Conversations

Season

1

Episode

51

|

Oct 31, 2024

The Origins of Life with Sara Imari Walker

In this episode of Conversations, we sit down with Dr. Sara Walker, an astrobiologist and theoretical physicist who is reshaping the way we think about life. From her work on the origins of life to her innovative assembly theory, Sara takes us on a journey through the possibilities of discovering alien life and challenges some of the core tenets of biology and physics. This conversation will make you rethink not just how life began, but what life is and does.

0:00/1:34

Conversations

Season

1

Episode

51

|

Oct 31, 2024

The Origins of Life with Sara Imari Walker

In this episode of Conversations, we sit down with Dr. Sara Walker, an astrobiologist and theoretical physicist who is reshaping the way we think about life. From her work on the origins of life to her innovative assembly theory, Sara takes us on a journey through the possibilities of discovering alien life and challenges some of the core tenets of biology and physics. This conversation will make you rethink not just how life began, but what life is and does.

0:00/1:34

Conversations

Season

1

Episode

51

|

10/31/24

The Origins of Life with Sara Imari Walker

In this episode of Conversations, we sit down with Dr. Sara Walker, an astrobiologist and theoretical physicist who is reshaping the way we think about life. From her work on the origins of life to her innovative assembly theory, Sara takes us on a journey through the possibilities of discovering alien life and challenges some of the core tenets of biology and physics. This conversation will make you rethink not just how life began, but what life is and does.

0:00/1:34

About Our Guest

Sara Walker is an astrobiologist and theoretical physicist interested in the origin of life and discovering alien life on other worlds. She is Deputy Director of the Beyond Center for Fundamental Concepts in Science and a Professor in the School of Earth and Space Exploration at Arizona State University. She is also a Fellow of the Berggruen Institute and a member of the External Faculty at the Santa Fe Institute. She is the recipient of the Stanley Miller Early Career award for her research on the origin of life, and her research team at ASU is internationally regarded as among the leading labs aiming to build fundamental theory for understanding what life is.

About The Show

The Nocturnists is an award-winning medical storytelling podcast, hosted by physician Emily Silverman. We feature personal stories from frontline clinicians, conversations with healthcare-related authors, and art-makers. Our mission is to humanize healthcare and foster joy, wonder, and curiosity among clinicians and patients alike.

resources

Credits

About Our Guest

Sara Walker is an astrobiologist and theoretical physicist interested in the origin of life and discovering alien life on other worlds. She is Deputy Director of the Beyond Center for Fundamental Concepts in Science and a Professor in the School of Earth and Space Exploration at Arizona State University. She is also a Fellow of the Berggruen Institute and a member of the External Faculty at the Santa Fe Institute. She is the recipient of the Stanley Miller Early Career award for her research on the origin of life, and her research team at ASU is internationally regarded as among the leading labs aiming to build fundamental theory for understanding what life is.

About The Show

The Nocturnists is an award-winning medical storytelling podcast, hosted by physician Emily Silverman. We feature personal stories from frontline clinicians, conversations with healthcare-related authors, and art-makers. Our mission is to humanize healthcare and foster joy, wonder, and curiosity among clinicians and patients alike.

resources

Credits

About Our Guest

Sara Walker is an astrobiologist and theoretical physicist interested in the origin of life and discovering alien life on other worlds. She is Deputy Director of the Beyond Center for Fundamental Concepts in Science and a Professor in the School of Earth and Space Exploration at Arizona State University. She is also a Fellow of the Berggruen Institute and a member of the External Faculty at the Santa Fe Institute. She is the recipient of the Stanley Miller Early Career award for her research on the origin of life, and her research team at ASU is internationally regarded as among the leading labs aiming to build fundamental theory for understanding what life is.

About The Show

The Nocturnists is an award-winning medical storytelling podcast, hosted by physician Emily Silverman. We feature personal stories from frontline clinicians, conversations with healthcare-related authors, and art-makers. Our mission is to humanize healthcare and foster joy, wonder, and curiosity among clinicians and patients alike.

resources

Credits

The Nocturnists is made possible by the California Medical Association, and donations from people like you!

Transcript

Note: The Nocturnists is created primarily as a listening experience. The audio contains emotion, emphasis, and soundscapes that are not easily transcribed. We encourage you to listen to the episode if at all possible. Our transcripts are produced using both speech recognition software and human copy editors, and may not be 100% accurate. Thank you for consulting the audio before quoting in print.


Emily Silverman  

You're listening to The Nocturnists: Conversations. I'm Emily Silverman. Today we're wrestling with one of the oldest questions in science: what exactly is life? For years, we've struggled to answer this question. We've said maybe living things are defined by reproduction or compartments or metabolism or being a self sustaining system, but it seems like there's always an exception to the rule. We either rule something in that clearly isn't life, like a car engine that metabolizes gasoline, or we rule something out that clearly is alive, like a mule, which is sterile and can't reproduce. It seems, no matter what box we try to draw around life, it breaks free from our definitions. But our guest today, astrobiologist and theoretical physicist Sara Walker wonders if maybe it's less important to define what life is and more important to define what life does. This is at the heart of her work on assembly theory, which we talk about in this episode. Sara is the Deputy Director of the Beyond Center for Fundamental Concepts in Science and a professor in the School of Earth and Space Exploration at Arizona State University. She's a fellow at the Berggruen Institute and member of the external faculty at the Santa Fe Institute. Her lab is internationally regarded as one of the leading labs building fundamental theories about life, and she recently wrote a book called "Life As No One Knows It: The Physics of Life's Emergence. I stumbled upon Sara on YouTube a few months ago and could not stop watching her talk. She's one of those trans-disciplinary polymaths who can speak about everything from physics to chemistry to biology to linguistics to artificial intelligence, dissolving the lines that humans have placed on reality and always seeking to unite, reduce, and simplify. I was very honored to have the opportunity to speak with her today about why traditional physics fails to describe life, the tenets of her assembly theory, the way she thinks about that magical phase change from a random chemical soup to life and its markers, her vision of life as a force that propagates through time as an exploding phylogenetic tree, her belief that human beings are actually some of the largest objects in the universe, containing 4 billion years of history and information all curled up inside and so much more. It's a dense conversation, but a fascinating one. So grab your coffee, buckle your seat belt, and enjoy this episode, which begins with Sara reading a short excerpt from her book, "Life As No One Knows It."

Sara Imari Walker  

The paradox of defining life. Look around you right now. I'll bet you can confidently catalog what is living and what is not. I could quiz a four year old and they would probably corroborate your classification. In fact, small children seem rather good at telling the difference between living things and inanimate objects with little explicit direction. Once a child is taught that plants are alive, they can readily tell those apart from non living things and act accordingly. For example, not stomping on the flower bed, but perhaps jumping in the mud, instead. Kids can also misclassify life by assuming inanimate objects like a favorite stuffed animal or life. Are they wrong in doing this, or are we wrong in correcting them?

Emily Silverman  

I am sitting here with Sara Walker. Sara, thank you so much for being here.

Sara Imari Walker  

Thanks for having me. I'm really excited about our conversation.

Emily Silverman  

I discovered you on YouTube, actually, before I knew that you had a book coming out and just listening to you talk, my face kind of melted off, to be honest, from how brilliant you are, not just the content of your work, which changed how I look at the world, but also the way that you're able to communicate that content. It's obviously very high level, complex stuff, but you just have such an ease as a science communicator that I was really excited to bring you on and have you share some of your work with The Nocturnists audience. So let's start with your background. You are a theoretical physicist, and you study the origin of life, which some people probably hear that and associate more with biology than with physics. So how does a physicist come to study the origin of life? 

Sara Imari Walker  

My first response to that was it was accidental, but I don't think it really was accidental. I started my university training and community college, and I was really interested in science, but I didn't know what I wanted to do, and my parents hadn't gone to college or anything, so I just took all the science classes, but I absolutely fell in love with physics. I just thought it was a phenomenal subject, mostly because I love this idea that we can come to deep, fundamental understanding of the nature of reality by thinking about it and by coupling our own thoughts with experiments and what we can test. And I really thought I was going to be a particle physicist or a cosmologist, but when I got to graduate school, my PhD advisor was studying the origin of life or getting interested in it, he was a cosmologist also. And I kind of reluctantly at first, actually started working on the problem, because I was like, this is not what physicists do, exactly what you're saying. But as I thought about that problem more and read more and read more and more of the literature, I realized that it was an incredibly open problem, and there was a lot of room for creativity. And I really felt that the training that theoretical physics provides, and in particular, like what I was really passionate about, which is deep explanations and discovering something new, like origin of life was wide open territory, in part because it doesn't really fit in any of our traditional disciplines, and we don't really have a right way of asking that question. So it seemed like there was a lot of room for the kind of science I wanted to do.

Emily Silverman  

Yeah, and I've heard you say this in interviews before that when people traditionally think of physics, they think of math equations. So probably a lot of people listening to this have taken basic college level physics. I think about Maxwell's equations, or I think about Newton's equations. And basically the way it's set up is you have this fixed law, and then you plug in a set of initial conditions, and it tells you what's going to happen to the apple as it falls from the tree to the ground, for example. But you've said that laws like that don't really work to explain life, and that, in fact, what we have to do is invent a new physics for that. So tell us about that. 

Sara Imari Walker  

Yeah, I have very fond memories of my physics classes, but I have some harrowing ones too. I think we all go through that. It's really interesting because the way that physics works. We're taught about disciplines, and we're taught like, if you're interested in life, go get a biology degree. If you're interested in stars and planets or mechanics or things, go get a physics degree. If you're interested in molecules, go get a chemistry degree. But obviously, Nature doesn't have any of those distinctions. And why I became a physicist was not necessarily because I was interested in a particular question. It was because I really was thinking about this idea that the human mind was so incredibly creative to come up with really abstract and universal understanding of the world. So to me, that's what physics is. And so this example that you're bringing up is that once you have the initial condition and law of motion, you've described the system for all time, and that comes from the era of Galileo and Newton, because they were trying to describe motion and gravitational physics in particular. And that kind of dynamic works very well for those kind of systems, because they have very regular behavior over time. But life does something very different. And I really love this very famous quote from Charles Darwin about "endless forms most beautiful keep evolving and continue to evolve." And he juxtaposes it to the fixed law of gravitation. So apparently he was a huge Newton fan, and I think he was hoping there would be like laws for evolution, like there were for gravity and motion. But he also recognized this very stark contrast, that biological forms seem to be changing all the time, and they seem to change in a way that is historical and path dependent, and this is not what we see in typical physical systems. They usually have the same kinds of rules and the same kind of behavior, no matter when you observe them. And life forms seem to have new rules and new behaviors emerging all the time. And biological evolution over geological timescales is probably the grandest stage for observing that. But even if you observe a cell, you know, it's a complex machine with all kinds of dynamical, non linear processes happening. And so it's been a real challenge for the standard conception of physics in terms of this initial condition and fixed law to describe it. And what I spent most of my career working on is thinking about, how do we actually understand a physics that has these kind of path dependent properties that we see in living systems.

Emily Silverman  

And one of the theories that you're working on is called assembly theory. And you talk in the book about how atoms are countable, we can see them on the periodic table, but once you level up to molecules, the space of possible combinations is huge, and the universe doesn't have the time or the resources to build everything that can theoretically be built. So you and your collaborator, Lee Cronin have developed this theory called assembly theory, which explains why certain things get to exist. So tell us about assembly theory, and particularly what you just said earlier about this idea of path dependence or historical contingency.

Sara Imari Walker  

The idea of path dependence is a nice segue into the example that you've given. Some of these approaches and ideas that I'm talking about seem very deep and abstract and theoretical, but one of the places that we're finding actual utility for them is in drug design, because if you imagine the combinatorial space of molecules that might actually have a biological function, the space of molecules is huge, and then finding that subset that might interact with our biochemistry in just the right way to get the right kind of health outcome is really challenging. There's a huge industry of people trying to tackle that problem. But the way that we think about this combinatorial space, from the perspective of thinking about origins of life, is there's a huge amount of possible molecules that could exist, and the way that you actually start to produce them is you have molecules that have been selected in a particular environment, and they scaffold into other structures. And so assembly theory captures this in a really rigorous way. And I'll explain a little bit about what our observables are and how we think about the theory, which is really related to the idea of, how do you detect life in chemistry, like if you look at a molecule, how do you know it's a product of life or not? But then I'll circle back to the question about the path dependency, and how this relates to a very non-Newtonian kind of physics, or at least not in the standard paradigm of physics. So the basics of the theory are, we have a conjecture that life is the only mechanism the universe has for generating complex objects. And complex objects are things like complex molecules, like DNA is obviously an example, or ATP even so, things that we find in a cell we don't find spontaneously produced in the environment of a planet like no one's observed these outside of biology. And you could take for granted that the universe could just produce those objects outside of life, but actually a key conjecture of assembly theory is that that's not possible. And that's one of the things that we actually aiming to formalize to test that you only get complexity as part of a living process. And the way that we formalize that, it's almost like thinking about Lego blocks and building up a LEGO Universe. Or if you play Minecraft, you have all the pieces you can put together to make all these complex structures. But we start with thinking about bonds, and how bonds are actually the fundamental unit of building molecules. So if you prefer the Lego analogy, you can think about your Lego blocks, and you can stick them together, and then you can reuse structures you've already built. And let's say that we have a target structure, like maybe we want to make the molecule ATP, or maybe want to make Hogwarts Lego Castle. A couple examples of complex objects in your mind. So you can take this idea of trying to build an object by putting pieces together and then taking those pieces and reusing them and get up to the final structure. And what we try to look for is actually the shortest pathway to do that, which we call the assembly index. So there's sort of a minimal constructed complexity for making the object, or a minimal set of constraints that's necessary minimal history to produce that particular structure. And then we also look for that structure in high abundance. So the idea being on our planet, there's not one Lego Hogwarts Castle, there's millions of boxes, potentially of them, and patterns on the planet so that they're repeatable structures. That's because they've actually been selected by the environment, by humans, by kids that are really big fans of Harry Potter to be a really highly abundant object on our planet. Whereas, if you took the equivalent amount of Legos, I think there's like a couple 1000 pieces in there, all the random structures that you can make out of that, that's a huge space, but most of those structures will never actually be built ever in the history of the universe. It's actually probably the number of structures is so large the universe doesn't have enough time, even if you were putting Lego blocks together every second, to exhaust that space. So we have the assembly index and the copy number as the two key variables that we're looking at, and we view those together as a signature of constructed complexity. Basically what life does. Life is the mechanism the universe has for making complex objects. And it's important that we have those two things, because we can actually go in the lab and observe them and measure them. So the way we do it with the assembly index's shortest path, is using mass spectrometry. And it turns out that with a mass spec, you can probe that feature of a molecule. And if you're doing mass spec, you already have to have thousands of copies of a molecule even to recognize that it's there and measure it just with standard laboratory equipment. And so what we were able to do was actually go in the lab, and this was an experiment done in Lee Cronin's lab. I think this was just such a cool experiment. What they did was actually take living and non living samples, and they figured out how to measure assembly in those samples, and they were able to show that above about an assembly index of 15. So thinking 15 steps where you reuse parts and build an object in chemical space. So you're using bonds to form these structures. About above 15, you only see molecules that were produced from the Living samples, and there were even blinded samples from NASA that were supposed to be confounding ones. It was like Murchison meteorite, which is like one of the most complex examples of chemistry in our solar system. And it didn't pass the test of producing high assembly molecules, like if you just have random structure making molecules, you can make all of the very simple molecules pretty easily. But what we see in prebiotic systems,when you don't have selection and you don't have evolution, or you don't have constraints, you just get tar, which is an undifferentiated goop of all of the kind of simple molecules. You don't get high abundance of very refined molecular structures like what we see in biochemistry. And so we think that unless you have selection and evolution, you can't cross a complexity threshold, which experimentally seems to be about 15 steps. That might not be a universal number. It's experimentally verified, but we have a lot of work to do, but we conjecture that there's always a threshold, and above that, you have to have a path dependent trajectory to make those objects, because they're just so complex, and the number of errors that you could have making the wrong part, and bifurcating off into all the other possibilities that could be built is so large that you would never expect that structure to exist unless something selected for it to exist, and it was built by an Information Processing System, or something that we might call alive. So that's sort of the key features of assembly theory, and why it's very different than that standard Newtonian paradigm is the universe of possibilities gets larger with every single step that you take trying to build a new object, because every time you could have put some other things together differently, and so you have to trace this path dependent trajectory out. And in a Newtonian world, the universe never gets bigger as a function of the complexity time always has the same size. But what we see in assembly theory is this notion of this combinatorial space that living things are moving through. And as it gets more complex, the space gets larger and larger, and that actually informs a lot of the foundational physics that we're thinking about and the philosophy that I'm most interested in.

Emily Silverman  

So the idea is that if you look at a molecule like Taxol, for example. I know as an example that you use a lot Taxol is, I believe, a molecule that was originally found in plants, yes, or a molecule like ATP, or even a giant molecule like a ribosome. It's like, once you get ATP and once you get the ribosome, all sorts of things become possible. How do you think about these stop gaps? I've heard you compare the ribosome kind of like a transistor, which is a piece of technology that, once that was invented, it was like, Okay, now we have the transistor. We can build all these other machines. Or once the light bulb was invented, it's like, Okay, now we have the light bulb. We can make all of these different types of lights. Do you think about those moments as special, where you create something that blooms out into more possibilities?

Sara Imari Walker  

I think that's just the way that our universe works, and it's the only way it can make complex objects. When we're thinking about life evolution, at least, we all have a picture in our mind of these branching phylogenetic trees that are constructed from genomic information, right? And before that, people were trying to do it by morphological similarity. So we've had this kind of tree structure in the way that we've been visualizing the nature of life for at least 150 years, since the time of Darwin. Well more than that now. And some people will be like, Well, what do you need assembly theory for we already have good enough evolutionary biology, but the point of assembly theory is actually try to look at selection and evolution before you have things like genomes, because we're trying to explain the origin of those things. So we need deeper evolutionary principles than what biological architectures provide us, because we're trying to explain the origin of things like the cell, not use the cell as a scaffold for talking about further complexity. And so what we see looking at assembly theory is, if you think about the universe as this giant space of possibilities that could exist, selection has to mediate these pathways to build up the next sort of scaffold. So I do like the example of transistors and the point that I would make there, because technology is very perceptible to us, is that most of our modern technologies are built on the transistor, because transistors have been a very prevalent component of a lot of technology. So now we have factories that can make them, and they make devices compatible with each other. And so it's selected out a whole bunch of other options that might have been equivalent or even better than transistors in terms of efficiency and all other kinds of technology, but we're kind of locked into using transistors now because they're so prevalent and we're on its trajectory now where transistors are going to be part of our future technology. And a similar transition happened very early in the evolution of life on Earth, with the invention of the ribosome. Evolution had to construct the ribosome. Once the ribosome and the rest of the translation machinery was constructed by evolution, it became solidified as a structure on our planet that could elaborate further complexity and kind of ruled out the other possibilities that maybe other kinds of cellular architectures or things could have evolved at the same period. But this became the one that was successful, and now we find it in all life on Earth. So we have a single branching point for life. But I think what assembly theory is doing is making that branching structure go deeper. And so this is how we're thinking about the nature of the actual structure of physical objects, and that they're all related to each other through this historical contingency of this reuse of parts and the causation. Propagating through all these objects.

