in a space that's exponentially growing. So, the space gets larger at every step that you take. So, if you do an assembly index increase of one, you've moved into an exponentially larger space every single step.
So, what we're measuring with this causal measure is actually this compression of selection into these objects that get to exist against this exponentially growing background. And our conjecture is, because of the structure of that space, there should be a threshold that we should never expect to observe an object unless it's the product of life. That there was actually a selective mechanism, information processing system. That was where we went into it.
And so, Lee's lab is very clever. So, this is just sort of a depiction that you can take a certain number of steps and then suddenly you might have something that's alive. Done for origami. The principles of assembly theory are quite general.
They actually went and tested it in the lab. So, this is a paper that was published in 2021, building a life meter with this idea of molecular assembly. And so, they went and they took a whole bunch of samples, abiotic, biological, dead. The conjecture actually held up to the experimental test.
So, what was done here is the molecular assembly that I just told you, the minimum number of steps, is shown on the x-axis on this plot that says life. The only molecules that we found above assembly index 15 were products of life.
So, if you take abiotic samples, and actually NASA was really interested in this, so they sent samples to Lee's lab that they had blinded, and they really tried to trick him. They sent him samples of Murchison, which is one of the messiest that's abiotic. It's a meteorite. Very messy. And what they were able to do is use this approach with the measurement techniques that they developed to be able to distinguish the biological from non-biological samples.
And it conformed to the theory's conjecture that there should be a very abrupt complexity threshold, and that living things were the only things that could produce things above this background, because the space is getting so exponentially large, we should never observe those objects in high abundance enough to detect them.
And if you go into pure abiotic examples like meteorites, what they are is what we call tar. They're just a total random mess of molecules. You can't actually even differentiate some of the molecules because there's just so much combinatorial complexity. And what life does is select out specific structures in high abundance that are very complex. So, we can measure that. That's pretty cool.
It's very cool for me, because as a theoretical physicist, I live in abstract land most of the time. What we're trying to do is actually look at this and use it now to build a theory. And one of the things that we're excited about is this idea of recognizing alien life as constructive complexity. So, there's clear applications to looking for life on other worlds if we don't know the molecules. We can actually go in and we can discover life as we don't know it, even life as no one knows it.
And that's exciting. But what I'm really after is that explanation for what we are, the fundamental physics of life. And I'm very excited about where assembly theory can take us with that.
So, what I want to get to now is using this fundamental idea of the kind of complexity assembly captures, this idea that there's a minimum number of causal steps to produce an object, and this somehow tells us about how hard it is for the universe to generate that structure. We can use it actually to formalize some really interesting theories and philosophies about the nature of life. And it's going to be a little bit counterintuitive. Most of the history of physics is.
And so, a lot of the ideas I'm going to talk about seem to be merging computation and material languages in a really weird way. And that's kind of on purpose, because I think what we're doing by taking these measurements that are of this causal structure, some people might call it a computational structure of an object. I don't like to think about it that way, but you might. What we're really doing is doing a unification of the ideas of information as an abstract property and thinking about it now as a physical property. And if you look at the history of physics, every time we do one of these kinds of unifications, it really changes our thinking, and so that's what I'm hoping is on the frontier here.
So, I'm going to talk about assembly space as if it's a physical space because that's the way that I think about it. So, we have invented many kinds of physical spaces in our history. One of them is coordinate space, and we invented that by inventing rulers and actually measuring the physical geometry of our environment. Assembly space, you can
think of this idea of the assembly index, actually measures things in this what seems to us to be an abstract space, but actually has become measurable through the advent of assembly theory, of how causally deep objects are.
And if we can measure that, then we can start to orient all of these complex objects in our world and root them in a physical space, and that's actually what we're trying to do. I'm showing here an assembly space for adenosine. It's a molecule that's very important in genetic systems, and what I want you to take from this is that this idea of the assembly space is actually a material property. It's one we can measure, and we can also think of it as an informational property.
Because what it's talking about, what it's capturing, is how much information and selection are in the history of that object in order for it to exist. So, it's a physical space, you can measure it, it's predictive.
What does it tell us about evolution and selection as fundamental mechanisms in our universe? So, in the introduction, Ben also mentioned this idea of selection before biology. And so, we're really after this idea that all the structures that the universe creates are selected to exist, and they might exist in a larger space. And so, we have some objects we can observe. They might be molecules, they might be a LEGO castle, and we can construct a space of this minimal causal history, the assembly space, what we call assembly observed.
This space tells us a lot about the structure of what doesn't exist, which is super interesting. And this is where it gets into a really intriguing philosophy and theory of information. So, I'll walk you through it very briefly and kind of quickly. We're in like the deep weeds of theoretical physics now. First layer that you can build out of that space, imagine I take all of the LEGO blocks and I build every single possible structure, but I don't obey the rules of the LEGO universe. I can just glue them together. I don't have to use those little nubs or like whatever they are. So, I can build the space of all imaginable structures. Some of them are non-physical by the laws of the LEGO universe. That's what we call assembly universe. It's an imagined space. So, somehow encoded in these objects is actually our ability as intelligent agents to extract this kind of space that it could exist, but it doesn't.
The second layer is assembly possible, which includes all the structures that are actually possible to construct with the rules of that universe. And so, for LEGO, it would be actually using the building block brick rules to construct objects, not just assuming that they could stick together in arbitrary ways. In chemistry, it's using real chemical bonds and looking at thermodynamic stability. So, there's a set of possible structures that are consistent with our laws of physics. Those would be consistent with standard physics also. But what assembly theory offers is a second layer inside of that.
So, this is kind of the third layer in the nested assembly universe, which we call assembly contingent, which is the observable structures that the universe actually can build because they're built along historically contingent trajectories. So, this idea of the assembly space and looking at the structure as actually being a physical feature means when you're actually building an evolutionary system, it has to scaffold on the layers that came before. And because that space is so exponentially large, it can't exhaust every layer. So, by the time you get to things like us, we are so historically contingent if you rewound history 100 years, we would be radically different right now. Because the space is so exponentially large, it becomes unpredictable.
So, the real history we observe is just one such history within the structure. And our ability to access some of the other histories by looking at our own causal structure is actually one of the mechanisms I think that we use to create novelty in the future. We can learn the rules of our own history, and we can use them to actually build into the future.
So, we have a way of measuring how deep you are in this assembly space. We do it with this equation we call the assembly equation. It really only has two properties you need to worry about. It's exponential in the number of steps because that space is exponentially expanding, and it cares about how many copies you see of an object. The reason being a one-off fluctuation doesn't imply a long lineage of things that actually could rebuild that object. So, what we're interested in is the causation on a planet like us that can make many things that are similar in structure. So, human beings are selected structures that exist on this planet. We're not exact copies, but there's many of us because evolution actually has a reliable mechanism for generating human-like structures. And so, this equation is actually capturing that feature in this very abstract space and allowing us to figure out ways to measure this as a generic property of physical systems.
Now, I mentioned this idea of a threshold in this exponentially expanding space. So, this is how we formalize the origin of life. The origin of life is when things become too complex that a specific trajectory has to actually scaffold on itself in order to get to that level of complexity. It has to build a history and a memory. Selection has to happen. There has to be an information processing system that feeds back on itself recursively in order to construct into the future in this space. That's the origin of life. Otherwise, you get a combinatorial mess of exploring all possibilities and our universe cannot do that for complex things because there are too many to exist all at once. So, it has to make choices. It's very anthropocentric language, but this is why we get to exist and other things don't because our entire history has led to our existence and in fact, we are our history. That is the physical thing that is us.
So, the definition of life that I'll offer tonight is that life is lineages of propagating information. It's actually these causal histories accrued over many generations and actually physical in objects. And known physics exist down here. We don't know this transition point yet, but I think assembly theory is allowing us to get close.
And the origin of life happens in chemistry originally. It's the first combinatorial space a planet generates where it can exhaust all possibilities and if you want to see complex structure, life has to emerge on a planet. And then, the origin of life actually is a continual process that happens anytime a new combinatorial space is built. So, I actually think the origin of life is a process that happens in languages and in technologies. It's a general process of generating novelty out of combinatorial spaces and then allowing complexity to emerge out of them.