Emily Silverman  

You wrote an article in Eon magazine called "Life is Not Alive," where you say that things like cats and dogs are alive, and things like sofas and rockets are life, and the reason that sofas and rockets are life is you don't just randomly find a sofa in the universe that you need life around to invent and build a sofa. And like you said, there's a high copy number. We see a sofa here and a sofa there, and there's green sofas and black sofas and velvet sofas. So I never really thought of it that way. I always thought of it as animate and inanimate, but you're bringing in this new category, which is inanimate matter that is life, because life needed to be around to create it. So talk a bit more about those distinctions and why they're so important. 

Sara Imari Walker  

I'm a professional astrobiologist. So the two problems that drive my career really the origin of life. How is the transition from non living to living matter, and then the other one being life detection on other worlds. And if you're thinking about detecting life, you don't necessarily care if you're finding an extant organism, something that's alive. Now in the Mars community, a lot of the focus is on finding fossils, or fossil evidence of life. But I think most of us would be convinced if we found a cell phone on Mars, it would be evidence that life was there. Like, we go find a robot, it's probably a human robot that was put there, not that geochemistry of Mars spontaneously fluctuated curiosity on the surface right, like it wouldn't be explanatory. It's probably the case that some intelligent planet nearby actually built the technology to put it there. So we recognize in our narratives already that there are these certain structures that must be constructed by humans or some human-like physical system, something with comparable intelligence, or are necessarily the product of evolution. That's kind of the idea of a bio signature. And so when we were thinking about this fundamental story about the physics of life, and thinking about the combinatorial universe of chemistry and all the structures that could be built out of it. Chemistry, at least, is iterable, right? So I like thinking about chemistry because you can talk about the atoms in the periodic table and the combinations, and maybe estimate the size of a chemical space using chnops, carbon, hydrogen, oxygen, phosphorus, nitrogen, sulfur, right? Like the standard biological elements, we can kind of imagine sticking them together by bonding rules and estimating how many structures there are. We don't know how to estimate how many kinds of life forms are out there, right? Like how many cell like structures, if you think about the Cambrian explosion and how many body plans were made, it could have been exponentially more than that was probably just what the resources and physical environment imposed. But most of those aren't around anymore. And if we think about all possible technologies, that's an even larger space. So all of these combinatorial spaces are quite large. And in assembly theory, we always think that there's this boundary that needs to be crossed, that only evolution can do the process we call evolution, but thinking about it much more deeply in terms of this historical contingency and the construction process, not as a passive filter. It's a very active mechanism of selection, selecting the next thing that the universe can construct. When you're thinking about that kind of structure, it's very clear that you have to have kind of an abrupt transition point, and when you cross that transition point, those objects can only exist in the presence of an evolutionary process or selection, because the space of possibilities they live in is so large that the universe could not possibly do an exhaustive search to discover them. It would have to do a directed, constructed process. And so we kind of differentiate all of the objects above that that are impossible to form randomly and require information for their construction to be life, anything in that space. And then alive to me is the actual active mechanism of building a trajectory into that space. So something like us is alive because we're creative in the universe, we're generating new possibilities. But things like water bottles, like I have on my desk that are artifacts of life, they're still life. Like bottles don't spontaneously come into the universe outside of things like us to generate them. So we have a causal pathway like this thing has to be built before this can be built, before this can be built. So objects actually necessitate their existence based on other structures that have been selected in the evolution of a biosphere, but it's the active process of building that's alive. So in some sense, it's like, is it the process or the thing? And we're really trying to merge both of them into the same story, but this is where you can have a more holistic view about the nature of the physics of life, really, just thinking about life is everything that's complex, too complex for the universe to generate spontaneously and requires some kind of construction and informational constraints. And then it's much more perplexing and interesting to find things like us that are alive. If we want to go look for aliens, probably we want to find alive ones and not dead ones. We had this joke in that essay, actually, about Schrodinger's cat. And I love this because, you know, it's a very physics example about is the cat alive or dead? But in assembly theory, in the way we think about life, the cat can be dead but still be life. It's not alive anymore, right? But it's, you know, a dead cat is a product of evolution in the same sense that a living cat is. They don't spontaneously form in the universe either. So you're kind of talking about two different levels of description of the same physics, but that's kind of the idea behind it.

Emily Silverman  

Yeah, and you joked about this in the talk that you gave at Mindfest, which I watched and which we'll link in the show notes, where you talk about this thought experiment of the Boltzmann brain. And so the idea for the audience is that human brains exist in the universe because we all have a human brain, therefore it is possible for a human brain to exist in the universe. Therefore, why shouldn't a human brain spontaneously fluctuate into existence? Because physics theoretically allows it. But then you say, and I quote, "People really intrinsically think the universe can generate anything, anywhere, for free. Where is all that information? Please tell me." And so where is the information? Like, where? Where is the locus of the information?

Sara Imari Walker  

A lot of my work is trying to build what seem like counterintuitive notions, but they're actually very intuitive, if you're willing to sort of break your training and how you've been taught to think. Why I was joking about that is because I think our standard education and the way that we talk about the world has, right now, at this moment in history, really trained us to think that any object can fluctuate into existence. It's very standard in physics, and I learned it in many physics classes that there was just a low probability for thermodynamic fluctuations or quantum fluctuations, out of the vacuum to produce objects of arbitrary complexity, just because it's written in the laws of physics as we understand them. Yet we have no observational evidence for that. It drives me mad, because I think about it this way now. But like, obviously I didn't think about it this way most of my career, like when I was a student, stuff, I never would have thought about this way. But it's like, seems so obvious to me, is like, in order to have any object fluctuate into existence at any point in space and time be a possibility, it means, like, everywhere has to have the information for every object, which means you have to have an infinite density of information in our universe at every point. And that's just totally unphysical. So we have, like, these really unphysical things just baked into the way that we talk about the world, and because they seem so obvious to everyone that that's the way the world works, nobody takes a step back and questions these basic assumptions. And I think this is actually really the art of theoretical physics. Like most of the major advances in theoretical physics have been someone that's willing to sit down and question the most basic assumption that everyone assumes is true. It's like Einstein with like, the constancy of the speed of light, right? That's like, a very obvious example, because even with the Michelson Morley experiments that validated experimentally with high precision that the speed of light was constant independent of the motion of the Earth, people still didn't want to accept it. They assumed the experiment was wrong. And Einstein was like, I'm just gonna assume it's right and just run with it, and we get relativity right. So it's kind of interesting, like there's just these things that we just assume. So that's one of them, for me, that I find kind of mind boggling now, is that the idea of Boltzmann brains is taken seriously in some of the physics community, and it's to the point like in my history as a cosmologist, like some people reason their cosmological models on the idea of trying to rule out Boltzmann brains being more prevalent than we are, because once you assume a brain can fluctuate into existence, it's very likely that any observer is a Boltzmann brain over cosmological timescales, instead of being something like us that was evolved on a planet over billions of years, and assembly theory says something very firm like that is not possible. The only way to get to something like us is through the process of evolution and construction, and to get something like us means we have to come with a whole bunch of other structures like us. So we emerge in populations and things. And it's like, very consistent with what we actually observe the world to be like.

Emily Silverman  

Right before, you need a human brain, you need a human body, you need a ribosome. You need it's like the Carl Sagan quote, where he says, "How do you bake an apple pie? First you have to invent the universe." 

Sara Imari Walker  

That's right, and then you have to go through all those evolutionary steps to get one right. Yeah, that's right. It's not just in the initial condition.

Emily Silverman  

I'm wondering if assembly theory could apply to abstract objects as well. So for example, this thought experiment of could a random letter generator left to operate for an infinite amount of time write the book of Hamlet for example. My husband and I actually have this ongoing fight about this question. 

Sara Imari Walker  

Oh, really?

Emily Silverman  

Yeah, you know, and I always felt like, yeah, maybe a random letter generator could produce a few words or some word fragments, but it's not going to write Hamlet. For it to write Hamlet, you need something else. You need consciousness or intent or something like that. And he would always argue like, no, no, you don't understand what infinity means. Like, if you let it run for infinity, by definition, it'll write Hamlet. And before you respond to that, I just wanted to share one more example that's tied to language, which is this short story The Library of Babel by an Argentinian writer, Jose Luis Borges. And the story describes a universe that consists of an infinitely large library filled with books, and the books are filled with random letters, and the librarians are driven to suicidal despair because they know that the library must contain every book that has ever been written, and yet they're surrounded by gibberish. And if they ever stumble upon a book that has a sentence in it, they kneel in worship. And so I was wondering if something like the assembly index threshold could apply to language too. Could you say that after 15 letters, it stops being a coincidence, and there's somebody writing?

Sara Imari Walker  

Yeah, so I don't think the value of 15 will be meaningful for language, because people start to assign a very particular value to 15 and like it becomes like the number 42. It's not the answer to life, the universe and everything. It's an empirical validation of this idea of a threshold. We don't know where the threshold is going to be in most systems, and that's actually something that I'm actively working on, is like a general theory of the idea of like a phase transition associated with the origin of life, and why does it happen at a particular complexity value in assembly theory? But to your point, yes, I think the ideas apply to language as much as they do to molecules. And for a long time, I had been writing about the nature of mathematics as a physical system, which is like, kind of a really weird way to think about it. But the reason this came about is because I was trying to think about, like, how could I really point out the discrepancy that we were talking about before, about, like, what physics says and what we need to explain life. And everybody thinks physics as we understand it now is right and, like, the ultimate explanation. So like, how do you build a thought experiment or something that turns the narrative for people to see that there's, like, a totally different way of looking at it? So I started doing these thought experiments on the theoretical physics of theoretical physicists as I was thinking about in my own head, but really thinking about, like, you know, the laws of physics themselves as physical objects that have merged in our biosphere like Newton law of gravitation obviously describes something we think is objective about the world, which is like gravity, but it's also an equation we write down on paper, and that equation has causal power, in the sense that we can teach it to undergrad physics students, and they can roll balls down inclined planes and test its properties, but it also allows us to launch satellites into space and do a myriad of other things that wouldn't be possible without that knowledge and that kind of small information packet. And so to me, that's a physical structure that exists on our planet, and I was really trying to elucidate that. So the reason I'm prefacing this and talking about language is we think about language as being abstract, but it must have some kind of physicality to it, because language actually influences the physical structure of reality. It changes the way we perceive reality, but also through our actions, we actually change the nature of what exists and what gets to happen. I actually even have an example in my book, which is in the middle of the book about trying to prompt people to turn the page is like a physical act, right? Like a word actually activates you to turn a page of a book. Like, how is that possible? That's like a really weird sort of thing.

Emily Silverman  

And in your MindFest talk, you take a moment and you say, "Everybody raise your hand" and then everyone in the room raises their hand, yeah? And so that's a physical act.

Sara Imari Walker  

aYeah, so I think we have evidence all around us. Yeah, that language has some physicality to it. As far as applying assembly theory, it's really non trivial, because, you know, I talked about this idea of assembly index and copy number, and they're not really separable. We can calculate the complexity of this molecule and count the number of copies. It's pretty easy for molecules, but when you move to other systems, like minerals or language that don't necessarily have natural boundaries for language. Is the boundary a word? Is it a meaningful segment of a sentence? Is it a sentence? Is it a paragraph? This is part of the foundation of trying to understand language, and the machine learning is that, like the things that actually become tokens are not necessarily the things that we prescribe meaning to, as far as the predictability of language. But the point there is just that we actually need to start thinking about language as a physical system. And one of the other tenants we have of assembly theory is if you're going to talk about assembly index and copy number, you have to be able to measure them as part of the systems. We have to think about what are interactional units with a language that we interact with. And so I do have work in my lab right now, actually trying to apply assembly theory to language, but applying the principles of the abstract theoretical physics to language is very non trivial, because we're really trying to think about language as a physical system. But I do think to your example about the Library of Babel, that there would be a threshold, and you would be able to say that this thing is definitively constructed by a meaning making system that was reusing elements that had specific associations with other ones, and that the structure of a language would emerge. Another thing that happened with large language models is that people were starting to realize that the structure of almost all human language is identical. If you remove the actual differences about like what words are used for what in different languages, there's this like topology to the structure of human language, about the associations between concepts that we build, that seems to be pretty universal, and I think assembly theory would be quite good at picking up those causal structures underlying language, which are the structure of association. So even when I use language, I don't really think about the meanings of words. I always think about structural associations, which is super important for my line of work, actually, because there are no predefined concepts. So I have to always know that the meanings of the words that I'm using are not the meanings that I intend, because we don't have words for these concepts yet. So I'm kind of playing around with the relations of words like you probably noticed already in this conversation. I'll interchange complexity, assembly, information, evolution, causation, constraint. And to me, they all mean the same thing. I know it's like dizzy,

Emily Silverman  

Yeah, consciousness, meaning making system. Yeah, yeah. 

Sara Imari Walker  

You have to kind of play around with the words. This is part of the thing about effective communication. When you do that, enough people start to build their own associations in their mind, and then they can start to see the picture of what you're talking about. So I think this is really important to communicating deep ideas effectively. Is not to use the same words all the time. Yeah,

Emily Silverman  

I love the idea of a theoretical physics lab collaborating with a team of linguists. I mean, to me, that is just the perfect example of how your work jumps from discipline to discipline. It cannot stay confined inside physics or chemistry or biology, and that's part of what makes your work so exciting to me. I wanted to ask about you mentioned large language models, and you talk a lot about technology, actually, in your book. And there's a couple questions I want to ask you, but maybe first I'll ask you about just how you see technology. You talk about how Earth has developed a biosphere, and that biosphere has now invented a technosphere.

Sara Imari Walker  

I think sometimes about what our biosphere is doing is unfolding structure over time, and so in some sense, and I actually was thinking about this quite a lot this morning, about the simulation argument, where people think reality is literally simulation, and it's so interesting that we can viscerally feel that right now, because we're seeing all of these objects coming into our environment that were once just ideas in human minds. And so pretty much everything in our environment is constructed nowadays by a human mind. I have a microphone sitting in front of me, a board behind me with a whole bunch of scribbles on it that were iterated by a human evolution took 4 billion years to construct something like us, and all of that selection and all of that historical contingency, of all those steps along the way are actually embedded in us as physical structures. So one of the concepts I've taken from Assembly theory, which I think is a direct consequence of how we measure attributes of the theory, is that evolved objects actually have a size and time, and that size and time, the more complex the object, the deeper in time that object is. So I would think of something like myself or you as an object that takes 4 billion years of history to be able to create you, but that's all rolled up recursively through this process of reusing parts like that stack of structure is you. Then we go out in the world and we create things. We're putting all of that history into these structures. So like the world is in some sense becoming more life, maybe not yet alive around us in the technology that we're creating, because it required so much evolution and so much packing in of all that selection and all of that time and structures like us to generate these things in the first place. So technology, to me, is just a continuation of the process of life on Earth. I think the more interesting question there, and the more interesting transition is, when do we say that our technology itself is alive too, in the sense that technology becomes a creative force in the universe, in the way that we are, it has causal power of its own, if you want to talk about it that way. And I don't know that that we're there yet with any individual device or technology, including large language models. I don't really see them as being incredibly creative beyond what has been put in, as far as what human language is already capable of. But I do think if you look at like the global ecosystem between technology and humans and like what we're creating, that that whole system is in some sense living like. Social systems are living, and they exist at different scales. So one of the other things that I think about quite a lot is how life is a process that recurs across scales, and what's the role of technology in continuing that process, and thinking about technology as a social scale or a cultural phenomenon. So I, like a lot of the discussion now, thinking about LLMs (large language models) as cultural technologies and these kind of things.

Emily Silverman  

Yeah, because there is this interesting thing that happens with Life where I'm a human being. I'm sitting here, I have a ton of cells in my body. I have a dendritic cell in my immune system, and that cell, you could argue, is alive, because it's constantly sampling its environment. It's making decisions about whether to travel to the lymph node to present the antigen and things like that. But that dendritic cell probably is not aware of Emily, like it is a living thing inside, a living thing that has a consciousness that maybe that one cell is not aware of. So do you think it's possible we could evolve in a direction where you and I are like the dendritic cells, and then God knows what, the new GPT, or whatever is the Emily. And maybe we're not even aware that it exists, and we're just kind of living our lives, swimming around. But there's like a new level.

Sara Imari Walker  

I think we're already kind of there. I mean, I mean, I think that's what human, social and cultural systems are. And a lot of people that work on complex systems really think about them as living, evolving entities. Like culture itself clearly has evolutionary properties, and it clearly controls the behavior of individual actors that are part of that social system. So if you think about like the cellular contract, cells form together to make tissues, and they're not going to apoptose unless there's some traumatic event that happens, or why would you even have programed cell deaths, unless you were a part of a society that needed to have that function. So I think societies already have some of that structure, but I think certainly what I view with modern technologies is they're becoming the informational architecture of these social entities. So something like a large language model cannot come into existence until you have a society that evolves language and then evolves technology to record vast volumes of that language in order to even train these models. And so the large language model is scaffolded on all of this human evolution, but where it scaffold is actually at the societal layer, not the individual. And so I think this is one of the reasons that they're deeply perplexing to us, because when we interact with them, and this is why it was such an epistemic shock for so many people the first time that they ever interacted with one is like you're taking something that has traditionally been part of a social system distributed across many human minds, language and the dynamic of that and its evolution has always been, we're the only substrate, and then you train a model on it, and you put it in a box, and then suddenly you can interact with that as like a dynamic. It's totally crazy, but I loved when that was coming out. A lot of people were pointing to how in ancient Greece, people were trying to not have people learn how to read and write, because they were afraid of the dead talking, and they thought this was a really dangerous technology, because the dead should not speak to the living, so you should not record what you've said as written text. So in the past, we've had ghosts. The Dead could talk through books over hundreds of years, and now it's become a process where now the dead have much more of a live interactive voice, because you can obviously train someone, if they have enough text outputs from their own vocalizations, you can train a digital twin of them. So now it's like the ghost has gotten two dimensional instead of one dimensional. But I think about it a little bit like how the genome had to evolve in early life, in some sense, because if you're thinking about a dynamic chemical system that has no record of its past history. It has to start building record keeping devices. This is where the genome and the translation machinery came really important, because they become the memory of the cell, or at least one of the ways that it can store it really robustly. And obviously, human language has evolved to allow us to take the dynamic of human cultural systems and store it robustly and allow some history to actually enter the modern cultural moment. Because we've recorded our history. We don't record it accurately, but we've recorded some of it like some of it's lost, and just like cellular architectures like that, process became much more dynamic over time. So if you look at all the editing functions of a eukaryotic cell versus a prokaryotic cell, the genome of a eukaryotic cell is a highly dynamic system. It's crazy with all the epigenetic modifications and all the other stuff, like how much editable information is there, so that becomes much more like a large language model to a book. These systems are just evolving, but there's social and cultural technologies. I think the way that we should think about these technologies is a progression of what life has been doing over billions of years. But as you're saying, we're not the Master and Commander of the ship. We're not the largest scale, most complex thing on the planet, and we probably haven't been for most of human evolution, at least since we became social and agricultural. But now we're really just seeing very strong evidence of that directly presented to us. So we might be a little bit different than that dendritic cell, because we have a bit more self awareness of it happening, and we're having this conversation. But whether our conversation of it is also accurate to like, what these higher scale systems that we don't know exactly how to talk about are doing is also another question, which I always worry about, but I think we're doing the best we can.