So, this is the physics that we're after is that actual process.
You might be scratching your head. It's a little abstract. It was quite funny when our paper came out. We had a paper in Nature that was very deeply theoretical, which was like very exciting. I have never had a scientific paper go viral before. So, this was my favorite meme. And that's the abstract of our paper introducing the ideas of assembly theory. Obviously, he's very confused. You might be saying similar things. It's a lot of things that we folded in to really try to get at this problem of how life arises in the universe. So, I think it's a complex set of topics.
I want to talk for a few minutes before wrapping up about the idea of assembly theory as an existential technology cuz I think one of the reasons that our paper was so confusing is the language didn't fit. We wrote the paper in what I felt like as a self-consistent language, and I remember these very deep arguments with Lee. I think we wrote 150 drafts of the paper with our collaborators, and it was very intense. The language, I think, was self-consistent, but we used words in ways that people weren't familiar with. And I think this is because it is a new idea, and we're trying to really pull a whole bunch of undercurrents and conceptual frameworks together and try to build a new language. And therefore, hopefully, it offers a new existential technology in the sense of the history of physics.
So, I already mentioned that new theories offer new concepts of time. Assembly theory has its own concept of time, which is time is material. Time is a property of objects in the sense that we're 4 billion years old.
And you might say, "I'm not allowed to invent new materials." I know some people that would beg to differ. Roger Penrose, Nobel laureate in physics, does not like the word material because it suggests we know what the material is. And Madonna lives in a material world. And I'm pretty sure social reality is equally real to elementary particles. And so, I think if you look at the history of physics, the things we call matter are things we invented measurements for and could embed abstract ideas in theories. So, why is it we use mass and acceleration for understanding motion? It's because we could measure those things in a very regular way, and they happen to be the right measurements to actually build into our theories of motion. But it took us a long time to figure out what measurements map to the abstractions. And so, inventing new material realities, I think, is something that's very consistent with the history of physics.
Another concept, fundamental unit of life is not the cell. It's the lineage. And as I mentioned, you have parts of you that are 4 billion years old. So, I have a friend, Michael Lachmann, that if I ask him how old he is, he'll say he's 3.8 billion years old. Well, that's when we think the origin of life happened. He looks good for his age. You all also look great for your age.
So, the idea being that we have been constructed on this planet across 4 billion years, and the reason we're not reducible to our atoms, why that fundamental layer doesn't describe us, is because when we do that, we remove all of that time and causation. And that's actually where we exist. That's where the causal structure is that is us. It is embedded in this history of time. And so, I think to understand life, we have to think about time in a real physical way.
And actually evolved objects, one of the reasons they're so perplexing for us to see is because they're bigger in time than space. Imagine putting 4 billion years in this tiny volume of my brain. Right? That's what we are. 4 billion years of constructed complexity inside a very small volume. I think this is one of the reasons that reality is looking increasingly as we become increasingly complex, because most of the objects in our world are now deep in time, and our sense perception has evolved to interact with the spatial dimension, not the temporal dimension. Right? And so, the power of theoretical physics has always been to build abstractions that correspond to reality and allow us to see reality in new ways. And we didn't really understand gravity, we didn't understand the curvature of the space-time around us for most of our history, and I think we don't understand the causal structures that we're embedded in and actually just how deep they are. And so, information looks like it hops around between physical materials because it's actually the structure that is embedded deep in time, and it's just around us everywhere.
If you adopt this view, the largest known causal structure in the universe, collectively, is our technosphere. So, if we looked at our universe in spatial volume, we look very small. But if we look at it in terms of causal time volume, our planet is huge. And in fact, I think right now on our planet, the largest density of causal volume per unit mass is the human brain. And this is one of the reasons that we can abstract so well. And we haven't manifest that yet in our technology because the sort of causal structure is much more distributed in terms of energy per unit mass, etc.
The universe doesn't predict the future, it constructs it. What it does is predict the past. The future is undetermined till it happens. Why? Because in the complex universe we live, the volume of things we can create is much larger than what exists now. There's not enough information existing now to specify where we're going. We actually live in an indeterministic universe. I think that's completely consistent with quantum mechanics. That's a whole separate conversation. Okay.
History and knowledge determine what's possible. So, I'm showing two processes that are consistent with the laws of physics, launching satellites into space, building high atomic number elements. Both of those things can happen on our planet because of the knowledge that's been acquired over billions of years by and then evolving intelligent systems like us that invent laws of physics that can launch satellites into space, something we don't see on any other planet, and can create elements that only exist here.
All right. Last thing I want to introduce, which is one of my favorite, is first contact in the lab. So, we talk a lot about AI and alien, right now, two concepts we don't understand. This is one of the reasons that I'm really interested in solving the origin of life. And most of the time when I say this, I go places and people like, "What are you talking about, Sara? What does that even mean?" I'm going to try to explain what I mean a little bit better than I have before. I'm hoping I'm getting there. But the idea is we actually want to understand how the universe constructs complexity, and we want to see that history. I want to see the causal structure underlying us. I want to understand what that physics is. So, quantifying the ghost in the machine is we see all of these things being animated in our environment right now. All of our technologies are coming alive, but we actually can't see the historical contingency that got there. We can trace information across genomes, but we can't trace information from genomic biology into technologies. But, there's information transfer between their substrates, right? Our technologies are part of our lineage. And so, we need to be able to actually see that underlying structure in order to understand how we're evolving and co-evolving with the technologies that we're creating.
So, I think we're not going to discover alien life on another planet first. I think we're going to discover it here on Earth first. Hopefully, if we're successful. By building experiments to search chemical space for new life forms. The idea being, if we can do that, we can isolate the physics of that origin of life transition and really understand how is it that our universe crosses that threshold and starts to build information processing systems and constructive complexity.
We're a tiny, tiny fragment of chemical space. This is the Earth's biosphere's complexity of chemistry drawn as a single network, planetary chemistry. We exist somewhere in this giant space. There are probably other kinds of chemistry. Planets can generate other kinds of living chemistry. So, could we build experiments that can build entirely new lineages totally different from the one that emerged on Earth without putting the design in? Without using ourselves as intelligent designers. We want to know how the universe designs itself.
So, we're trying to build experiments that allow searching for the origin of life in the lab and understanding the source of novelty and non-determinism underlying creation. I actually think there's no description of the universe from the outside. It's constructing itself, and we want to learn the rules of that.
So, first contact for me is an event. It's actually contact like realizing we make a discovery of alien life, but more important is that first contact is an existential technology.
And to conclude, I'm going to use this like very good in this crowd, about the sort of structure of the different layers that we all exist in. And part of my reason for this is I love fashion, and I love theoretical physics, so it's the top and bottom layer. But, I also really deeply think, as I mentioned before, that if we don't work on the base level of our understanding of
reality, the entire structure starts to become incongruent. So, I think this idea that we need new fundamental theoretical physics to understand the time that we're living in now, and what life is, and where we're going is critically important for this reason. And so, with that, I'm going to thank my lab and my amazing collaborators and all of you, and thank you guys.
That was amazing.
Thank you.
It's always so fun to... They're great. Always so fun to listen to you. I always learn so much every time. So, I have obviously a million questions. Let me maybe start with the technosphere and this really wonderful sort of paradox that actually turns out not to be a paradox after all. That the technosphere, which has the largest causal structure embedded within it, which is usually thought of as sort of the newest thing in this world, is actually the oldest thing in the universe because it has the most causal structure running with it. And I wonder if you could talk a little bit about, differentiate, or maybe it's the same thing, whether or not the causality, that sort of accumulated causation that has accrued over these billions of years that is embedded within this, that's one kind of causality.
But now the technosphere itself is also causing things, right? It has a kind of agency, perhaps mindless agency, that's able to do things that also previously couldn't be done. So, in your mind, is there a kind of necessary symmetry between the amount of, let's say, you know, involved causality that's in something and the amount of causal agency that thing might have?