Emily Silverman  

I want to ask you about the fate of life, or the fate of the universe, you mentioned earlier, this idea of a phylogenetic tree. So you have this explosion of different branches of the tree. You have amphibians, you have reptiles, you have mammals, and it's just branching, branching, branching forward, and life is continuing to generate more and more complexity, until question mark. And there's this idea in your book of the complexity hump, kind of like mixing together milk and coffee. So you have your cup of coffee, you pour the milk in, and that's kind of like the Big Bang, and then you get these beautiful swirling, complex patterns, which is, I guess, kind of like complex things existing in the universe, and then, because of entropy, things ultimately blend into a uniform light brown in your coffee cup. So is it possible that this exploding, branching phylogenetic tree could ever stop diverging and start converging back into oneness?

Sara Imari Walker  

The idea of this complexity hump. It comes from standard physics that if you want to get to a complex state, you need to have a low entropy initial condition. So low entropy would be like the cream separate from your coffee, because it requires someone to have filtered the two to be separate. So that's like a highly ordered state. And then when you mix them, there's this disorder that emerges because they're starting to swirl together, but you still can tell your cream is separate from the coffee. And so an observer would need a lot of information to specify where those exact atoms are. And then when you get to the warm, soft, brown, well mixed state, that's the high entropy state, because at that point to describe that system, you don't need to have any precision of knowing where any of the molecules are, because most of the configurations are just going to have them be random, and therefore you would get something that looked like it was well mixed. So that's sort of the standard description of how you get complexity. Complexity can be these intermediate states from between order and ultimate disorder. And so because of the second law of thermodynamics, and we think that things are always trending and toward disorder, that must be the way that it went. And so this requires fine tuning of the initial conditions of the universe to be low entropy, which is a big problem in cosmology, but also thinking about the nature of life, and is our universe fine tuned for life. So I find all of that discussion really problematic in a lot of ways, because you actually need a human to have the cream separate from the coffee and mix them to begin with. So like these, analogies are only so good as you actually really carefully consider all the assumptions you're putting into them. Which goes back to this idea we were talking about, about spontaneous fluctuation of brains. If you really spend some time with that idea, you can come to some pretty basic reasons why it's ridiculous, and we should make sure our laws of physics don't permit that, because it's ludicrous, instead of assuming that it must be true, because the laws of physics say and so the laws of physics right now say that entropy is increasing all the time, and that this is the narrative as heretical as a theoretical physicist I am, I don't really accept the second law of thermodynamics as like God given truth about the way the universe works. So in so many physics classes, there's a quote from Eddington that they call you should die in shame if your experiment or theory goes against the second law of thermodynamics. That's the sentiment of the quote. I'd have to pull the quote to actually remember it. But when you learn thermodynamics, you're taught this, you should not question the second law. You're be over the head with a ruler. Do not question the second law. And it's really perplexing to me, because as I was starting to think more about life. One of the big paradoxes is it seems to violate the second law, because it's building complexity and kind of a sustained way. And so the idea was it's consistent with the laws of physics, because this idea of fluctuation, you can always have complexity in the middle, but the universe in the long term future is going to be a boring place. I just don't think that's right, actually, like on any level. And part of the reason that I have a hard problem with the second law of thermodynamics, I think it's in a problem. Think it's an approximate law, like I think it works in some situations, but it's a very subjective law. It's actually the most subjective law of physics there is. And I think in some ways, it's even more subjective than issues with like quantum observers and things that happen quantum mechanics, because we don't even know how to interpret quantum mechanics because of the observer problem. But in thermodynamics it's kind of hidden in because you need somebody to actually label the states of your system, and the labels are actually relative to a measuring device or an observer. So you end up how you define something as being high entropy depends on like, how you label the configurations of your system and how you choose to measure them. So that already suggests to me it's a problem. But then there's other issues, like you have to assume you have many copies of the same system to talk about entropy increasing overall of all the systems, and we only have one universe and a particle in a box model. It works great. And for simple systems that have no memory in them, it works great. But I think the narrative that I see with assembly theory is that actually what's happening is the universe is trying to maximize. Amount of stuff that can exist, or maximize the number of configurations that are actually physical. And that process is what we see in our biosphere. And we don't know where that process is going, but it seems to be one that would forever trend in more complexity, and there will be energy limitations on that. And so that's where the interesting thing comes in. But I really like David Deutsch's idea the beginning of infinity that like, once you have systems like us that can build knowledge like the potential of what can happen in the future is limitless and Freeman Dyson also talks about life infinitely far in the future. In his writing, he had this great paper about analog life forms living in an exponentially expanding universe. It's like you have a metabolism that goes slower and slower and slower and slower and slower. It could just persist infinitely far in the future. So I guess my perspective on it is life is, I think, much deeper than current physics in terms of the structure of the universe we live in. I think it's very deeply embedded, because it's about causation, and it's about the structure of the existence of objects, and there are fundamental laws there that could have a totally different explanation and a different projection for what the future looks like than what our current laws of physics say. And I think there will be some interesting intersections of them. One is this idea of virtualization running faster and faster life forms in computers is kind of one way that you could have life persist infinitely far in the future in smaller volumes of space, because you put all this evolution, it takes billions of years of evolution to get to something like us, to build virtual simulators of the past history of our planet. And so now you can put all of that time in a small volume of space. And I think this is like the physical to digital transition. So that's one way you can think about life constructing more, but it would just become more insular and more virtual in a particular physical object. But I think there's other ways that we can think about this expanding possibility space actually being a real, physical thing, that we see the universe getting more complex into the future. So I have two ways of thinking about it, one very virtual and one very physical, but I think they're kind of the same thing, and I don't really know how to reconcile that, so it's a set of thought experiments I'm playing around with right now, actually.

Emily Silverman  

When I was reading your book, I wrote down a couple examples, and I labeled them "The Magic of life," which I've heard you say that magic isn't a word that you love, but that's just how I labeled it.

Sara Imari Walker  

That's okay. I like, I like the feeling of magic. I don't like magic as a substitute for trying to understand things.

Emily Silverman  

For explanation?

Sara Imari Walker  

 Yeah.

Emily Silverman  

The first example I wrote down is from when I was pregnant, and obviously this baby was forming inside of me, and so I got really excited about embryology, and I went on YouTube, and I started watching these embryology videos and learning more about how the thing organizes itself. And I was reading about something called the primitive streak, which is this axis that develops very early on in the embryo. And they've studied it with chicken embryos, and I've read descriptions of it, and they're really magical. I can't think of a better word to describe it, where they talk about the cells like dancing and twirling, and they migrate to the midline, and then they stop, and then they reverse, and they change direction as the embryo forms, and then it thickens, and then it folds. And I know that you've talked about how a lot of the information or blueprint or memory or knowledge that is needed to construct an animal, like a chicken or like a human being, is stored in the genome is stored as DNA. But there is something else about the way that the matter is animated and comes alive and starts dancing unfolding that is clearly not Newtonian motion. It's something else happening. And so what do you think is happening as that blueprint is then actually being built?

Sara Imari Walker  

I use DNA as an example and genomes because people are familiar with it, but I by no means think that all the information or causation that's relevant in biology is in the genome, and I think very little of it is actually in the genome. So the way that I think about biological structures, or like any object that evolution builds, is you have this underlying assembly processes, recursive construction of these objects, and the object itself is actually not the thing that you observe, like if you think of a molecule, it's not the three dimensional molecule, and the configuration of where the bonds are that's relevant, or atomic composition, or its molecular weight, or any of these things you might think of as physical features. The thing that's physically relevant for the processes that we want to think about as being fundamental to life and evolution is actually that construction history. So that assembly space is what we call it, the space of the way that you can actually make this object from smaller parts by reusing them to make this final object. And so if you think objects have a size and time like they have an assembly space underlying them, when you're looking at all these component parts, these are objects that are deep in time, that are actually interacting across that temporal scale at the same time that they're interacting in the three dimensional space. If you think about the three dimensional space as one layer. But then you also think about time is actually a physical component inside the objects.

Emily Silverman  

This is really trippy. Like, keep going.

Sara Imari Walker  

Yeah, it's okay. I mean, it sounds totally crazy, and I hear myself, and I know what I'm saying, and I know it sounds totally- but it's so consistent with what we're doing with how we measure assembly in the lab. This is the physical attribute we look at for molecules. But also when I think about formalizing the origin of life transition and what physics is necessary, this is really the physical description and what gives objects agency, and all of this causation that you're seeing with these things dancing around to assemble something that becomes an embryo. We all start as a single cell, which is just totally mind blowing. Right? Every single living thing on this planet goes through the phase of being one single cell. Development is crazy. How is that even possible? How do we not just, like stick cells together that are already of the type they are, but it's because we're evolved and constructed things, and so all of these things had to be scaffolded on the layers that came before, and so that information still persists in those structures. And so the reason that development can happen, and also evolution can happen so fast, I think, is because of this temporal size of these objects, and that they don't lose their history. It's all bundled up in there, so they're already causally contingent. And the causation is in the structure, and

Emily Silverman  

it's where? -If it's not in the genome, where is it? 

Sara Imari Walker  

The information is in every single physical piece of the cells. I think even the genome we don't actually have right every single piece of the genome of any length is undergoing selection all the time. But we tend to fixate on genes as the functional units and only talk about those. We don't talk about this hierarchical structure where selection is happening across every component of every length. That's easy to think of as a sequence, but now, if you think about the cell, all of these molecules that are reacting all the time, every single scale is undergoing selection from the cell itself, the tissue, higher level pattern the cell, and then the chemistry inside the cell, and then the parts of the molecules inside the molecules. So the way that we think about that is to look at the structure and then build down into the assembly space, and then think about the interactions that are actually happening across the layers of the assembly space, which is like if you could unfold a structure across time, you would have this much larger space, and then you could talk about all the interactions across that space, and that would allow you to explain more of the dynamics that you're actually observing. This is very cutting edge. We haven't really gotten this kind of work out yet, but this is where I imagine the dynamics of assembly spaces are actually going to be able to provide models for what we call agency, or how we view causation more broadly in living systems. 

Emily Silverman  

Yeah, I mean, the second example that I wrote down here is the ATP synthetase machine.

Sara Imari Walker  

Yeah, that one's crazy. 

Emily Silverman  

I remember learning about it in biochemistry, and it's this molecular motor. And I was Googling it, and I came across this article written by this guy, Jonathan McClatchy, who's like a Christian writer slash scientist, and he's trying to puzzle through how this molecular motor could have been built. And so he says one hypothesis is that a proton motor came to be associated with some kind of helicase, and the ATPase of the helicase was driven forcibly in reverse by the proton motor. And that was one option. The other scenario he talks about is maybe there was a translocating protein that got stuck in the protein trans locace, and then subsequently evolved into the central stock of the motor. But then you have to couple that stock to the ring in order for it to rotate and so on. And each example just seems more and more far fetched, and then a lot of story building. He says "it's a marvel of nano technology, an amazing molecular energy turbine. One is led unsurprisingly, to the conclusion that such a machine is best explained by intelligent design." Yeah. So what say you? How does this thing get built? Is it reaching back 3.8 billion years into history, and just like a hand, like puts it together? Or how does that happen? 

Sara Imari Walker  

I think the universe is constructing itself. So I don't find it helpful to explain what exists by assuming that there's some entity outside of our universe, or some explanation that comes from the outside the boundary of our universe, which even standard physics would put because you have to set the because you have to set the initial condition of the universe, and you have to also decide where the laws of physics come from. I don't think any of that's necessary. I think our universe is literally constructing itself. And if you want to understand why certain structures like ATPase come into existence, you actually have to look at the causal contingency about how such a structure could be built, which is like the structure of the assembly space, and you have to look for what objects could co construct with it. This is sort of, I think, the issue why everything looks intelligent in design is, if you look at any object in isolation, of course, it's like, where did that thing come from? But it had to be built by the things in its environment. And so really, what we need to do, and what we're trying to do with assembly theory is look at objects that were co constructed and actually construct what we call the joint assembly space of the construction histories, and use that as a model for how we think about explaining why these objects come into existence. And one of the key features also this goes back in the long history literature of the origin of life, is like this idea of auto catalytic sets that you need systems that one molecule can produce the next molecule can produce the next molecule, and they form closed cycles. We don't exactly think about it that way in assembly theory, but there is this idea that, like when you're moving into this combinatorial space of these incredibly intricate, complex objects, a lot of it involves trial and error and making things that just can't persist. They're not able to persist in time because there's no selection mechanism to reinforce them. What happens is there's a random search, and you get a collapse on the space of a set of objects or molecules that can reinforce their own existence. And so ATPase cannot come into existence on its own. It has to be part of this other structure where it's all self reinforcing. One way I think about it is like we're all hanging on to being able to exist by coexisting with a bunch of things that help reinforce our existence, right? You imagine the universe is like random out there, and there's this insanely large combinatorial space of possibilities. The only things that can be structured are things that can be co built and actually exist along these same lineages. So usually I describe life as lineage, as a propagating information or these sort of construction histories or trajectories we were talking about. So design emerges along those because it starts to look like things have function because they're reinforcing the structure and existence of other objects. And so I think very much a function is an emergent property of this kind of dynamics, not as fundamental, but you would expect to see things over time that look more and more designed because they have more and more history and more and more selection and more and more constraints that were ruled out in order to make that specific object. The only reason that object exists and not some other object is just because of the particular contingency in the historical pathway to get there. So we're ruling out a whole bunch of objects that will never exist, like the transistor example, like all these technologies, that will never exist because they're just not along our particular pathway. But it doesn't mean that anything in that path was designed by something outside of it was all internally constructed, and all has to be self consistent along that particular trajectory.

Emily Silverman  

I once heard you ask in an interview, and I think you were asking this question to yourself, "is there a problem that the universe has that life solves?" I'm wondering where you've landed on that you know, we have these exploding phylogenetic trees of increasing complexity and novelty. What is the problem in the universe that life is solving? 

Sara Imari Walker  

Well, one part of it is, I was told by a very prominent senior physicist, you're not supposed to ask "why" questions after I gave a lecture once, and I thought that was really hilarious, and I actually was doubly bad, because he was like, the last person to ask a question, like, this was Boltzmann, and you know, it happened to him. Oh, my God, are you kidding me? But also, you know, people have been asking, what is life for so long? This is a famous Schrodinger book, and the what is life question is just like over and over again, we're not asking, Why does life exist, or what does life do that nothing else in the universe could do? And I think those questions are actually more scientifically productive, because the what usually follows the why and the what does it do, kind of question. So I thought a lot about the nature of life and like, what are the things about reality we couldn't explain without a fundamental mechanism for life. And like, what is life doing? And the thing I've always landed on, and I had ways of talking about it before assembly theory, and I have ways of talking about it after assembly theory that I think are much more precise and concrete, but the way that I've always thought about it is life is sort of maximizing the number of possibilities that could exist. That was an early thought I had, because there's a lot of stuff that can be created on our planet that cannot be created anywhere else in the universe. And some of those things you might really like, like the Library of Babel or Shakespeare's works, and some of them you might not like Tiktok or pick your favorite political campaign. Like there's lots of good and bad things that people love or hate. I'm pretty agnostic on everything on this planet, but they can't exist anywhere else. They can only exist here. And I think this is really profound. So that was one part of it. But then as I was thinking more deeply and thinking about really, just how large the space of possibilities is, which is when I really started working on assembly and really thinking about the size of chemical space, which is just like, ridiculously huge. I mean, I already talked about this pharmaceutical drug design problem. It's like, epically huge, like, you can't train AI on things that are that big, that you don't even know how to make them think about that. You know, it really struck me that what life is doing is maximizing the number of things that get to exist. And so you think about Darwin's in the struggle for existence, every form is out competing each other. But what assembly theory gives you, that's, I think is a deeper story is there's also a competition to get to exist at all. It's not even like once you exist, you get to keep existing. You want to keep existing. Also, there's a drive to try to make more things exist. And I think this is also one of the reasons that evolved objects like us are so efficient for the universe to maximize existence, because all of those layers of evolution and history and contingency now get packed in a small volume of space, like I'm five foot three, so it's kind of tiny volume, but I represent 4 billion years of evolution packed into one physical structure. And there's a huge amount of recursive layers of structure in me, and all of that gets to coexist in one physical object. So that feature, I think is really important that the universe is actually increasing the recursivity in objects as they get deeper and deeper in time. So they're becoming much larger temporal structure, small volume of space maximize existence, but also being able to use those structures to generate more complexity and more stuff. And so I guess I would think about the universe as kind of like the universe is what exists in a local volume that we can observe. There's the Hubble volume, which is like observational boundary of our universe, and it's just trying to pack as much stuff as possible in there, because everything wants to exist. Because the other option is we don't know what non existence, but non existence is a concept that exists, so I don't actually know what it is to not exist.

Emily Silverman  

Well, you said something once about how a lot of people look up at the night sky and they feel so small, they feel like a spec and so on. But you're really flipping that idea on its head and saying, Actually, we are some of the largest structures that exist in the universe. And I really loved that. In these last couple minutes. I just wanted to ask you about aliens, because we have to people think of aliens as life that evolved off Earth, but you have a slightly different way of thinking about it. So tell us how you think about what an alien is, and then how you're going about trying to discover/ find aliens.

Sara Imari Walker  

I never liked the idea of defining life. A lot of people my field try to do it. I think assembly theory has a natural definition for alien life, and alien life is something that emerged from the random froth of chemistry along a different causal history than we did. And so the origin of life as an experimental paradigm becomes an interesting problem, because now you can talk about this huge possibility space, like this huge volume of things that the universe could create, and you can treat it like a search problem. If you could build a chemical search engine that could actually iterate through different chemical possibilities, how long would it take that search to actually find an alien life form? And so this is actually not my idea. This is Lee Cronin idea, and he's been working on building a technology for that for more than 10 years. He's a pioneer in digital chemistry, and the whole reason that he even got into that field was because he wanted to build robots that could solve the origin of life by trying to scale the technology to do chemical searches. So right now, state of the art in the chemistry lab is you have somebody pipetting individual reactions and things. And so he's basically built a technology that automates that process, and the idea is that we'll be able to scale it. And so we're collaborating on this idea of trying to build large scale, origin life experiments guided by theories. So we have Assembly theories, the abstract, fundamental formalism, and we want to build these very large, robotically driven experiments to look for aliens in the lab. It's a little bit like how particle physics has the theory coupled with massive experiments. And we started a company called chemify, which is trying to scale up the technology. It's actually a drug design company. They build the technology to build molecules, and I'm writing a book to try to get people excited about it. So that's what we're trying to do to get there.

Emily Silverman  

Should we be scared?

Sara Imari Walker  

 So it's like, no, I don't know. I'm terrified, but I don't know how to be scared of understanding things. I love this Marie Curie quote, it's like, now is the time to understand more so we can fear less. And I really feel that very deeply. And so I guess for me, when I talk about the things that we're envisioning doing and the ideas that we're excited about, a lot of people have this kind of fear response, oh my god. Like, what kind of synthetic life forms are gonna come out of these labs, and what kind of technology are you building? But really the motivation is the fundamental understanding. And I think we're so existentially traumatized by these living technologies and all these other things that we don't understand, that unless we really understand fundamentally what life is, we're not going to be able to understand how we're talking about aliens or what they are. We're not gonna be able to understand how we're talking about artificial intelligence as an artificial life or what they are. This is a really existential problem we have to solve, and I would be more afraid of walking into the future not knowing what it is that we're talking about than doing this kind of fundamental science to really understand the core of this question. And so I think there probably are absolutely technological repercussions that would be scary, but I also think that there's a lot of them that are incredibly optimistic. And ultimately, I think as humans, we build our own future. And so I want to keep choosing the optimistic one and the one that gives us the most possibilities for the future, because I think that's what life does. And so I'm going to keep trying to work in that direction. And I hope other people do too, because I think our future could be insanely bright if we want to build it that way. 

Emily Silverman  

Boom, yeah. Well, this has been so much fun for me. I've been looking forward to this for a long time, and I'm just so happy to have you on the show. And wanted to encourage the audience to please consider going out and buying Sara's amazing new book, "Life As No One Knows It." You'll learn a ton. I also recommend just checking her out on YouTube and watching some of her talks, because she's a great writer, but as you can tell from this interview, she is also just so skilled at communicating complex scientific topics to mortals like me. So thank you again, so much, Sara for coming onto the show, and I can't wait to keep tracking your work and see what you do next. 

Sara Imari Walker  

Thanks, it was a lovely conversation. Really enjoyed it.

Emily Silverman  

This episode of The Nocturnists was produced by me and Jon Oliver. Jon also edited and mixed. Our executive producer is Ali Block. Our head of story development is Molly Rose-Williams and Ashley Pettit is our program manager. Original theme music was composed by Yosef Munro, and additional music comes from Blue Dot sessions. The Nocturnist is made possible by the California Medical Association, a physician led organization that works tirelessly to make sure that the doctor patient relationship remains at the center of medicine. To learn more about the CMA, visit CMAdocs.org, The Nocturnists is also made possible by donations from listeners like you. Thank you so much for supporting our work in storytelling. If you enjoyed this episode, please, Like, Share, Subscribe and help others find us by giving us a rating and review in your favorite podcast app, to contribute your voice to an upcoming project or to make a donation, visit our website at the nocturnist.org. I'm your host, Emily Silverman, see you next week. 

Note: The Nocturnists is created primarily as a listening experience. The audio contains emotion, emphasis, and soundscapes that are not easily transcribed. We encourage you to listen to the episode if at all possible. Our transcripts are produced using both speech recognition software and human copy editors, and may not be 100% accurate. Thank you for consulting the audio before quoting in print.


Emily Silverman  

You're listening to The Nocturnists: Conversations. I'm Emily Silverman. Today we're wrestling with one of the oldest questions in science: what exactly is life? For years, we've struggled to answer this question. We've said maybe living things are defined by reproduction or compartments or metabolism or being a self sustaining system, but it seems like there's always an exception to the rule. We either rule something in that clearly isn't life, like a car engine that metabolizes gasoline, or we rule something out that clearly is alive, like a mule, which is sterile and can't reproduce. It seems, no matter what box we try to draw around life, it breaks free from our definitions. But our guest today, astrobiologist and theoretical physicist Sara Walker wonders if maybe it's less important to define what life is and more important to define what life does. This is at the heart of her work on assembly theory, which we talk about in this episode. Sara is the Deputy Director of the Beyond Center for Fundamental Concepts in Science and a professor in the School of Earth and Space Exploration at Arizona State University. She's a fellow at the Berggruen Institute and member of the external faculty at the Santa Fe Institute. Her lab is internationally regarded as one of the leading labs building fundamental theories about life, and she recently wrote a book called "Life As No One Knows It: The Physics of Life's Emergence. I stumbled upon Sara on YouTube a few months ago and could not stop watching her talk. She's one of those trans-disciplinary polymaths who can speak about everything from physics to chemistry to biology to linguistics to artificial intelligence, dissolving the lines that humans have placed on reality and always seeking to unite, reduce, and simplify. I was very honored to have the opportunity to speak with her today about why traditional physics fails to describe life, the tenets of her assembly theory, the way she thinks about that magical phase change from a random chemical soup to life and its markers, her vision of life as a force that propagates through time as an exploding phylogenetic tree, her belief that human beings are actually some of the largest objects in the universe, containing 4 billion years of history and information all curled up inside and so much more. It's a dense conversation, but a fascinating one. So grab your coffee, buckle your seat belt, and enjoy this episode, which begins with Sara reading a short excerpt from her book, "Life As No One Knows It."

Sara Imari Walker  

The paradox of defining life. Look around you right now. I'll bet you can confidently catalog what is living and what is not. I could quiz a four year old and they would probably corroborate your classification. In fact, small children seem rather good at telling the difference between living things and inanimate objects with little explicit direction. Once a child is taught that plants are alive, they can readily tell those apart from non living things and act accordingly. For example, not stomping on the flower bed, but perhaps jumping in the mud, instead. Kids can also misclassify life by assuming inanimate objects like a favorite stuffed animal or life. Are they wrong in doing this, or are we wrong in correcting them?

Emily Silverman  

I am sitting here with Sara Walker. Sara, thank you so much for being here.

Sara Imari Walker  

Thanks for having me. I'm really excited about our conversation.

Emily Silverman  

I discovered you on YouTube, actually, before I knew that you had a book coming out and just listening to you talk, my face kind of melted off, to be honest, from how brilliant you are, not just the content of your work, which changed how I look at the world, but also the way that you're able to communicate that content. It's obviously very high level, complex stuff, but you just have such an ease as a science communicator that I was really excited to bring you on and have you share some of your work with The Nocturnists audience. So let's start with your background. You are a theoretical physicist, and you study the origin of life, which some people probably hear that and associate more with biology than with physics. So how does a physicist come to study the origin of life? 

Sara Imari Walker  

My first response to that was it was accidental, but I don't think it really was accidental. I started my university training and community college, and I was really interested in science, but I didn't know what I wanted to do, and my parents hadn't gone to college or anything, so I just took all the science classes, but I absolutely fell in love with physics. I just thought it was a phenomenal subject, mostly because I love this idea that we can come to deep, fundamental understanding of the nature of reality by thinking about it and by coupling our own thoughts with experiments and what we can test. And I really thought I was going to be a particle physicist or a cosmologist, but when I got to graduate school, my PhD advisor was studying the origin of life or getting interested in it, he was a cosmologist also. And I kind of reluctantly at first, actually started working on the problem, because I was like, this is not what physicists do, exactly what you're saying. But as I thought about that problem more and read more and read more and more of the literature, I realized that it was an incredibly open problem, and there was a lot of room for creativity. And I really felt that the training that theoretical physics provides, and in particular, like what I was really passionate about, which is deep explanations and discovering something new, like origin of life was wide open territory, in part because it doesn't really fit in any of our traditional disciplines, and we don't really have a right way of asking that question. So it seemed like there was a lot of room for the kind of science I wanted to do.

Emily Silverman  

Yeah, and I've heard you say this in interviews before that when people traditionally think of physics, they think of math equations. So probably a lot of people listening to this have taken basic college level physics. I think about Maxwell's equations, or I think about Newton's equations. And basically the way it's set up is you have this fixed law, and then you plug in a set of initial conditions, and it tells you what's going to happen to the apple as it falls from the tree to the ground, for example. But you've said that laws like that don't really work to explain life, and that, in fact, what we have to do is invent a new physics for that. So tell us about that. 

Sara Imari Walker  

Yeah, I have very fond memories of my physics classes, but I have some harrowing ones too. I think we all go through that. It's really interesting because the way that physics works. We're taught about disciplines, and we're taught like, if you're interested in life, go get a biology degree. If you're interested in stars and planets or mechanics or things, go get a physics degree. If you're interested in molecules, go get a chemistry degree. But obviously, Nature doesn't have any of those distinctions. And why I became a physicist was not necessarily because I was interested in a particular question. It was because I really was thinking about this idea that the human mind was so incredibly creative to come up with really abstract and universal understanding of the world. So to me, that's what physics is. And so this example that you're bringing up is that once you have the initial condition and law of motion, you've described the system for all time, and that comes from the era of Galileo and Newton, because they were trying to describe motion and gravitational physics in particular. And that kind of dynamic works very well for those kind of systems, because they have very regular behavior over time. But life does something very different. And I really love this very famous quote from Charles Darwin about "endless forms most beautiful keep evolving and continue to evolve." And he juxtaposes it to the fixed law of gravitation. So apparently he was a huge Newton fan, and I think he was hoping there would be like laws for evolution, like there were for gravity and motion. But he also recognized this very stark contrast, that biological forms seem to be changing all the time, and they seem to change in a way that is historical and path dependent, and this is not what we see in typical physical systems. They usually have the same kinds of rules and the same kind of behavior, no matter when you observe them. And life forms seem to have new rules and new behaviors emerging all the time. And biological evolution over geological timescales is probably the grandest stage for observing that. But even if you observe a cell, you know, it's a complex machine with all kinds of dynamical, non linear processes happening. And so it's been a real challenge for the standard conception of physics in terms of this initial condition and fixed law to describe it. And what I spent most of my career working on is thinking about, how do we actually understand a physics that has these kind of path dependent properties that we see in living systems.

Emily Silverman  

And one of the theories that you're working on is called assembly theory. And you talk in the book about how atoms are countable, we can see them on the periodic table, but once you level up to molecules, the space of possible combinations is huge, and the universe doesn't have the time or the resources to build everything that can theoretically be built. So you and your collaborator, Lee Cronin have developed this theory called assembly theory, which explains why certain things get to exist. So tell us about assembly theory, and particularly what you just said earlier about this idea of path dependence or historical contingency.

Sara Imari Walker  

The idea of path dependence is a nice segue into the example that you've given. Some of these approaches and ideas that I'm talking about seem very deep and abstract and theoretical, but one of the places that we're finding actual utility for them is in drug design, because if you imagine the combinatorial space of molecules that might actually have a biological function, the space of molecules is huge, and then finding that subset that might interact with our biochemistry in just the right way to get the right kind of health outcome is really challenging. There's a huge industry of people trying to tackle that problem. But the way that we think about this combinatorial space, from the perspective of thinking about origins of life, is there's a huge amount of possible molecules that could exist, and the way that you actually start to produce them is you have molecules that have been selected in a particular environment, and they scaffold into other structures. And so assembly theory captures this in a really rigorous way. And I'll explain a little bit about what our observables are and how we think about the theory, which is really related to the idea of, how do you detect life in chemistry, like if you look at a molecule, how do you know it's a product of life or not? But then I'll circle back to the question about the path dependency, and how this relates to a very non-Newtonian kind of physics, or at least not in the standard paradigm of physics. So the basics of the theory are, we have a conjecture that life is the only mechanism the universe has for generating complex objects. And complex objects are things like complex molecules, like DNA is obviously an example, or ATP even so, things that we find in a cell we don't find spontaneously produced in the environment of a planet like no one's observed these outside of biology. And you could take for granted that the universe could just produce those objects outside of life, but actually a key conjecture of assembly theory is that that's not possible. And that's one of the things that we actually aiming to formalize to test that you only get complexity as part of a living process. And the way that we formalize that, it's almost like thinking about Lego blocks and building up a LEGO Universe. Or if you play Minecraft, you have all the pieces you can put together to make all these complex structures. But we start with thinking about bonds, and how bonds are actually the fundamental unit of building molecules. So if you prefer the Lego analogy, you can think about your Lego blocks, and you can stick them together, and then you can reuse structures you've already built. And let's say that we have a target structure, like maybe we want to make the molecule ATP, or maybe want to make Hogwarts Lego Castle. A couple examples of complex objects in your mind. So you can take this idea of trying to build an object by putting pieces together and then taking those pieces and reusing them and get up to the final structure. And what we try to look for is actually the shortest pathway to do that, which we call the assembly index. So there's sort of a minimal constructed complexity for making the object, or a minimal set of constraints that's necessary minimal history to produce that particular structure. And then we also look for that structure in high abundance. So the idea being on our planet, there's not one Lego Hogwarts Castle, there's millions of boxes, potentially of them, and patterns on the planet so that they're repeatable structures. That's because they've actually been selected by the environment, by humans, by kids that are really big fans of Harry Potter to be a really highly abundant object on our planet. Whereas, if you took the equivalent amount of Legos, I think there's like a couple 1000 pieces in there, all the random structures that you can make out of that, that's a huge space, but most of those structures will never actually be built ever in the history of the universe. It's actually probably the number of structures is so large the universe doesn't have enough time, even if you were putting Lego blocks together every second, to exhaust that space. So we have the assembly index and the copy number as the two key variables that we're looking at, and we view those together as a signature of constructed complexity. Basically what life does. Life is the mechanism the universe has for making complex objects. And it's important that we have those two things, because we can actually go in the lab and observe them and measure them. So the way we do it with the assembly index's shortest path, is using mass spectrometry. And it turns out that with a mass spec, you can probe that feature of a molecule. And if you're doing mass spec, you already have to have thousands of copies of a molecule even to recognize that it's there and measure it just with standard laboratory equipment. And so what we were able to do was actually go in the lab, and this was an experiment done in Lee Cronin's lab. I think this was just such a cool experiment. What they did was actually take living and non living samples, and they figured out how to measure assembly in those samples, and they were able to show that above about an assembly index of 15. So thinking 15 steps where you reuse parts and build an object in chemical space. So you're using bonds to form these structures. About above 15, you only see molecules that were produced from the Living samples, and there were even blinded samples from NASA that were supposed to be confounding ones. It was like Murchison meteorite, which is like one of the most complex examples of chemistry in our solar system. And it didn't pass the test of producing high assembly molecules, like if you just have random structure making molecules, you can make all of the very simple molecules pretty easily. But what we see in prebiotic systems,when you don't have selection and you don't have evolution, or you don't have constraints, you just get tar, which is an undifferentiated goop of all of the kind of simple molecules. You don't get high abundance of very refined molecular structures like what we see in biochemistry. And so we think that unless you have selection and evolution, you can't cross a complexity threshold, which experimentally seems to be about 15 steps. That might not be a universal number. It's experimentally verified, but we have a lot of work to do, but we conjecture that there's always a threshold, and above that, you have to have a path dependent trajectory to make those objects, because they're just so complex, and the number of errors that you could have making the wrong part, and bifurcating off into all the other possibilities that could be built is so large that you would never expect that structure to exist unless something selected for it to exist, and it was built by an Information Processing System, or something that we might call alive. So that's sort of the key features of assembly theory, and why it's very different than that standard Newtonian paradigm is the universe of possibilities gets larger with every single step that you take trying to build a new object, because every time you could have put some other things together differently, and so you have to trace this path dependent trajectory out. And in a Newtonian world, the universe never gets bigger as a function of the complexity time always has the same size. But what we see in assembly theory is this notion of this combinatorial space that living things are moving through. And as it gets more complex, the space gets larger and larger, and that actually informs a lot of the foundational physics that we're thinking about and the philosophy that I'm most interested in.

Emily Silverman  

So the idea is that if you look at a molecule like Taxol, for example. I know as an example that you use a lot Taxol is, I believe, a molecule that was originally found in plants, yes, or a molecule like ATP, or even a giant molecule like a ribosome. It's like, once you get ATP and once you get the ribosome, all sorts of things become possible. How do you think about these stop gaps? I've heard you compare the ribosome kind of like a transistor, which is a piece of technology that, once that was invented, it was like, Okay, now we have the transistor. We can build all these other machines. Or once the light bulb was invented, it's like, Okay, now we have the light bulb. We can make all of these different types of lights. Do you think about those moments as special, where you create something that blooms out into more possibilities?

Sara Imari Walker  

I think that's just the way that our universe works, and it's the only way it can make complex objects. When we're thinking about life evolution, at least, we all have a picture in our mind of these branching phylogenetic trees that are constructed from genomic information, right? And before that, people were trying to do it by morphological similarity. So we've had this kind of tree structure in the way that we've been visualizing the nature of life for at least 150 years, since the time of Darwin. Well more than that now. And some people will be like, Well, what do you need assembly theory for we already have good enough evolutionary biology, but the point of assembly theory is actually try to look at selection and evolution before you have things like genomes, because we're trying to explain the origin of those things. So we need deeper evolutionary principles than what biological architectures provide us, because we're trying to explain the origin of things like the cell, not use the cell as a scaffold for talking about further complexity. And so what we see looking at assembly theory is, if you think about the universe as this giant space of possibilities that could exist, selection has to mediate these pathways to build up the next sort of scaffold. So I do like the example of transistors and the point that I would make there, because technology is very perceptible to us, is that most of our modern technologies are built on the transistor, because transistors have been a very prevalent component of a lot of technology. So now we have factories that can make them, and they make devices compatible with each other. And so it's selected out a whole bunch of other options that might have been equivalent or even better than transistors in terms of efficiency and all other kinds of technology, but we're kind of locked into using transistors now because they're so prevalent and we're on its trajectory now where transistors are going to be part of our future technology. And a similar transition happened very early in the evolution of life on Earth, with the invention of the ribosome. Evolution had to construct the ribosome. Once the ribosome and the rest of the translation machinery was constructed by evolution, it became solidified as a structure on our planet that could elaborate further complexity and kind of ruled out the other possibilities that maybe other kinds of cellular architectures or things could have evolved at the same period. But this became the one that was successful, and now we find it in all life on Earth. So we have a single branching point for life. But I think what assembly theory is doing is making that branching structure go deeper. And so this is how we're thinking about the nature of the actual structure of physical objects, and that they're all related to each other through this historical contingency of this reuse of parts and the causation. Propagating through all these objects.

Emily Silverman  

You wrote an article in Eon magazine called "Life is Not Alive," where you say that things like cats and dogs are alive, and things like sofas and rockets are life, and the reason that sofas and rockets are life is you don't just randomly find a sofa in the universe that you need life around to invent and build a sofa. And like you said, there's a high copy number. We see a sofa here and a sofa there, and there's green sofas and black sofas and velvet sofas. So I never really thought of it that way. I always thought of it as animate and inanimate, but you're bringing in this new category, which is inanimate matter that is life, because life needed to be around to create it. So talk a bit more about those distinctions and why they're so important. 