Yes.
Okay, so we're going faster than I thought.
Yeah, now it was fun while you were talking, which often happens when I have conversations with you. I build really nice visuals in my head. So, I think the way... So, they're absolutely related. So, things that have the most causal power in the universe are things that are deep in time. But it's actually very interesting because most of the time you can actually act on your past history. So, where your causal... are the things that are behind you in time, and the things ahead of you are kind of this noisy horizon. So, I don't think the technosphere will have a causal power of its own until there's a larger structure on it. So, it's almost like it's post-selected.
Until there's a larger structure on it.
Yes.
In the future at some point.
Yes.
What would that be?
...still have most of the causation.
Does that make you happy or sad?
Well, both. I mean, I suppose like you... What you also talked about, you ended with saying that we are co-evolving with our technologies, right? And that it's selecting on us and we're selecting on it, right, as well. So, maybe it's both.
Yes.
Yeah.
I mean, the whole system is being co-constructed.
For sure.
And so, I think... Yes, including us. And so, I think the question of agency is quite hard because agency is very distributed.
For sure.
And it's also temporally distributed. So, I think it's always interesting, you know, the places it also comes up is like in questions of free will, and there's like an issue about whether you have free will or not, but free will is a temporally distributed property and it depends on the actual causal act that you're aiming to do. So, you know, I couldn't be in Arizona right now because I'm in San Francisco, but I could be in Arizona tomorrow because I planned ahead, right? So, your free will is distributed over time, and I think it's the same with these causal structures, that the actual causation they have is distributed in time. So, it's recursively deep. You could think about one structure, but it's many such interacting structures and they're all layered.
Yeah. Yeah.
That's clear to me anyway. I don't know if any of you...
...made that up.
I should have been an artist. It would have been much easier to communicate.
Yeah.
One of the things that, you know, you've been working on lately is this idea of the, I think, perceptual horizon.
Yes.
...horizon, right? Maybe I ask you to explain that a little bit more. As I understand it, it has to do with the ways in which, and this goes a bit to the history of science that you showed, that different technologies allow us to perceive the universe in different ways. And because we can perceive the universe in different ways, we can measure it in different ways. And this in essence both enables and constrains the kind of science that's possible, right? And so, this horizon is expanding, right? And perhaps computational simulations are the latest version of it. But the idea is that this is not only changing, but in essence getting wider or getting bigger.
Yeah.
And so, I wanted to ask you to explain that a little bit more, but in the context of this question. You've said a number of times that, you know, you imagine that when we, or when you, perhaps, you and Lee, discover what life really is, it'll be something more like the discovery of gravity.
Mhm.
Will that only be possible because the perceptual horizon has expanded to such a degree that that level of conceptual granularity is possible?
Yeah. Yeah.
So, let me explain, for people that haven't heard me talk about that idea before, where it comes from. So, I usually use atoms as an example, right? So, when atoms as elements were discovered, we thought they were the base level of reality, right? So, atom comes from the ancient Greek as a word, and when we discovered the atomic elements, we called them atoms because we thought they were the substance everything was made of. But, subsequently, we developed technology that allowed us to see substructure in the atom. And right now, the sort of leading theory of a fundamental theory of physics is string theory. And the only reason that theory is not accepted right now is because we can't measure that deep into it. So, it's a technological hurdle to test ideas of that theory.
And so, this brings to the forefront this idea that the horizon of what we call fundamental actually evolves with our technology, because it evolves with the way that we can actually measure and perceive the world. LIGO is another example. Gravitational waves have been permeating this planet for billions of years, but it took, you know, the invention of Einstein's general theory of relativity and then LIGO to detect gravitational waves. All right, so we build all of these technologies to perceive the world in new ways. And this led me to think that the fundamental things are not at the horizon. They're all the constructive complexity in the horizon, because the way that we actually expand our ways of knowing, if we think about ourselves in this causal bubble, is by building new apparatus to explore the boundaries of our bubble. And so, if we're doing that, we need to be large enough as a causal structure to see the things smaller than us. And I think this is the transition we're undergoing now, that technology is animating more things as alive, that what we're doing now is going to fold back and allow us to understand what life is.