Sara Imari Walker  

I'm a professional astrobiologist. So the two problems that drive my career really the origin of life. How is the transition from non living to living matter, and then the other one being life detection on other worlds. And if you're thinking about detecting life, you don't necessarily care if you're finding an extant organism, something that's alive. Now in the Mars community, a lot of the focus is on finding fossils, or fossil evidence of life. But I think most of us would be convinced if we found a cell phone on Mars, it would be evidence that life was there. Like, we go find a robot, it's probably a human robot that was put there, not that geochemistry of Mars spontaneously fluctuated curiosity on the surface right, like it wouldn't be explanatory. It's probably the case that some intelligent planet nearby actually built the technology to put it there. So we recognize in our narratives already that there are these certain structures that must be constructed by humans or some human-like physical system, something with comparable intelligence, or are necessarily the product of evolution. That's kind of the idea of a bio signature. And so when we were thinking about this fundamental story about the physics of life, and thinking about the combinatorial universe of chemistry and all the structures that could be built out of it. Chemistry, at least, is iterable, right? So I like thinking about chemistry because you can talk about the atoms in the periodic table and the combinations, and maybe estimate the size of a chemical space using chnops, carbon, hydrogen, oxygen, phosphorus, nitrogen, sulfur, right? Like the standard biological elements, we can kind of imagine sticking them together by bonding rules and estimating how many structures there are. We don't know how to estimate how many kinds of life forms are out there, right? Like how many cell like structures, if you think about the Cambrian explosion and how many body plans were made, it could have been exponentially more than that was probably just what the resources and physical environment imposed. But most of those aren't around anymore. And if we think about all possible technologies, that's an even larger space. So all of these combinatorial spaces are quite large. And in assembly theory, we always think that there's this boundary that needs to be crossed, that only evolution can do the process we call evolution, but thinking about it much more deeply in terms of this historical contingency and the construction process, not as a passive filter. It's a very active mechanism of selection, selecting the next thing that the universe can construct. When you're thinking about that kind of structure, it's very clear that you have to have kind of an abrupt transition point, and when you cross that transition point, those objects can only exist in the presence of an evolutionary process or selection, because the space of possibilities they live in is so large that the universe could not possibly do an exhaustive search to discover them. It would have to do a directed, constructed process. And so we kind of differentiate all of the objects above that that are impossible to form randomly and require information for their construction to be life, anything in that space. And then alive to me is the actual active mechanism of building a trajectory into that space. So something like us is alive because we're creative in the universe, we're generating new possibilities. But things like water bottles, like I have on my desk that are artifacts of life, they're still life. Like bottles don't spontaneously come into the universe outside of things like us to generate them. So we have a causal pathway like this thing has to be built before this can be built, before this can be built. So objects actually necessitate their existence based on other structures that have been selected in the evolution of a biosphere, but it's the active process of building that's alive. So in some sense, it's like, is it the process or the thing? And we're really trying to merge both of them into the same story, but this is where you can have a more holistic view about the nature of the physics of life, really, just thinking about life is everything that's complex, too complex for the universe to generate spontaneously and requires some kind of construction and informational constraints. And then it's much more perplexing and interesting to find things like us that are alive. If we want to go look for aliens, probably we want to find alive ones and not dead ones. We had this joke in that essay, actually, about Schrodinger's cat. And I love this because, you know, it's a very physics example about is the cat alive or dead? But in assembly theory, in the way we think about life, the cat can be dead but still be life. It's not alive anymore, right? But it's, you know, a dead cat is a product of evolution in the same sense that a living cat is. They don't spontaneously form in the universe either. So you're kind of talking about two different levels of description of the same physics, but that's kind of the idea behind it.

Emily Silverman  

Yeah, and you joked about this in the talk that you gave at Mindfest, which I watched and which we'll link in the show notes, where you talk about this thought experiment of the Boltzmann brain. And so the idea for the audience is that human brains exist in the universe because we all have a human brain, therefore it is possible for a human brain to exist in the universe. Therefore, why shouldn't a human brain spontaneously fluctuate into existence? Because physics theoretically allows it. But then you say, and I quote, "People really intrinsically think the universe can generate anything, anywhere, for free. Where is all that information? Please tell me." And so where is the information? Like, where? Where is the locus of the information?

Sara Imari Walker  

A lot of my work is trying to build what seem like counterintuitive notions, but they're actually very intuitive, if you're willing to sort of break your training and how you've been taught to think. Why I was joking about that is because I think our standard education and the way that we talk about the world has, right now, at this moment in history, really trained us to think that any object can fluctuate into existence. It's very standard in physics, and I learned it in many physics classes that there was just a low probability for thermodynamic fluctuations or quantum fluctuations, out of the vacuum to produce objects of arbitrary complexity, just because it's written in the laws of physics as we understand them. Yet we have no observational evidence for that. It drives me mad, because I think about it this way now. But like, obviously I didn't think about it this way most of my career, like when I was a student, stuff, I never would have thought about this way. But it's like, seems so obvious to me, is like, in order to have any object fluctuate into existence at any point in space and time be a possibility, it means, like, everywhere has to have the information for every object, which means you have to have an infinite density of information in our universe at every point. And that's just totally unphysical. So we have, like, these really unphysical things just baked into the way that we talk about the world, and because they seem so obvious to everyone that that's the way the world works, nobody takes a step back and questions these basic assumptions. And I think this is actually really the art of theoretical physics. Like most of the major advances in theoretical physics have been someone that's willing to sit down and question the most basic assumption that everyone assumes is true. It's like Einstein with like, the constancy of the speed of light, right? That's like, a very obvious example, because even with the Michelson Morley experiments that validated experimentally with high precision that the speed of light was constant independent of the motion of the Earth, people still didn't want to accept it. They assumed the experiment was wrong. And Einstein was like, I'm just gonna assume it's right and just run with it, and we get relativity right. So it's kind of interesting, like there's just these things that we just assume. So that's one of them, for me, that I find kind of mind boggling now, is that the idea of Boltzmann brains is taken seriously in some of the physics community, and it's to the point like in my history as a cosmologist, like some people reason their cosmological models on the idea of trying to rule out Boltzmann brains being more prevalent than we are, because once you assume a brain can fluctuate into existence, it's very likely that any observer is a Boltzmann brain over cosmological timescales, instead of being something like us that was evolved on a planet over billions of years, and assembly theory says something very firm like that is not possible. The only way to get to something like us is through the process of evolution and construction, and to get something like us means we have to come with a whole bunch of other structures like us. So we emerge in populations and things. And it's like, very consistent with what we actually observe the world to be like.

Emily Silverman  

Right before, you need a human brain, you need a human body, you need a ribosome. You need it's like the Carl Sagan quote, where he says, "How do you bake an apple pie? First you have to invent the universe." 

Sara Imari Walker  

That's right, and then you have to go through all those evolutionary steps to get one right. Yeah, that's right. It's not just in the initial condition.

Emily Silverman  

I'm wondering if assembly theory could apply to abstract objects as well. So for example, this thought experiment of could a random letter generator left to operate for an infinite amount of time write the book of Hamlet for example. My husband and I actually have this ongoing fight about this question. 

Sara Imari Walker  

Oh, really?

Emily Silverman  

Yeah, you know, and I always felt like, yeah, maybe a random letter generator could produce a few words or some word fragments, but it's not going to write Hamlet. For it to write Hamlet, you need something else. You need consciousness or intent or something like that. And he would always argue like, no, no, you don't understand what infinity means. Like, if you let it run for infinity, by definition, it'll write Hamlet. And before you respond to that, I just wanted to share one more example that's tied to language, which is this short story The Library of Babel by an Argentinian writer, Jose Luis Borges. And the story describes a universe that consists of an infinitely large library filled with books, and the books are filled with random letters, and the librarians are driven to suicidal despair because they know that the library must contain every book that has ever been written, and yet they're surrounded by gibberish. And if they ever stumble upon a book that has a sentence in it, they kneel in worship. And so I was wondering if something like the assembly index threshold could apply to language too. Could you say that after 15 letters, it stops being a coincidence, and there's somebody writing?

Sara Imari Walker  

Yeah, so I don't think the value of 15 will be meaningful for language, because people start to assign a very particular value to 15 and like it becomes like the number 42. It's not the answer to life, the universe and everything. It's an empirical validation of this idea of a threshold. We don't know where the threshold is going to be in most systems, and that's actually something that I'm actively working on, is like a general theory of the idea of like a phase transition associated with the origin of life, and why does it happen at a particular complexity value in assembly theory? But to your point, yes, I think the ideas apply to language as much as they do to molecules. And for a long time, I had been writing about the nature of mathematics as a physical system, which is like, kind of a really weird way to think about it. But the reason this came about is because I was trying to think about, like, how could I really point out the discrepancy that we were talking about before, about, like, what physics says and what we need to explain life. And everybody thinks physics as we understand it now is right and, like, the ultimate explanation. So like, how do you build a thought experiment or something that turns the narrative for people to see that there's, like, a totally different way of looking at it? So I started doing these thought experiments on the theoretical physics of theoretical physicists as I was thinking about in my own head, but really thinking about, like, you know, the laws of physics themselves as physical objects that have merged in our biosphere like Newton law of gravitation obviously describes something we think is objective about the world, which is like gravity, but it's also an equation we write down on paper, and that equation has causal power, in the sense that we can teach it to undergrad physics students, and they can roll balls down inclined planes and test its properties, but it also allows us to launch satellites into space and do a myriad of other things that wouldn't be possible without that knowledge and that kind of small information packet. And so to me, that's a physical structure that exists on our planet, and I was really trying to elucidate that. So the reason I'm prefacing this and talking about language is we think about language as being abstract, but it must have some kind of physicality to it, because language actually influences the physical structure of reality. It changes the way we perceive reality, but also through our actions, we actually change the nature of what exists and what gets to happen. I actually even have an example in my book, which is in the middle of the book about trying to prompt people to turn the page is like a physical act, right? Like a word actually activates you to turn a page of a book. Like, how is that possible? That's like a really weird sort of thing.

Emily Silverman  

And in your MindFest talk, you take a moment and you say, "Everybody raise your hand" and then everyone in the room raises their hand, yeah? And so that's a physical act.

Sara Imari Walker  

aYeah, so I think we have evidence all around us. Yeah, that language has some physicality to it. As far as applying assembly theory, it's really non trivial, because, you know, I talked about this idea of assembly index and copy number, and they're not really separable. We can calculate the complexity of this molecule and count the number of copies. It's pretty easy for molecules, but when you move to other systems, like minerals or language that don't necessarily have natural boundaries for language. Is the boundary a word? Is it a meaningful segment of a sentence? Is it a sentence? Is it a paragraph? This is part of the foundation of trying to understand language, and the machine learning is that, like the things that actually become tokens are not necessarily the things that we prescribe meaning to, as far as the predictability of language. But the point there is just that we actually need to start thinking about language as a physical system. And one of the other tenants we have of assembly theory is if you're going to talk about assembly index and copy number, you have to be able to measure them as part of the systems. We have to think about what are interactional units with a language that we interact with. And so I do have work in my lab right now, actually trying to apply assembly theory to language, but applying the principles of the abstract theoretical physics to language is very non trivial, because we're really trying to think about language as a physical system. But I do think to your example about the Library of Babel, that there would be a threshold, and you would be able to say that this thing is definitively constructed by a meaning making system that was reusing elements that had specific associations with other ones, and that the structure of a language would emerge. Another thing that happened with large language models is that people were starting to realize that the structure of almost all human language is identical. If you remove the actual differences about like what words are used for what in different languages, there's this like topology to the structure of human language, about the associations between concepts that we build, that seems to be pretty universal, and I think assembly theory would be quite good at picking up those causal structures underlying language, which are the structure of association. So even when I use language, I don't really think about the meanings of words. I always think about structural associations, which is super important for my line of work, actually, because there are no predefined concepts. So I have to always know that the meanings of the words that I'm using are not the meanings that I intend, because we don't have words for these concepts yet. So I'm kind of playing around with the relations of words like you probably noticed already in this conversation. I'll interchange complexity, assembly, information, evolution, causation, constraint. And to me, they all mean the same thing. I know it's like dizzy,

Emily Silverman  

Yeah, consciousness, meaning making system. Yeah, yeah. 

Sara Imari Walker  

You have to kind of play around with the words. This is part of the thing about effective communication. When you do that, enough people start to build their own associations in their mind, and then they can start to see the picture of what you're talking about. So I think this is really important to communicating deep ideas effectively. Is not to use the same words all the time. Yeah,

Emily Silverman  

I love the idea of a theoretical physics lab collaborating with a team of linguists. I mean, to me, that is just the perfect example of how your work jumps from discipline to discipline. It cannot stay confined inside physics or chemistry or biology, and that's part of what makes your work so exciting to me. I wanted to ask about you mentioned large language models, and you talk a lot about technology, actually, in your book. And there's a couple questions I want to ask you, but maybe first I'll ask you about just how you see technology. You talk about how Earth has developed a biosphere, and that biosphere has now invented a technosphere.

Sara Imari Walker  

I think sometimes about what our biosphere is doing is unfolding structure over time, and so in some sense, and I actually was thinking about this quite a lot this morning, about the simulation argument, where people think reality is literally simulation, and it's so interesting that we can viscerally feel that right now, because we're seeing all of these objects coming into our environment that were once just ideas in human minds. And so pretty much everything in our environment is constructed nowadays by a human mind. I have a microphone sitting in front of me, a board behind me with a whole bunch of scribbles on it that were iterated by a human evolution took 4 billion years to construct something like us, and all of that selection and all of that historical contingency, of all those steps along the way are actually embedded in us as physical structures. So one of the concepts I've taken from Assembly theory, which I think is a direct consequence of how we measure attributes of the theory, is that evolved objects actually have a size and time, and that size and time, the more complex the object, the deeper in time that object is. So I would think of something like myself or you as an object that takes 4 billion years of history to be able to create you, but that's all rolled up recursively through this process of reusing parts like that stack of structure is you. Then we go out in the world and we create things. We're putting all of that history into these structures. So like the world is in some sense becoming more life, maybe not yet alive around us in the technology that we're creating, because it required so much evolution and so much packing in of all that selection and all of that time and structures like us to generate these things in the first place. So technology, to me, is just a continuation of the process of life on Earth. I think the more interesting question there, and the more interesting transition is, when do we say that our technology itself is alive too, in the sense that technology becomes a creative force in the universe, in the way that we are, it has causal power of its own, if you want to talk about it that way. And I don't know that that we're there yet with any individual device or technology, including large language models. I don't really see them as being incredibly creative beyond what has been put in, as far as what human language is already capable of. But I do think if you look at like the global ecosystem between technology and humans and like what we're creating, that that whole system is in some sense living like. Social systems are living, and they exist at different scales. So one of the other things that I think about quite a lot is how life is a process that recurs across scales, and what's the role of technology in continuing that process, and thinking about technology as a social scale or a cultural phenomenon. So I, like a lot of the discussion now, thinking about LLMs (large language models) as cultural technologies and these kind of things.

Emily Silverman  

Yeah, because there is this interesting thing that happens with Life where I'm a human being. I'm sitting here, I have a ton of cells in my body. I have a dendritic cell in my immune system, and that cell, you could argue, is alive, because it's constantly sampling its environment. It's making decisions about whether to travel to the lymph node to present the antigen and things like that. But that dendritic cell probably is not aware of Emily, like it is a living thing inside, a living thing that has a consciousness that maybe that one cell is not aware of. So do you think it's possible we could evolve in a direction where you and I are like the dendritic cells, and then God knows what, the new GPT, or whatever is the Emily. And maybe we're not even aware that it exists, and we're just kind of living our lives, swimming around. But there's like a new level.

Sara Imari Walker  

I think we're already kind of there. I mean, I mean, I think that's what human, social and cultural systems are. And a lot of people that work on complex systems really think about them as living, evolving entities. Like culture itself clearly has evolutionary properties, and it clearly controls the behavior of individual actors that are part of that social system. So if you think about like the cellular contract, cells form together to make tissues, and they're not going to apoptose unless there's some traumatic event that happens, or why would you even have programed cell deaths, unless you were a part of a society that needed to have that function. So I think societies already have some of that structure, but I think certainly what I view with modern technologies is they're becoming the informational architecture of these social entities. So something like a large language model cannot come into existence until you have a society that evolves language and then evolves technology to record vast volumes of that language in order to even train these models. And so the large language model is scaffolded on all of this human evolution, but where it scaffold is actually at the societal layer, not the individual. And so I think this is one of the reasons that they're deeply perplexing to us, because when we interact with them, and this is why it was such an epistemic shock for so many people the first time that they ever interacted with one is like you're taking something that has traditionally been part of a social system distributed across many human minds, language and the dynamic of that and its evolution has always been, we're the only substrate, and then you train a model on it, and you put it in a box, and then suddenly you can interact with that as like a dynamic. It's totally crazy, but I loved when that was coming out. A lot of people were pointing to how in ancient Greece, people were trying to not have people learn how to read and write, because they were afraid of the dead talking, and they thought this was a really dangerous technology, because the dead should not speak to the living, so you should not record what you've said as written text. So in the past, we've had ghosts. The Dead could talk through books over hundreds of years, and now it's become a process where now the dead have much more of a live interactive voice, because you can obviously train someone, if they have enough text outputs from their own vocalizations, you can train a digital twin of them. So now it's like the ghost has gotten two dimensional instead of one dimensional. But I think about it a little bit like how the genome had to evolve in early life, in some sense, because if you're thinking about a dynamic chemical system that has no record of its past history. It has to start building record keeping devices. This is where the genome and the translation machinery came really important, because they become the memory of the cell, or at least one of the ways that it can store it really robustly. And obviously, human language has evolved to allow us to take the dynamic of human cultural systems and store it robustly and allow some history to actually enter the modern cultural moment. Because we've recorded our history. We don't record it accurately, but we've recorded some of it like some of it's lost, and just like cellular architectures like that, process became much more dynamic over time. So if you look at all the editing functions of a eukaryotic cell versus a prokaryotic cell, the genome of a eukaryotic cell is a highly dynamic system. It's crazy with all the epigenetic modifications and all the other stuff, like how much editable information is there, so that becomes much more like a large language model to a book. These systems are just evolving, but there's social and cultural technologies. I think the way that we should think about these technologies is a progression of what life has been doing over billions of years. But as you're saying, we're not the Master and Commander of the ship. We're not the largest scale, most complex thing on the planet, and we probably haven't been for most of human evolution, at least since we became social and agricultural. But now we're really just seeing very strong evidence of that directly presented to us. So we might be a little bit different than that dendritic cell, because we have a bit more self awareness of it happening, and we're having this conversation. But whether our conversation of it is also accurate to like, what these higher scale systems that we don't know exactly how to talk about are doing is also another question, which I always worry about, but I think we're doing the best we can.

Emily Silverman  

I want to ask you about the fate of life, or the fate of the universe, you mentioned earlier, this idea of a phylogenetic tree. So you have this explosion of different branches of the tree. You have amphibians, you have reptiles, you have mammals, and it's just branching, branching, branching forward, and life is continuing to generate more and more complexity, until question mark. And there's this idea in your book of the complexity hump, kind of like mixing together milk and coffee. So you have your cup of coffee, you pour the milk in, and that's kind of like the Big Bang, and then you get these beautiful swirling, complex patterns, which is, I guess, kind of like complex things existing in the universe, and then, because of entropy, things ultimately blend into a uniform light brown in your coffee cup. So is it possible that this exploding, branching phylogenetic tree could ever stop diverging and start converging back into oneness?