Because we have to have a large enough horizon in order to see that physics, right? So, you can't understand patterns unless you've observed enough history. All right? So, we don't even know the patterns in our own history and culture because human history hasn't been around long enough for us to actually see regularity there.
So, okay. So, there's a question to come back to in a different way, of how you define the future in relation to this. But what I hear you say is that technology is itself a form of life, but that there needs to be this expansion of this horizon so that that particular form of life can in essence recursively look back at itself around it. So, the question I wanted to ask in this is, what would that technology be like? What might it be? And maybe you already answered part of this. What would be the preconditions to its appearance?
So, I think this is why I was talking about theories as existential technologies, because I think what good theories do is they allow you to see the world in a new way, and that's actually the scaffold for embedding the measurement. So, I see as one of the critical technologies the conceptual development of assembly theory, which is, you know, a multi-collaborative effort. But also I think some of them are going to be the actual instruments in the lab, and artificial intelligence to some extent as a tool for measurement.
Well, let me ask you then about the artificial intelligence, because it seems, you know, this is an area where there's a lot of attention already paid to a kind of co-evolutionary dynamic of this as well. And there's in many ways a lot of focus on the role of AI as a kind of collaborator within science, and you're sort of suggesting that...
...about them as more like a microscope.
Uh-huh.
So, if you're looking at structures that are so large, you need to be able to process the data and call it... like microscopes allow us to see tiny structures and telescopes allow us to see distant structures. Artificial intelligence algorithms allow us to see large patterns, like the periphery of very large causal structures, but just the surface layer. And so, I think the idea of having a theory that allows us to understand the historical contingency that gets to these very large complex systems is the other piece of it. So, assembly theory to me is the underlying architecture of how you get to this data layer that we exist in now, and AI is the technology of seeing that data layer.
Uh-huh. Yeah, I like that. What is the role of mind in assembly theory?
Yeah, this is something I've thought a lot about, and Andy and I have talked a lot about, and also with our groups. I think minds are very deep in assembly space. So, I think the examples I like to give are things like perfect circles.
Hm.
Perfect circles are objects that can't exist as physical structures in the world, right? Because you can't have infinite precision to make them perfect, but they're an idea that exists in our minds. So, I think I talked about this idea of the virtualization of reality. I think there's a lot of the space that exists in this kind of virtual causal structure that actually can't be physical objects on their own, but they can exist in that virtual causal time as a space. Sorry, not physical coordinate space, but like, you know, this combinatorial space. And so, I think what's interesting about minds is, because of this very large depth in time and the ability to reach into those very abstract layers, there's many more things that we can manipulate about our environment and much more creativity we can induce in the world, because we have access to these internally very large causal spaces. So, to me, our ability to abstract is just evidence of how deep in time we are.
And those sort of abstractions, though, they're kind of idealizations. You give the example of the circle, but you would differentiate that, I believe. I mean, perhaps mathematics would count as one of these abstractions. You would differentiate this from, you know, Platonism.
Yes.
Or, you know, the followers of Neo-Plato.
Right. So, yeah, I'm not a Platonist. I have like an allergic reaction to Platonism. I don't know what it is. Like, I'm literally allergic to it. Intellectually. The reason being, I think it's much more constructive, and this is probably the materialist in me, and I'm not sure if it's the Madonna or Penrose version of the materialist, because both are juxtaposed in my mind. To me, reality is great, and there's no reason for thinking that anything exists more than what exists here. And I think if we fold all of that in, and we really think about the things in our mind and these abstract ideas we have as physical features of our reality and not in some imagined reality out there, it tells us much more about what's actually happening.
And so the way that I think about abstractions is I want to know what the abstractions themselves are as physical architectures.
Right.
And so mathematics to me is particularly interesting, because mathematics is supposed to be a universal language and it corresponds to a lot of objects in the world, but really what it is, is our minds' ability to actually causally influence the future because of the kind of patterns that exist in our own architecture. And so I see math, and science actually generally, not as predictive but as constructive, as I mentioned in the talk, that these things that we do are actually part of the mechanism of how our planet is generating possibilities and moving into the future. And I think that's actually much more helpful, because it gives us a lot of agency and control over things, and a lot more understanding about us fundamentally, than assuming there's some reality out there that we don't have access to, yet that reality is somehow supervening on us and controlling it.
And this is also where it corresponds with your insistence about the materiality of information.
Yes.
Right? This is a real thing.
I mean, if it kicks, it's real. I mean, I don't think there's anybody who would doubt that information is influencing their life. I just think there's no sense to think that these things are not physical.
Yeah.
This goes a little bit to the relationship between assembly theory and philosophy of science once again. What are the falsifiable predictions that assembly theory can make?
So, one of them is that you could have arbitrarily high complexity objects happen outside of an evolutionary process. And so far, we haven't observed that. So, we don't observe, you know, cell phones spontaneously fluctuating into existence on Mars, for example. They exist on our planet. That would be counter to assembly theory. I also think, you know, one of the tests that Lee and I have been talking about is the idea that there's a minimal construction time for any object. And so, you know, Lee's a brilliant experimentalist. I think if anybody could figure out how to do it, he would, but it's a tough problem. But if you could devise an experiment to actually demonstrate that something can't fluctuate into existence and it requires some history. So, what would that experiment be?
You know, we get asked a lot about experimental tests, and I think there are many that we will devise, just like gravity has many tests, but there's also this idea of explanations, right? And so, gravity is a language that we now use to describe this regularity of why we're stuck to our chairs, but it's also something our ancestors observed and their ancestors observed. And so, I think there are deep things we know about the world that we don't have languages for, and I think assembly theory is a language for this deep regularity. And so, in some sense, I don't know how you test an explanation that seems obvious. Do you see what I mean? It's a hard thing to do. And so, you can test the sort of features of the mathematical theory and if it fits, but I also really believe that this deep layer of what science is doing is building better explanations, and the explanations are what carry us forward. And so, I don't know what to do with that. I think science is much more varied than people traditionally... I really like this idea of epistemological anarchy in science.
Paul Feyerabend. Yeah, yeah, yeah.
Yeah.
Okay, here's an experiment for you. Mass spectrometry may help measure assembly index for molecules, but how do you measure more complex or abstract forms of, quote unquote, life, like a technology or a culture?
So, the challenge we have with assembly theory right now is we think the theory is quite general, but developing it for new substrates to test the principles is quite hard. So, we've done it for two in a rigorous way so far. One is molecules, and the second one is crystal materials like minerals and silicon chips. And so, there you can tell the difference between a natural milk
mineral and a technologically engineered mineral, which we call silicon in a very precise way in assembly theory. But the challenge we have is this idea of assembly index has to be embedded in a measurement scheme. So, you have to have some way of actually deconstructing what you think the causation in the object is, not your model of the causation, but how the universe actually generates that structure in terms of what causal constraints, what laws of physics are at play. And you have to be able to look at that as a substrate-specific property.
So, once you can build the assembly space for the material, then you can apply assembly theory. But the assembly space is very non-trivial, and this is where assembly theory I think fundamentally differs from computation, because computation is a... like we can build computational models for everything, but there's one assembly space for a particular selective mechanism.
And so, the universe might be a self-constructing system and all of these spaces have different sets of laws and constraints in them and different kinds of contingency, and assembly spaces can be nested in other assembly spaces. And so, I think what we need to do is to demonstrate it in enough materials to show that life is a general phenomena. That's what I'm interested in actually.
So, I think the question is also about things that you would call real and physical like a culture. Yep. But aren't necessarily material in the same sort of way. And so, the question is like presumably some cultures are more like some languages are more complex than other languages.
But the question becomes hard, right? So, which substrate do you use for constructing the assembly space of language? Do you use vocalization? Do you use written word? Do you use what's actually imprinted in computers? And so, without actually understanding the architecture of how these things are embedded in the substrate, like we don't know what they are. So, you know, I think like language for example, like people think we understand language. Language is a set of patterns that some entities on this planet iterate, like utter, to other entities on the planet, right? So, if we're looking at that as an objective thing outside of ourselves and asking what is that as a physical causal structure, it's a highly non-trivial problem. Yeah. And so, I think it's possible. But I don't want to give you easy answers to it because I think it's a really hard challenge.