Sara Imari Walker  

The idea of this complexity hump. It comes from standard physics that if you want to get to a complex state, you need to have a low entropy initial condition. So low entropy would be like the cream separate from your coffee, because it requires someone to have filtered the two to be separate. So that's like a highly ordered state. And then when you mix them, there's this disorder that emerges because they're starting to swirl together, but you still can tell your cream is separate from the coffee. And so an observer would need a lot of information to specify where those exact atoms are. And then when you get to the warm, soft, brown, well mixed state, that's the high entropy state, because at that point to describe that system, you don't need to have any precision of knowing where any of the molecules are, because most of the configurations are just going to have them be random, and therefore you would get something that looked like it was well mixed. So that's sort of the standard description of how you get complexity. Complexity can be these intermediate states from between order and ultimate disorder. And so because of the second law of thermodynamics, and we think that things are always trending and toward disorder, that must be the way that it went. And so this requires fine tuning of the initial conditions of the universe to be low entropy, which is a big problem in cosmology, but also thinking about the nature of life, and is our universe fine tuned for life. So I find all of that discussion really problematic in a lot of ways, because you actually need a human to have the cream separate from the coffee and mix them to begin with. So like these, analogies are only so good as you actually really carefully consider all the assumptions you're putting into them. Which goes back to this idea we were talking about, about spontaneous fluctuation of brains. If you really spend some time with that idea, you can come to some pretty basic reasons why it's ridiculous, and we should make sure our laws of physics don't permit that, because it's ludicrous, instead of assuming that it must be true, because the laws of physics say and so the laws of physics right now say that entropy is increasing all the time, and that this is the narrative as heretical as a theoretical physicist I am, I don't really accept the second law of thermodynamics as like God given truth about the way the universe works. So in so many physics classes, there's a quote from Eddington that they call you should die in shame if your experiment or theory goes against the second law of thermodynamics. That's the sentiment of the quote. I'd have to pull the quote to actually remember it. But when you learn thermodynamics, you're taught this, you should not question the second law. You're be over the head with a ruler. Do not question the second law. And it's really perplexing to me, because as I was starting to think more about life. One of the big paradoxes is it seems to violate the second law, because it's building complexity and kind of a sustained way. And so the idea was it's consistent with the laws of physics, because this idea of fluctuation, you can always have complexity in the middle, but the universe in the long term future is going to be a boring place. I just don't think that's right, actually, like on any level. And part of the reason that I have a hard problem with the second law of thermodynamics, I think it's in a problem. Think it's an approximate law, like I think it works in some situations, but it's a very subjective law. It's actually the most subjective law of physics there is. And I think in some ways, it's even more subjective than issues with like quantum observers and things that happen quantum mechanics, because we don't even know how to interpret quantum mechanics because of the observer problem. But in thermodynamics it's kind of hidden in because you need somebody to actually label the states of your system, and the labels are actually relative to a measuring device or an observer. So you end up how you define something as being high entropy depends on like, how you label the configurations of your system and how you choose to measure them. So that already suggests to me it's a problem. But then there's other issues, like you have to assume you have many copies of the same system to talk about entropy increasing overall of all the systems, and we only have one universe and a particle in a box model. It works great. And for simple systems that have no memory in them, it works great. But I think the narrative that I see with assembly theory is that actually what's happening is the universe is trying to maximize. Amount of stuff that can exist, or maximize the number of configurations that are actually physical. And that process is what we see in our biosphere. And we don't know where that process is going, but it seems to be one that would forever trend in more complexity, and there will be energy limitations on that. And so that's where the interesting thing comes in. But I really like David Deutsch's idea the beginning of infinity that like, once you have systems like us that can build knowledge like the potential of what can happen in the future is limitless and Freeman Dyson also talks about life infinitely far in the future. In his writing, he had this great paper about analog life forms living in an exponentially expanding universe. It's like you have a metabolism that goes slower and slower and slower and slower and slower. It could just persist infinitely far in the future. So I guess my perspective on it is life is, I think, much deeper than current physics in terms of the structure of the universe we live in. I think it's very deeply embedded, because it's about causation, and it's about the structure of the existence of objects, and there are fundamental laws there that could have a totally different explanation and a different projection for what the future looks like than what our current laws of physics say. And I think there will be some interesting intersections of them. One is this idea of virtualization running faster and faster life forms in computers is kind of one way that you could have life persist infinitely far in the future in smaller volumes of space, because you put all this evolution, it takes billions of years of evolution to get to something like us, to build virtual simulators of the past history of our planet. And so now you can put all of that time in a small volume of space. And I think this is like the physical to digital transition. So that's one way you can think about life constructing more, but it would just become more insular and more virtual in a particular physical object. But I think there's other ways that we can think about this expanding possibility space actually being a real, physical thing, that we see the universe getting more complex into the future. So I have two ways of thinking about it, one very virtual and one very physical, but I think they're kind of the same thing, and I don't really know how to reconcile that, so it's a set of thought experiments I'm playing around with right now, actually.

Emily Silverman  

When I was reading your book, I wrote down a couple examples, and I labeled them "The Magic of life," which I've heard you say that magic isn't a word that you love, but that's just how I labeled it.

Sara Imari Walker  

That's okay. I like, I like the feeling of magic. I don't like magic as a substitute for trying to understand things.

Emily Silverman  

For explanation?

Sara Imari Walker  

 Yeah.

Emily Silverman  

The first example I wrote down is from when I was pregnant, and obviously this baby was forming inside of me, and so I got really excited about embryology, and I went on YouTube, and I started watching these embryology videos and learning more about how the thing organizes itself. And I was reading about something called the primitive streak, which is this axis that develops very early on in the embryo. And they've studied it with chicken embryos, and I've read descriptions of it, and they're really magical. I can't think of a better word to describe it, where they talk about the cells like dancing and twirling, and they migrate to the midline, and then they stop, and then they reverse, and they change direction as the embryo forms, and then it thickens, and then it folds. And I know that you've talked about how a lot of the information or blueprint or memory or knowledge that is needed to construct an animal, like a chicken or like a human being, is stored in the genome is stored as DNA. But there is something else about the way that the matter is animated and comes alive and starts dancing unfolding that is clearly not Newtonian motion. It's something else happening. And so what do you think is happening as that blueprint is then actually being built?

Sara Imari Walker  

I use DNA as an example and genomes because people are familiar with it, but I by no means think that all the information or causation that's relevant in biology is in the genome, and I think very little of it is actually in the genome. So the way that I think about biological structures, or like any object that evolution builds, is you have this underlying assembly processes, recursive construction of these objects, and the object itself is actually not the thing that you observe, like if you think of a molecule, it's not the three dimensional molecule, and the configuration of where the bonds are that's relevant, or atomic composition, or its molecular weight, or any of these things you might think of as physical features. The thing that's physically relevant for the processes that we want to think about as being fundamental to life and evolution is actually that construction history. So that assembly space is what we call it, the space of the way that you can actually make this object from smaller parts by reusing them to make this final object. And so if you think objects have a size and time like they have an assembly space underlying them, when you're looking at all these component parts, these are objects that are deep in time, that are actually interacting across that temporal scale at the same time that they're interacting in the three dimensional space. If you think about the three dimensional space as one layer. But then you also think about time is actually a physical component inside the objects.

Emily Silverman  

This is really trippy. Like, keep going.

Sara Imari Walker  

Yeah, it's okay. I mean, it sounds totally crazy, and I hear myself, and I know what I'm saying, and I know it sounds totally- but it's so consistent with what we're doing with how we measure assembly in the lab. This is the physical attribute we look at for molecules. But also when I think about formalizing the origin of life transition and what physics is necessary, this is really the physical description and what gives objects agency, and all of this causation that you're seeing with these things dancing around to assemble something that becomes an embryo. We all start as a single cell, which is just totally mind blowing. Right? Every single living thing on this planet goes through the phase of being one single cell. Development is crazy. How is that even possible? How do we not just, like stick cells together that are already of the type they are, but it's because we're evolved and constructed things, and so all of these things had to be scaffolded on the layers that came before, and so that information still persists in those structures. And so the reason that development can happen, and also evolution can happen so fast, I think, is because of this temporal size of these objects, and that they don't lose their history. It's all bundled up in there, so they're already causally contingent. And the causation is in the structure, and

Emily Silverman  

it's where? -If it's not in the genome, where is it? 

Sara Imari Walker  

The information is in every single physical piece of the cells. I think even the genome we don't actually have right every single piece of the genome of any length is undergoing selection all the time. But we tend to fixate on genes as the functional units and only talk about those. We don't talk about this hierarchical structure where selection is happening across every component of every length. That's easy to think of as a sequence, but now, if you think about the cell, all of these molecules that are reacting all the time, every single scale is undergoing selection from the cell itself, the tissue, higher level pattern the cell, and then the chemistry inside the cell, and then the parts of the molecules inside the molecules. So the way that we think about that is to look at the structure and then build down into the assembly space, and then think about the interactions that are actually happening across the layers of the assembly space, which is like if you could unfold a structure across time, you would have this much larger space, and then you could talk about all the interactions across that space, and that would allow you to explain more of the dynamics that you're actually observing. This is very cutting edge. We haven't really gotten this kind of work out yet, but this is where I imagine the dynamics of assembly spaces are actually going to be able to provide models for what we call agency, or how we view causation more broadly in living systems. 

Emily Silverman  

Yeah, I mean, the second example that I wrote down here is the ATP synthetase machine.

Sara Imari Walker  

Yeah, that one's crazy. 

Emily Silverman  

I remember learning about it in biochemistry, and it's this molecular motor. And I was Googling it, and I came across this article written by this guy, Jonathan McClatchy, who's like a Christian writer slash scientist, and he's trying to puzzle through how this molecular motor could have been built. And so he says one hypothesis is that a proton motor came to be associated with some kind of helicase, and the ATPase of the helicase was driven forcibly in reverse by the proton motor. And that was one option. The other scenario he talks about is maybe there was a translocating protein that got stuck in the protein trans locace, and then subsequently evolved into the central stock of the motor. But then you have to couple that stock to the ring in order for it to rotate and so on. And each example just seems more and more far fetched, and then a lot of story building. He says "it's a marvel of nano technology, an amazing molecular energy turbine. One is led unsurprisingly, to the conclusion that such a machine is best explained by intelligent design." Yeah. So what say you? How does this thing get built? Is it reaching back 3.8 billion years into history, and just like a hand, like puts it together? Or how does that happen? 

Sara Imari Walker  

I think the universe is constructing itself. So I don't find it helpful to explain what exists by assuming that there's some entity outside of our universe, or some explanation that comes from the outside the boundary of our universe, which even standard physics would put because you have to set the because you have to set the initial condition of the universe, and you have to also decide where the laws of physics come from. I don't think any of that's necessary. I think our universe is literally constructing itself. And if you want to understand why certain structures like ATPase come into existence, you actually have to look at the causal contingency about how such a structure could be built, which is like the structure of the assembly space, and you have to look for what objects could co construct with it. This is sort of, I think, the issue why everything looks intelligent in design is, if you look at any object in isolation, of course, it's like, where did that thing come from? But it had to be built by the things in its environment. And so really, what we need to do, and what we're trying to do with assembly theory is look at objects that were co constructed and actually construct what we call the joint assembly space of the construction histories, and use that as a model for how we think about explaining why these objects come into existence. And one of the key features also this goes back in the long history literature of the origin of life, is like this idea of auto catalytic sets that you need systems that one molecule can produce the next molecule can produce the next molecule, and they form closed cycles. We don't exactly think about it that way in assembly theory, but there is this idea that, like when you're moving into this combinatorial space of these incredibly intricate, complex objects, a lot of it involves trial and error and making things that just can't persist. They're not able to persist in time because there's no selection mechanism to reinforce them. What happens is there's a random search, and you get a collapse on the space of a set of objects or molecules that can reinforce their own existence. And so ATPase cannot come into existence on its own. It has to be part of this other structure where it's all self reinforcing. One way I think about it is like we're all hanging on to being able to exist by coexisting with a bunch of things that help reinforce our existence, right? You imagine the universe is like random out there, and there's this insanely large combinatorial space of possibilities. The only things that can be structured are things that can be co built and actually exist along these same lineages. So usually I describe life as lineage, as a propagating information or these sort of construction histories or trajectories we were talking about. So design emerges along those because it starts to look like things have function because they're reinforcing the structure and existence of other objects. And so I think very much a function is an emergent property of this kind of dynamics, not as fundamental, but you would expect to see things over time that look more and more designed because they have more and more history and more and more selection and more and more constraints that were ruled out in order to make that specific object. The only reason that object exists and not some other object is just because of the particular contingency in the historical pathway to get there. So we're ruling out a whole bunch of objects that will never exist, like the transistor example, like all these technologies, that will never exist because they're just not along our particular pathway. But it doesn't mean that anything in that path was designed by something outside of it was all internally constructed, and all has to be self consistent along that particular trajectory.

Emily Silverman  

I once heard you ask in an interview, and I think you were asking this question to yourself, "is there a problem that the universe has that life solves?" I'm wondering where you've landed on that you know, we have these exploding phylogenetic trees of increasing complexity and novelty. What is the problem in the universe that life is solving? 

Sara Imari Walker  

Well, one part of it is, I was told by a very prominent senior physicist, you're not supposed to ask "why" questions after I gave a lecture once, and I thought that was really hilarious, and I actually was doubly bad, because he was like, the last person to ask a question, like, this was Boltzmann, and you know, it happened to him. Oh, my God, are you kidding me? But also, you know, people have been asking, what is life for so long? This is a famous Schrodinger book, and the what is life question is just like over and over again, we're not asking, Why does life exist, or what does life do that nothing else in the universe could do? And I think those questions are actually more scientifically productive, because the what usually follows the why and the what does it do, kind of question. So I thought a lot about the nature of life and like, what are the things about reality we couldn't explain without a fundamental mechanism for life. And like, what is life doing? And the thing I've always landed on, and I had ways of talking about it before assembly theory, and I have ways of talking about it after assembly theory that I think are much more precise and concrete, but the way that I've always thought about it is life is sort of maximizing the number of possibilities that could exist. That was an early thought I had, because there's a lot of stuff that can be created on our planet that cannot be created anywhere else in the universe. And some of those things you might really like, like the Library of Babel or Shakespeare's works, and some of them you might not like Tiktok or pick your favorite political campaign. Like there's lots of good and bad things that people love or hate. I'm pretty agnostic on everything on this planet, but they can't exist anywhere else. They can only exist here. And I think this is really profound. So that was one part of it. But then as I was thinking more deeply and thinking about really, just how large the space of possibilities is, which is when I really started working on assembly and really thinking about the size of chemical space, which is just like, ridiculously huge. I mean, I already talked about this pharmaceutical drug design problem. It's like, epically huge, like, you can't train AI on things that are that big, that you don't even know how to make them think about that. You know, it really struck me that what life is doing is maximizing the number of things that get to exist. And so you think about Darwin's in the struggle for existence, every form is out competing each other. But what assembly theory gives you, that's, I think is a deeper story is there's also a competition to get to exist at all. It's not even like once you exist, you get to keep existing. You want to keep existing. Also, there's a drive to try to make more things exist. And I think this is also one of the reasons that evolved objects like us are so efficient for the universe to maximize existence, because all of those layers of evolution and history and contingency now get packed in a small volume of space, like I'm five foot three, so it's kind of tiny volume, but I represent 4 billion years of evolution packed into one physical structure. And there's a huge amount of recursive layers of structure in me, and all of that gets to coexist in one physical object. So that feature, I think is really important that the universe is actually increasing the recursivity in objects as they get deeper and deeper in time. So they're becoming much larger temporal structure, small volume of space maximize existence, but also being able to use those structures to generate more complexity and more stuff. And so I guess I would think about the universe as kind of like the universe is what exists in a local volume that we can observe. There's the Hubble volume, which is like observational boundary of our universe, and it's just trying to pack as much stuff as possible in there, because everything wants to exist. Because the other option is we don't know what non existence, but non existence is a concept that exists, so I don't actually know what it is to not exist.

Emily Silverman  

Well, you said something once about how a lot of people look up at the night sky and they feel so small, they feel like a spec and so on. But you're really flipping that idea on its head and saying, Actually, we are some of the largest structures that exist in the universe. And I really loved that. In these last couple minutes. I just wanted to ask you about aliens, because we have to people think of aliens as life that evolved off Earth, but you have a slightly different way of thinking about it. So tell us how you think about what an alien is, and then how you're going about trying to discover/ find aliens.

Sara Imari Walker  

I never liked the idea of defining life. A lot of people my field try to do it. I think assembly theory has a natural definition for alien life, and alien life is something that emerged from the random froth of chemistry along a different causal history than we did. And so the origin of life as an experimental paradigm becomes an interesting problem, because now you can talk about this huge possibility space, like this huge volume of things that the universe could create, and you can treat it like a search problem. If you could build a chemical search engine that could actually iterate through different chemical possibilities, how long would it take that search to actually find an alien life form? And so this is actually not my idea. This is Lee Cronin idea, and he's been working on building a technology for that for more than 10 years. He's a pioneer in digital chemistry, and the whole reason that he even got into that field was because he wanted to build robots that could solve the origin of life by trying to scale the technology to do chemical searches. So right now, state of the art in the chemistry lab is you have somebody pipetting individual reactions and things. And so he's basically built a technology that automates that process, and the idea is that we'll be able to scale it. And so we're collaborating on this idea of trying to build large scale, origin life experiments guided by theories. So we have Assembly theories, the abstract, fundamental formalism, and we want to build these very large, robotically driven experiments to look for aliens in the lab. It's a little bit like how particle physics has the theory coupled with massive experiments. And we started a company called chemify, which is trying to scale up the technology. It's actually a drug design company. They build the technology to build molecules, and I'm writing a book to try to get people excited about it. So that's what we're trying to do to get there.

Emily Silverman  

Should we be scared?

Sara Imari Walker  

 So it's like, no, I don't know. I'm terrified, but I don't know how to be scared of understanding things. I love this Marie Curie quote, it's like, now is the time to understand more so we can fear less. And I really feel that very deeply. And so I guess for me, when I talk about the things that we're envisioning doing and the ideas that we're excited about, a lot of people have this kind of fear response, oh my god. Like, what kind of synthetic life forms are gonna come out of these labs, and what kind of technology are you building? But really the motivation is the fundamental understanding. And I think we're so existentially traumatized by these living technologies and all these other things that we don't understand, that unless we really understand fundamentally what life is, we're not going to be able to understand how we're talking about aliens or what they are. We're not gonna be able to understand how we're talking about artificial intelligence as an artificial life or what they are. This is a really existential problem we have to solve, and I would be more afraid of walking into the future not knowing what it is that we're talking about than doing this kind of fundamental science to really understand the core of this question. And so I think there probably are absolutely technological repercussions that would be scary, but I also think that there's a lot of them that are incredibly optimistic. And ultimately, I think as humans, we build our own future. And so I want to keep choosing the optimistic one and the one that gives us the most possibilities for the future, because I think that's what life does. And so I'm going to keep trying to work in that direction. And I hope other people do too, because I think our future could be insanely bright if we want to build it that way. 

Emily Silverman  

Boom, yeah. Well, this has been so much fun for me. I've been looking forward to this for a long time, and I'm just so happy to have you on the show. And wanted to encourage the audience to please consider going out and buying Sara's amazing new book, "Life As No One Knows It." You'll learn a ton. I also recommend just checking her out on YouTube and watching some of her talks, because she's a great writer, but as you can tell from this interview, she is also just so skilled at communicating complex scientific topics to mortals like me. So thank you again, so much, Sara for coming onto the show, and I can't wait to keep tracking your work and see what you do next. 

Sara Imari Walker  

Thanks, it was a lovely conversation. Really enjoyed it.

Emily Silverman  

This episode of The Nocturnists was produced by me and Jon Oliver. Jon also edited and mixed. Our executive producer is Ali Block. Our head of story development is Molly Rose-Williams and Ashley Pettit is our program manager. Original theme music was composed by Yosef Munro, and additional music comes from Blue Dot sessions. The Nocturnist is made possible by the California Medical Association, a physician led organization that works tirelessly to make sure that the doctor patient relationship remains at the center of medicine. To learn more about the CMA, visit CMAdocs.org, The Nocturnists is also made possible by donations from listeners like you. Thank you so much for supporting our work in storytelling. If you enjoyed this episode, please, Like, Share, Subscribe and help others find us by giving us a rating and review in your favorite podcast app, to contribute your voice to an upcoming project or to make a donation, visit our website at the nocturnist.org. I'm your host, Emily Silverman, see you next week. 

Transcript

Note: The Nocturnists is created primarily as a listening experience. The audio contains emotion, emphasis, and soundscapes that are not easily transcribed. We encourage you to listen to the episode if at all possible. Our transcripts are produced using both speech recognition software and human copy editors, and may not be 100% accurate. Thank you for consulting the audio before quoting in print.


Emily Silverman  

You're listening to The Nocturnists: Conversations. I'm Emily Silverman. Today we're wrestling with one of the oldest questions in science: what exactly is life? For years, we've struggled to answer this question. We've said maybe living things are defined by reproduction or compartments or metabolism or being a self sustaining system, but it seems like there's always an exception to the rule. We either rule something in that clearly isn't life, like a car engine that metabolizes gasoline, or we rule something out that clearly is alive, like a mule, which is sterile and can't reproduce. It seems, no matter what box we try to draw around life, it breaks free from our definitions. But our guest today, astrobiologist and theoretical physicist Sara Walker wonders if maybe it's less important to define what life is and more important to define what life does. This is at the heart of her work on assembly theory, which we talk about in this episode. Sara is the Deputy Director of the Beyond Center for Fundamental Concepts in Science and a professor in the School of Earth and Space Exploration at Arizona State University. She's a fellow at the Berggruen Institute and member of the external faculty at the Santa Fe Institute. Her lab is internationally regarded as one of the leading labs building fundamental theories about life, and she recently wrote a book called "Life As No One Knows It: The Physics of Life's Emergence. I stumbled upon Sara on YouTube a few months ago and could not stop watching her talk. She's one of those trans-disciplinary polymaths who can speak about everything from physics to chemistry to biology to linguistics to artificial intelligence, dissolving the lines that humans have placed on reality and always seeking to unite, reduce, and simplify. I was very honored to have the opportunity to speak with her today about why traditional physics fails to describe life, the tenets of her assembly theory, the way she thinks about that magical phase change from a random chemical soup to life and its markers, her vision of life as a force that propagates through time as an exploding phylogenetic tree, her belief that human beings are actually some of the largest objects in the universe, containing 4 billion years of history and information all curled up inside and so much more. It's a dense conversation, but a fascinating one. So grab your coffee, buckle your seat belt, and enjoy this episode, which begins with Sara reading a short excerpt from her book, "Life As No One Knows It."

Sara Imari Walker  

The paradox of defining life. Look around you right now. I'll bet you can confidently catalog what is living and what is not. I could quiz a four year old and they would probably corroborate your classification. In fact, small children seem rather good at telling the difference between living things and inanimate objects with little explicit direction. Once a child is taught that plants are alive, they can readily tell those apart from non living things and act accordingly. For example, not stomping on the flower bed, but perhaps jumping in the mud, instead. Kids can also misclassify life by assuming inanimate objects like a favorite stuffed animal or life. Are they wrong in doing this, or are we wrong in correcting them?

Emily Silverman  

I am sitting here with Sara Walker. Sara, thank you so much for being here.

Sara Imari Walker  

Thanks for having me. I'm really excited about our conversation.

Emily Silverman  

I discovered you on YouTube, actually, before I knew that you had a book coming out and just listening to you talk, my face kind of melted off, to be honest, from how brilliant you are, not just the content of your work, which changed how I look at the world, but also the way that you're able to communicate that content. It's obviously very high level, complex stuff, but you just have such an ease as a science communicator that I was really excited to bring you on and have you share some of your work with The Nocturnists audience. So let's start with your background. You are a theoretical physicist, and you study the origin of life, which some people probably hear that and associate more with biology than with physics. So how does a physicist come to study the origin of life? 

Sara Imari Walker  

My first response to that was it was accidental, but I don't think it really was accidental. I started my university training and community college, and I was really interested in science, but I didn't know what I wanted to do, and my parents hadn't gone to college or anything, so I just took all the science classes, but I absolutely fell in love with physics. I just thought it was a phenomenal subject, mostly because I love this idea that we can come to deep, fundamental understanding of the nature of reality by thinking about it and by coupling our own thoughts with experiments and what we can test. And I really thought I was going to be a particle physicist or a cosmologist, but when I got to graduate school, my PhD advisor was studying the origin of life or getting interested in it, he was a cosmologist also. And I kind of reluctantly at first, actually started working on the problem, because I was like, this is not what physicists do, exactly what you're saying. But as I thought about that problem more and read more and read more and more of the literature, I realized that it was an incredibly open problem, and there was a lot of room for creativity. And I really felt that the training that theoretical physics provides, and in particular, like what I was really passionate about, which is deep explanations and discovering something new, like origin of life was wide open territory, in part because it doesn't really fit in any of our traditional disciplines, and we don't really have a right way of asking that question. So it seemed like there was a lot of room for the kind of science I wanted to do.

Emily Silverman  

Yeah, and I've heard you say this in interviews before that when people traditionally think of physics, they think of math equations. So probably a lot of people listening to this have taken basic college level physics. I think about Maxwell's equations, or I think about Newton's equations. And basically the way it's set up is you have this fixed law, and then you plug in a set of initial conditions, and it tells you what's going to happen to the apple as it falls from the tree to the ground, for example. But you've said that laws like that don't really work to explain life, and that, in fact, what we have to do is invent a new physics for that. So tell us about that. 

Sara Imari Walker  

Yeah, I have very fond memories of my physics classes, but I have some harrowing ones too. I think we all go through that. It's really interesting because the way that physics works. We're taught about disciplines, and we're taught like, if you're interested in life, go get a biology degree. If you're interested in stars and planets or mechanics or things, go get a physics degree. If you're interested in molecules, go get a chemistry degree. But obviously, Nature doesn't have any of those distinctions. And why I became a physicist was not necessarily because I was interested in a particular question. It was because I really was thinking about this idea that the human mind was so incredibly creative to come up with really abstract and universal understanding of the world. So to me, that's what physics is. And so this example that you're bringing up is that once you have the initial condition and law of motion, you've described the system for all time, and that comes from the era of Galileo and Newton, because they were trying to describe motion and gravitational physics in particular. And that kind of dynamic works very well for those kind of systems, because they have very regular behavior over time. But life does something very different. And I really love this very famous quote from Charles Darwin about "endless forms most beautiful keep evolving and continue to evolve." And he juxtaposes it to the fixed law of gravitation. So apparently he was a huge Newton fan, and I think he was hoping there would be like laws for evolution, like there were for gravity and motion. But he also recognized this very stark contrast, that biological forms seem to be changing all the time, and they seem to change in a way that is historical and path dependent, and this is not what we see in typical physical systems. They usually have the same kinds of rules and the same kind of behavior, no matter when you observe them. And life forms seem to have new rules and new behaviors emerging all the time. And biological evolution over geological timescales is probably the grandest stage for observing that. But even if you observe a cell, you know, it's a complex machine with all kinds of dynamical, non linear processes happening. And so it's been a real challenge for the standard conception of physics in terms of this initial condition and fixed law to describe it. And what I spent most of my career working on is thinking about, how do we actually understand a physics that has these kind of path dependent properties that we see in living systems.

Emily Silverman  

And one of the theories that you're working on is called assembly theory. And you talk in the book about how atoms are countable, we can see them on the periodic table, but once you level up to molecules, the space of possible combinations is huge, and the universe doesn't have the time or the resources to build everything that can theoretically be built. So you and your collaborator, Lee Cronin have developed this theory called assembly theory, which explains why certain things get to exist. So tell us about assembly theory, and particularly what you just said earlier about this idea of path dependence or historical contingency.

Sara Imari Walker  

The idea of path dependence is a nice segue into the example that you've given. Some of these approaches and ideas that I'm talking about seem very deep and abstract and theoretical, but one of the places that we're finding actual utility for them is in drug design, because if you imagine the combinatorial space of molecules that might actually have a biological function, the space of molecules is huge, and then finding that subset that might interact with our biochemistry in just the right way to get the right kind of health outcome is really challenging. There's a huge industry of people trying to tackle that problem. But the way that we think about this combinatorial space, from the perspective of thinking about origins of life, is there's a huge amount of possible molecules that could exist, and the way that you actually start to produce them is you have molecules that have been selected in a particular environment, and they scaffold into other structures. And so assembly theory captures this in a really rigorous way. And I'll explain a little bit about what our observables are and how we think about the theory, which is really related to the idea of, how do you detect life in chemistry, like if you look at a molecule, how do you know it's a product of life or not? But then I'll circle back to the question about the path dependency, and how this relates to a very non-Newtonian kind of physics, or at least not in the standard paradigm of physics. So the basics of the theory are, we have a conjecture that life is the only mechanism the universe has for generating complex objects. And complex objects are things like complex molecules, like DNA is obviously an example, or ATP even so, things that we find in a cell we don't find spontaneously produced in the environment of a planet like no one's observed these outside of biology. And you could take for granted that the universe could just produce those objects outside of life, but actually a key conjecture of assembly theory is that that's not possible. And that's one of the things that we actually aiming to formalize to test that you only get complexity as part of a living process. And the way that we formalize that, it's almost like thinking about Lego blocks and building up a LEGO Universe. Or if you play Minecraft, you have all the pieces you can put together to make all these complex structures. But we start with thinking about bonds, and how bonds are actually the fundamental unit of building molecules. So if you prefer the Lego analogy, you can think about your Lego blocks, and you can stick them together, and then you can reuse structures you've already built. And let's say that we have a target structure, like maybe we want to make the molecule ATP, or maybe want to make Hogwarts Lego Castle. A couple examples of complex objects in your mind. So you can take this idea of trying to build an object by putting pieces together and then taking those pieces and reusing them and get up to the final structure. And what we try to look for is actually the shortest pathway to do that, which we call the assembly index. So there's sort of a minimal constructed complexity for making the object, or a minimal set of constraints that's necessary minimal history to produce that particular structure. And then we also look for that structure in high abundance. So the idea being on our planet, there's not one Lego Hogwarts Castle, there's millions of boxes, potentially of them, and patterns on the planet so that they're repeatable structures. That's because they've actually been selected by the environment, by humans, by kids that are really big fans of Harry Potter to be a really highly abundant object on our planet. Whereas, if you took the equivalent amount of Legos, I think there's like a couple 1000 pieces in there, all the random structures that you can make out of that, that's a huge space, but most of those structures will never actually be built ever in the history of the universe. It's actually probably the number of structures is so large the universe doesn't have enough time, even if you were putting Lego blocks together every second, to exhaust that space. So we have the assembly index and the copy number as the two key variables that we're looking at, and we view those together as a signature of constructed complexity. Basically what life does. Life is the mechanism the universe has for making complex objects. And it's important that we have those two things, because we can actually go in the lab and observe them and measure them. So the way we do it with the assembly index's shortest path, is using mass spectrometry. And it turns out that with a mass spec, you can probe that feature of a molecule. And if you're doing mass spec, you already have to have thousands of copies of a molecule even to recognize that it's there and measure it just with standard laboratory equipment. And so what we were able to do was actually go in the lab, and this was an experiment done in Lee Cronin's lab. I think this was just such a cool experiment. What they did was actually take living and non living samples, and they figured out how to measure assembly in those samples, and they were able to show that above about an assembly index of 15. So thinking 15 steps where you reuse parts and build an object in chemical space. So you're using bonds to form these structures. About above 15, you only see molecules that were produced from the Living samples, and there were even blinded samples from NASA that were supposed to be confounding ones. It was like Murchison meteorite, which is like one of the most complex examples of chemistry in our solar system. And it didn't pass the test of producing high assembly molecules, like if you just have random structure making molecules, you can make all of the very simple molecules pretty easily. But what we see in prebiotic systems,when you don't have selection and you don't have evolution, or you don't have constraints, you just get tar, which is an undifferentiated goop of all of the kind of simple molecules. You don't get high abundance of very refined molecular structures like what we see in biochemistry. And so we think that unless you have selection and evolution, you can't cross a complexity threshold, which experimentally seems to be about 15 steps. That might not be a universal number. It's experimentally verified, but we have a lot of work to do, but we conjecture that there's always a threshold, and above that, you have to have a path dependent trajectory to make those objects, because they're just so complex, and the number of errors that you could have making the wrong part, and bifurcating off into all the other possibilities that could be built is so large that you would never expect that structure to exist unless something selected for it to exist, and it was built by an Information Processing System, or something that we might call alive. So that's sort of the key features of assembly theory, and why it's very different than that standard Newtonian paradigm is the universe of possibilities gets larger with every single step that you take trying to build a new object, because every time you could have put some other things together differently, and so you have to trace this path dependent trajectory out. And in a Newtonian world, the universe never gets bigger as a function of the complexity time always has the same size. But what we see in assembly theory is this notion of this combinatorial space that living things are moving through. And as it gets more complex, the space gets larger and larger, and that actually informs a lot of the foundational physics that we're thinking about and the philosophy that I'm most interested in.

Emily Silverman  

So the idea is that if you look at a molecule like Taxol, for example. I know as an example that you use a lot Taxol is, I believe, a molecule that was originally found in plants, yes, or a molecule like ATP, or even a giant molecule like a ribosome. It's like, once you get ATP and once you get the ribosome, all sorts of things become possible. How do you think about these stop gaps? I've heard you compare the ribosome kind of like a transistor, which is a piece of technology that, once that was invented, it was like, Okay, now we have the transistor. We can build all these other machines. Or once the light bulb was invented, it's like, Okay, now we have the light bulb. We can make all of these different types of lights. Do you think about those moments as special, where you create something that blooms out into more possibilities?

Sara Imari Walker  

I think that's just the way that our universe works, and it's the only way it can make complex objects. When we're thinking about life evolution, at least, we all have a picture in our mind of these branching phylogenetic trees that are constructed from genomic information, right? And before that, people were trying to do it by morphological similarity. So we've had this kind of tree structure in the way that we've been visualizing the nature of life for at least 150 years, since the time of Darwin. Well more than that now. And some people will be like, Well, what do you need assembly theory for we already have good enough evolutionary biology, but the point of assembly theory is actually try to look at selection and evolution before you have things like genomes, because we're trying to explain the origin of those things. So we need deeper evolutionary principles than what biological architectures provide us, because we're trying to explain the origin of things like the cell, not use the cell as a scaffold for talking about further complexity. And so what we see looking at assembly theory is, if you think about the universe as this giant space of possibilities that could exist, selection has to mediate these pathways to build up the next sort of scaffold. So I do like the example of transistors and the point that I would make there, because technology is very perceptible to us, is that most of our modern technologies are built on the transistor, because transistors have been a very prevalent component of a lot of technology. So now we have factories that can make them, and they make devices compatible with each other. And so it's selected out a whole bunch of other options that might have been equivalent or even better than transistors in terms of efficiency and all other kinds of technology, but we're kind of locked into using transistors now because they're so prevalent and we're on its trajectory now where transistors are going to be part of our future technology. And a similar transition happened very early in the evolution of life on Earth, with the invention of the ribosome. Evolution had to construct the ribosome. Once the ribosome and the rest of the translation machinery was constructed by evolution, it became solidified as a structure on our planet that could elaborate further complexity and kind of ruled out the other possibilities that maybe other kinds of cellular architectures or things could have evolved at the same period. But this became the one that was successful, and now we find it in all life on Earth. So we have a single branching point for life. But I think what assembly theory is doing is making that branching structure go deeper. And so this is how we're thinking about the nature of the actual structure of physical objects, and that they're all related to each other through this historical contingency of this reuse of parts and the causation. Propagating through all these objects.

Emily Silverman  

You wrote an article in Eon magazine called "Life is Not Alive," where you say that things like cats and dogs are alive, and things like sofas and rockets are life, and the reason that sofas and rockets are life is you don't just randomly find a sofa in the universe that you need life around to invent and build a sofa. And like you said, there's a high copy number. We see a sofa here and a sofa there, and there's green sofas and black sofas and velvet sofas. So I never really thought of it that way. I always thought of it as animate and inanimate, but you're bringing in this new category, which is inanimate matter that is life, because life needed to be around to create it. So talk a bit more about those distinctions and why they're so important. 

Sara Imari Walker  

I'm a professional astrobiologist. So the two problems that drive my career really the origin of life. How is the transition from non living to living matter, and then the other one being life detection on other worlds. And if you're thinking about detecting life, you don't necessarily care if you're finding an extant organism, something that's alive. Now in the Mars community, a lot of the focus is on finding fossils, or fossil evidence of life. But I think most of us would be convinced if we found a cell phone on Mars, it would be evidence that life was there. Like, we go find a robot, it's probably a human robot that was put there, not that geochemistry of Mars spontaneously fluctuated curiosity on the surface right, like it wouldn't be explanatory. It's probably the case that some intelligent planet nearby actually built the technology to put it there. So we recognize in our narratives already that there are these certain structures that must be constructed by humans or some human-like physical system, something with comparable intelligence, or are necessarily the product of evolution. That's kind of the idea of a bio signature. And so when we were thinking about this fundamental story about the physics of life, and thinking about the combinatorial universe of chemistry and all the structures that could be built out of it. Chemistry, at least, is iterable, right? So I like thinking about chemistry because you can talk about the atoms in the periodic table and the combinations, and maybe estimate the size of a chemical space using chnops, carbon, hydrogen, oxygen, phosphorus, nitrogen, sulfur, right? Like the standard biological elements, we can kind of imagine sticking them together by bonding rules and estimating how many structures there are. We don't know how to estimate how many kinds of life forms are out there, right? Like how many cell like structures, if you think about the Cambrian explosion and how many body plans were made, it could have been exponentially more than that was probably just what the resources and physical environment imposed. But most of those aren't around anymore. And if we think about all possible technologies, that's an even larger space. So all of these combinatorial spaces are quite large. And in assembly theory, we always think that there's this boundary that needs to be crossed, that only evolution can do the process we call evolution, but thinking about it much more deeply in terms of this historical contingency and the construction process, not as a passive filter. It's a very active mechanism of selection, selecting the next thing that the universe can construct. When you're thinking about that kind of structure, it's very clear that you have to have kind of an abrupt transition point, and when you cross that transition point, those objects can only exist in the presence of an evolutionary process or selection, because the space of possibilities they live in is so large that the universe could not possibly do an exhaustive search to discover them. It would have to do a directed, constructed process. And so we kind of differentiate all of the objects above that that are impossible to form randomly and require information for their construction to be life, anything in that space. And then alive to me is the actual active mechanism of building a trajectory into that space. So something like us is alive because we're creative in the universe, we're generating new possibilities. But things like water bottles, like I have on my desk that are artifacts of life, they're still life. Like bottles don't spontaneously come into the universe outside of things like us to generate them. So we have a causal pathway like this thing has to be built before this can be built, before this can be built. So objects actually necessitate their existence based on other structures that have been selected in the evolution of a biosphere, but it's the active process of building that's alive. So in some sense, it's like, is it the process or the thing? And we're really trying to merge both of them into the same story, but this is where you can have a more holistic view about the nature of the physics of life, really, just thinking about life is everything that's complex, too complex for the universe to generate spontaneously and requires some kind of construction and informational constraints. And then it's much more perplexing and interesting to find things like us that are alive. If we want to go look for aliens, probably we want to find alive ones and not dead ones. We had this joke in that essay, actually, about Schrodinger's cat. And I love this because, you know, it's a very physics example about is the cat alive or dead? But in assembly theory, in the way we think about life, the cat can be dead but still be life. It's not alive anymore, right? But it's, you know, a dead cat is a product of evolution in the same sense that a living cat is. They don't spontaneously form in the universe either. So you're kind of talking about two different levels of description of the same physics, but that's kind of the idea behind it.