Mhm. Yeah, fair enough. If there's other substrates on which life can form, are there other substrates that can hold minds? Mhm. And the other question, which is probably related to this, which is phrased in an interesting way: When is AI life?
Okay, so both good. When is an interesting way of putting that. So I think any substrate that has enough combinatorial richness, kind of an open-ended, maybe an uncomputable one. So I talked about the fact that all possible molecules are actually not computable. Like we don't have enough resources in the entire planet to generate all of chemical space in a model. There are some... So those are the ones that I think are interesting cuz they're capable of open-ended complexity. I think that's actually the one that we're interested in with respect to life.
Now, one place that we've applied assembly theory, actually there's one other substrate I didn't mention, which is an atmosphere. So molecules, minerals, and atmospheres I think we're pretty confident on. Atmospheres because we're interested in life detection on exoplanets. And atmospheres are kind of interesting because they're not combinatorially rich. There's about 16,000 volatile molecules that probably could be in an atmosphere. So it's a really small space. And so I think if you look at the transition to life in an exoplanet atmosphere, the sort of selection constraints you see are very weak. So I think we can pick up signatures of it, but it's not this very sharp transition. So I think living worlds in terms of the imprint of selection in their atmosphere are on kind of a continuum with non-living worlds. But I think there's still a transition we can pick up.
But substrates that have very deep combinatorially rich structure can enter an open-ended cascade. Those are the things that we would really call living. Again, distinction between life and living. And the ones that are very deep structures in that space can have minds.
And I think AI is shallow right now. But I think the integrated structure of our technologies on this planet is getting deeper. But any particular model to me is still a very shallow system. But again, it goes back to this of like are you processing the data at the outer periphery, which I think is what we're doing right now, or are you building structures that are deep? And so, it's very unclear to me about how the predictive algorithms are, are they things that we're interacting with and we're seeing them as being causally deep because we're deep, but the actual embedding of that in the physical architecture might be still quite shallow.
But this is the question. Are they experiencing, are they actually... like what is the reality of a silicon chip processing a large language model? You don't want to ask about the output, right? The output is what we read on a screen. It's a predictive... is it like to be a chip? What is it like to be a large language model embedded in a physical architecture? Mhm. I don't think any of us know what that is.
No, I don't think any of us do know what it is, but presumably on the depth question, you know, if it is essentially something that we have made, we sort of invested our depth in this in a way where we're part of the technosphere, that there would be a lot of evolutionary time embedded in these large models. Like the thing that makes...
You put a lot of time in a small space, but most of that time is just the linguistic component right now, or certain kinds of structure, just to use LLMs as an example because... Yeah, there's a lot of evolution of language in that regard.
So, and I think the thing that's interesting to me is the interaction with words, right? So, words are shifting meaning all the time, and if I asked everybody in... well, we did ask this: what is life? Like there were probably as many different answers for that as there are people in the room, right? So, the fact that there's a predictive correlation between that word and other words doesn't necessarily capture all the conceptual foundations underneath it. And so, I think again it goes back to the causal richness of the history, and how we actually build meaning is mostly about our physical architecture, not these predictive associations of words.
Yes. Which is why we have disinformation issues and all these other things, because you have people with just different causal histories interacting and the words mean totally different things, and they're trying to like cloud each other's perceptions of reality, and it's really easy to do at the surface layer if you're not seeing the structure underneath, or if you're not seeing the experience of an individual, you can assume they have bad intentions when they don't.
Yeah, I mean there's a whole another script about spatial AI or analogy, which has to do with more sort of embeddedness, but that's another talk. We need to wrap it up here, and I just want to first of all say thank you on behalf of the whole audience for such a rich body of ideas to think with, assembly theory as a scaffold for many things to come next, and thanks for your time.
Yeah. Thank you.
Article published