Emily Silverman  

Yeah, and you joked about this in the talk that you gave at Mindfest, which I watched and which we'll link in the show notes, where you talk about this thought experiment of the Boltzmann brain. And so the idea for the audience is that human brains exist in the universe because we all have a human brain, therefore it is possible for a human brain to exist in the universe. Therefore, why shouldn't a human brain spontaneously fluctuate into existence? Because physics theoretically allows it. But then you say, and I quote, "People really intrinsically think the universe can generate anything, anywhere, for free. Where is all that information? Please tell me." And so where is the information? Like, where? Where is the locus of the information?

Sara Imari Walker  

A lot of my work is trying to build what seem like counterintuitive notions, but they're actually very intuitive, if you're willing to sort of break your training and how you've been taught to think. Why I was joking about that is because I think our standard education and the way that we talk about the world has, right now, at this moment in history, really trained us to think that any object can fluctuate into existence. It's very standard in physics, and I learned it in many physics classes that there was just a low probability for thermodynamic fluctuations or quantum fluctuations, out of the vacuum to produce objects of arbitrary complexity, just because it's written in the laws of physics as we understand them. Yet we have no observational evidence for that. It drives me mad, because I think about it this way now. But like, obviously I didn't think about it this way most of my career, like when I was a student, stuff, I never would have thought about this way. But it's like, seems so obvious to me, is like, in order to have any object fluctuate into existence at any point in space and time be a possibility, it means, like, everywhere has to have the information for every object, which means you have to have an infinite density of information in our universe at every point. And that's just totally unphysical. So we have, like, these really unphysical things just baked into the way that we talk about the world, and because they seem so obvious to everyone that that's the way the world works, nobody takes a step back and questions these basic assumptions. And I think this is actually really the art of theoretical physics. Like most of the major advances in theoretical physics have been someone that's willing to sit down and question the most basic assumption that everyone assumes is true. It's like Einstein with like, the constancy of the speed of light, right? That's like, a very obvious example, because even with the Michelson Morley experiments that validated experimentally with high precision that the speed of light was constant independent of the motion of the Earth, people still didn't want to accept it. They assumed the experiment was wrong. And Einstein was like, I'm just gonna assume it's right and just run with it, and we get relativity right. So it's kind of interesting, like there's just these things that we just assume. So that's one of them, for me, that I find kind of mind boggling now, is that the idea of Boltzmann brains is taken seriously in some of the physics community, and it's to the point like in my history as a cosmologist, like some people reason their cosmological models on the idea of trying to rule out Boltzmann brains being more prevalent than we are, because once you assume a brain can fluctuate into existence, it's very likely that any observer is a Boltzmann brain over cosmological timescales, instead of being something like us that was evolved on a planet over billions of years, and assembly theory says something very firm like that is not possible. The only way to get to something like us is through the process of evolution and construction, and to get something like us means we have to come with a whole bunch of other structures like us. So we emerge in populations and things. And it's like, very consistent with what we actually observe the world to be like.

Emily Silverman  

Right before, you need a human brain, you need a human body, you need a ribosome. You need it's like the Carl Sagan quote, where he says, "How do you bake an apple pie? First you have to invent the universe." 

Sara Imari Walker  

That's right, and then you have to go through all those evolutionary steps to get one right. Yeah, that's right. It's not just in the initial condition.

Emily Silverman  

I'm wondering if assembly theory could apply to abstract objects as well. So for example, this thought experiment of could a random letter generator left to operate for an infinite amount of time write the book of Hamlet for example. My husband and I actually have this ongoing fight about this question. 

Sara Imari Walker  

Oh, really?

Emily Silverman  

Yeah, you know, and I always felt like, yeah, maybe a random letter generator could produce a few words or some word fragments, but it's not going to write Hamlet. For it to write Hamlet, you need something else. You need consciousness or intent or something like that. And he would always argue like, no, no, you don't understand what infinity means. Like, if you let it run for infinity, by definition, it'll write Hamlet. And before you respond to that, I just wanted to share one more example that's tied to language, which is this short story The Library of Babel by an Argentinian writer, Jose Luis Borges. And the story describes a universe that consists of an infinitely large library filled with books, and the books are filled with random letters, and the librarians are driven to suicidal despair because they know that the library must contain every book that has ever been written, and yet they're surrounded by gibberish. And if they ever stumble upon a book that has a sentence in it, they kneel in worship. And so I was wondering if something like the assembly index threshold could apply to language too. Could you say that after 15 letters, it stops being a coincidence, and there's somebody writing?

Sara Imari Walker  

Yeah, so I don't think the value of 15 will be meaningful for language, because people start to assign a very particular value to 15 and like it becomes like the number 42. It's not the answer to life, the universe and everything. It's an empirical validation of this idea of a threshold. We don't know where the threshold is going to be in most systems, and that's actually something that I'm actively working on, is like a general theory of the idea of like a phase transition associated with the origin of life, and why does it happen at a particular complexity value in assembly theory? But to your point, yes, I think the ideas apply to language as much as they do to molecules. And for a long time, I had been writing about the nature of mathematics as a physical system, which is like, kind of a really weird way to think about it. But the reason this came about is because I was trying to think about, like, how could I really point out the discrepancy that we were talking about before, about, like, what physics says and what we need to explain life. And everybody thinks physics as we understand it now is right and, like, the ultimate explanation. So like, how do you build a thought experiment or something that turns the narrative for people to see that there's, like, a totally different way of looking at it? So I started doing these thought experiments on the theoretical physics of theoretical physicists as I was thinking about in my own head, but really thinking about, like, you know, the laws of physics themselves as physical objects that have merged in our biosphere like Newton law of gravitation obviously describes something we think is objective about the world, which is like gravity, but it's also an equation we write down on paper, and that equation has causal power, in the sense that we can teach it to undergrad physics students, and they can roll balls down inclined planes and test its properties, but it also allows us to launch satellites into space and do a myriad of other things that wouldn't be possible without that knowledge and that kind of small information packet. And so to me, that's a physical structure that exists on our planet, and I was really trying to elucidate that. So the reason I'm prefacing this and talking about language is we think about language as being abstract, but it must have some kind of physicality to it, because language actually influences the physical structure of reality. It changes the way we perceive reality, but also through our actions, we actually change the nature of what exists and what gets to happen. I actually even have an example in my book, which is in the middle of the book about trying to prompt people to turn the page is like a physical act, right? Like a word actually activates you to turn a page of a book. Like, how is that possible? That's like a really weird sort of thing.

Emily Silverman  

And in your MindFest talk, you take a moment and you say, "Everybody raise your hand" and then everyone in the room raises their hand, yeah? And so that's a physical act.

Sara Imari Walker  

aYeah, so I think we have evidence all around us. Yeah, that language has some physicality to it. As far as applying assembly theory, it's really non trivial, because, you know, I talked about this idea of assembly index and copy number, and they're not really separable. We can calculate the complexity of this molecule and count the number of copies. It's pretty easy for molecules, but when you move to other systems, like minerals or language that don't necessarily have natural boundaries for language. Is the boundary a word? Is it a meaningful segment of a sentence? Is it a sentence? Is it a paragraph? This is part of the foundation of trying to understand language, and the machine learning is that, like the things that actually become tokens are not necessarily the things that we prescribe meaning to, as far as the predictability of language. But the point there is just that we actually need to start thinking about language as a physical system. And one of the other tenants we have of assembly theory is if you're going to talk about assembly index and copy number, you have to be able to measure them as part of the systems. We have to think about what are interactional units with a language that we interact with. And so I do have work in my lab right now, actually trying to apply assembly theory to language, but applying the principles of the abstract theoretical physics to language is very non trivial, because we're really trying to think about language as a physical system. But I do think to your example about the Library of Babel, that there would be a threshold, and you would be able to say that this thing is definitively constructed by a meaning making system that was reusing elements that had specific associations with other ones, and that the structure of a language would emerge. Another thing that happened with large language models is that people were starting to realize that the structure of almost all human language is identical. If you remove the actual differences about like what words are used for what in different languages, there's this like topology to the structure of human language, about the associations between concepts that we build, that seems to be pretty universal, and I think assembly theory would be quite good at picking up those causal structures underlying language, which are the structure of association. So even when I use language, I don't really think about the meanings of words. I always think about structural associations, which is super important for my line of work, actually, because there are no predefined concepts. So I have to always know that the meanings of the words that I'm using are not the meanings that I intend, because we don't have words for these concepts yet. So I'm kind of playing around with the relations of words like you probably noticed already in this conversation. I'll interchange complexity, assembly, information, evolution, causation, constraint. And to me, they all mean the same thing. I know it's like dizzy,

Emily Silverman  

Yeah, consciousness, meaning making system. Yeah, yeah. 

Sara Imari Walker  

You have to kind of play around with the words. This is part of the thing about effective communication. When you do that, enough people start to build their own associations in their mind, and then they can start to see the picture of what you're talking about. So I think this is really important to communicating deep ideas effectively. Is not to use the same words all the time. Yeah,

Emily Silverman  

I love the idea of a theoretical physics lab collaborating with a team of linguists. I mean, to me, that is just the perfect example of how your work jumps from discipline to discipline. It cannot stay confined inside physics or chemistry or biology, and that's part of what makes your work so exciting to me. I wanted to ask about you mentioned large language models, and you talk a lot about technology, actually, in your book. And there's a couple questions I want to ask you, but maybe first I'll ask you about just how you see technology. You talk about how Earth has developed a biosphere, and that biosphere has now invented a technosphere.

Sara Imari Walker  

I think sometimes about what our biosphere is doing is unfolding structure over time, and so in some sense, and I actually was thinking about this quite a lot this morning, about the simulation argument, where people think reality is literally simulation, and it's so interesting that we can viscerally feel that right now, because we're seeing all of these objects coming into our environment that were once just ideas in human minds. And so pretty much everything in our environment is constructed nowadays by a human mind. I have a microphone sitting in front of me, a board behind me with a whole bunch of scribbles on it that were iterated by a human evolution took 4 billion years to construct something like us, and all of that selection and all of that historical contingency, of all those steps along the way are actually embedded in us as physical structures. So one of the concepts I've taken from Assembly theory, which I think is a direct consequence of how we measure attributes of the theory, is that evolved objects actually have a size and time, and that size and time, the more complex the object, the deeper in time that object is. So I would think of something like myself or you as an object that takes 4 billion years of history to be able to create you, but that's all rolled up recursively through this process of reusing parts like that stack of structure is you. Then we go out in the world and we create things. We're putting all of that history into these structures. So like the world is in some sense becoming more life, maybe not yet alive around us in the technology that we're creating, because it required so much evolution and so much packing in of all that selection and all of that time and structures like us to generate these things in the first place. So technology, to me, is just a continuation of the process of life on Earth. I think the more interesting question there, and the more interesting transition is, when do we say that our technology itself is alive too, in the sense that technology becomes a creative force in the universe, in the way that we are, it has causal power of its own, if you want to talk about it that way. And I don't know that that we're there yet with any individual device or technology, including large language models. I don't really see them as being incredibly creative beyond what has been put in, as far as what human language is already capable of. But I do think if you look at like the global ecosystem between technology and humans and like what we're creating, that that whole system is in some sense living like. Social systems are living, and they exist at different scales. So one of the other things that I think about quite a lot is how life is a process that recurs across scales, and what's the role of technology in continuing that process, and thinking about technology as a social scale or a cultural phenomenon. So I, like a lot of the discussion now, thinking about LLMs (large language models) as cultural technologies and these kind of things.

Emily Silverman  

Yeah, because there is this interesting thing that happens with Life where I'm a human being. I'm sitting here, I have a ton of cells in my body. I have a dendritic cell in my immune system, and that cell, you could argue, is alive, because it's constantly sampling its environment. It's making decisions about whether to travel to the lymph node to present the antigen and things like that. But that dendritic cell probably is not aware of Emily, like it is a living thing inside, a living thing that has a consciousness that maybe that one cell is not aware of. So do you think it's possible we could evolve in a direction where you and I are like the dendritic cells, and then God knows what, the new GPT, or whatever is the Emily. And maybe we're not even aware that it exists, and we're just kind of living our lives, swimming around. But there's like a new level.

Sara Imari Walker  

I think we're already kind of there. I mean, I mean, I think that's what human, social and cultural systems are. And a lot of people that work on complex systems really think about them as living, evolving entities. Like culture itself clearly has evolutionary properties, and it clearly controls the behavior of individual actors that are part of that social system. So if you think about like the cellular contract, cells form together to make tissues, and they're not going to apoptose unless there's some traumatic event that happens, or why would you even have programed cell deaths, unless you were a part of a society that needed to have that function. So I think societies already have some of that structure, but I think certainly what I view with modern technologies is they're becoming the informational architecture of these social entities. So something like a large language model cannot come into existence until you have a society that evolves language and then evolves technology to record vast volumes of that language in order to even train these models. And so the large language model is scaffolded on all of this human evolution, but where it scaffold is actually at the societal layer, not the individual. And so I think this is one of the reasons that they're deeply perplexing to us, because when we interact with them, and this is why it was such an epistemic shock for so many people the first time that they ever interacted with one is like you're taking something that has traditionally been part of a social system distributed across many human minds, language and the dynamic of that and its evolution has always been, we're the only substrate, and then you train a model on it, and you put it in a box, and then suddenly you can interact with that as like a dynamic. It's totally crazy, but I loved when that was coming out. A lot of people were pointing to how in ancient Greece, people were trying to not have people learn how to read and write, because they were afraid of the dead talking, and they thought this was a really dangerous technology, because the dead should not speak to the living, so you should not record what you've said as written text. So in the past, we've had ghosts. The Dead could talk through books over hundreds of years, and now it's become a process where now the dead have much more of a live interactive voice, because you can obviously train someone, if they have enough text outputs from their own vocalizations, you can train a digital twin of them. So now it's like the ghost has gotten two dimensional instead of one dimensional. But I think about it a little bit like how the genome had to evolve in early life, in some sense, because if you're thinking about a dynamic chemical system that has no record of its past history. It has to start building record keeping devices. This is where the genome and the translation machinery came really important, because they become the memory of the cell, or at least one of the ways that it can store it really robustly. And obviously, human language has evolved to allow us to take the dynamic of human cultural systems and store it robustly and allow some history to actually enter the modern cultural moment. Because we've recorded our history. We don't record it accurately, but we've recorded some of it like some of it's lost, and just like cellular architectures like that, process became much more dynamic over time. So if you look at all the editing functions of a eukaryotic cell versus a prokaryotic cell, the genome of a eukaryotic cell is a highly dynamic system. It's crazy with all the epigenetic modifications and all the other stuff, like how much editable information is there, so that becomes much more like a large language model to a book. These systems are just evolving, but there's social and cultural technologies. I think the way that we should think about these technologies is a progression of what life has been doing over billions of years. But as you're saying, we're not the Master and Commander of the ship. We're not the largest scale, most complex thing on the planet, and we probably haven't been for most of human evolution, at least since we became social and agricultural. But now we're really just seeing very strong evidence of that directly presented to us. So we might be a little bit different than that dendritic cell, because we have a bit more self awareness of it happening, and we're having this conversation. But whether our conversation of it is also accurate to like, what these higher scale systems that we don't know exactly how to talk about are doing is also another question, which I always worry about, but I think we're doing the best we can.

Emily Silverman  

I want to ask you about the fate of life, or the fate of the universe, you mentioned earlier, this idea of a phylogenetic tree. So you have this explosion of different branches of the tree. You have amphibians, you have reptiles, you have mammals, and it's just branching, branching, branching forward, and life is continuing to generate more and more complexity, until question mark. And there's this idea in your book of the complexity hump, kind of like mixing together milk and coffee. So you have your cup of coffee, you pour the milk in, and that's kind of like the Big Bang, and then you get these beautiful swirling, complex patterns, which is, I guess, kind of like complex things existing in the universe, and then, because of entropy, things ultimately blend into a uniform light brown in your coffee cup. So is it possible that this exploding, branching phylogenetic tree could ever stop diverging and start converging back into oneness?

Sara Imari Walker  

The idea of this complexity hump. It comes from standard physics that if you want to get to a complex state, you need to have a low entropy initial condition. So low entropy would be like the cream separate from your coffee, because it requires someone to have filtered the two to be separate. So that's like a highly ordered state. And then when you mix them, there's this disorder that emerges because they're starting to swirl together, but you still can tell your cream is separate from the coffee. And so an observer would need a lot of information to specify where those exact atoms are. And then when you get to the warm, soft, brown, well mixed state, that's the high entropy state, because at that point to describe that system, you don't need to have any precision of knowing where any of the molecules are, because most of the configurations are just going to have them be random, and therefore you would get something that looked like it was well mixed. So that's sort of the standard description of how you get complexity. Complexity can be these intermediate states from between order and ultimate disorder. And so because of the second law of thermodynamics, and we think that things are always trending and toward disorder, that must be the way that it went. And so this requires fine tuning of the initial conditions of the universe to be low entropy, which is a big problem in cosmology, but also thinking about the nature of life, and is our universe fine tuned for life. So I find all of that discussion really problematic in a lot of ways, because you actually need a human to have the cream separate from the coffee and mix them to begin with. So like these, analogies are only so good as you actually really carefully consider all the assumptions you're putting into them. Which goes back to this idea we were talking about, about spontaneous fluctuation of brains. If you really spend some time with that idea, you can come to some pretty basic reasons why it's ridiculous, and we should make sure our laws of physics don't permit that, because it's ludicrous, instead of assuming that it must be true, because the laws of physics say and so the laws of physics right now say that entropy is increasing all the time, and that this is the narrative as heretical as a theoretical physicist I am, I don't really accept the second law of thermodynamics as like God given truth about the way the universe works. So in so many physics classes, there's a quote from Eddington that they call you should die in shame if your experiment or theory goes against the second law of thermodynamics. That's the sentiment of the quote. I'd have to pull the quote to actually remember it. But when you learn thermodynamics, you're taught this, you should not question the second law. You're be over the head with a ruler. Do not question the second law. And it's really perplexing to me, because as I was starting to think more about life. One of the big paradoxes is it seems to violate the second law, because it's building complexity and kind of a sustained way. And so the idea was it's consistent with the laws of physics, because this idea of fluctuation, you can always have complexity in the middle, but the universe in the long term future is going to be a boring place. I just don't think that's right, actually, like on any level. And part of the reason that I have a hard problem with the second law of thermodynamics, I think it's in a problem. Think it's an approximate law, like I think it works in some situations, but it's a very subjective law. It's actually the most subjective law of physics there is. And I think in some ways, it's even more subjective than issues with like quantum observers and things that happen quantum mechanics, because we don't even know how to interpret quantum mechanics because of the observer problem. But in thermodynamics it's kind of hidden in because you need somebody to actually label the states of your system, and the labels are actually relative to a measuring device or an observer. So you end up how you define something as being high entropy depends on like, how you label the configurations of your system and how you choose to measure them. So that already suggests to me it's a problem. But then there's other issues, like you have to assume you have many copies of the same system to talk about entropy increasing overall of all the systems, and we only have one universe and a particle in a box model. It works great. And for simple systems that have no memory in them, it works great. But I think the narrative that I see with assembly theory is that actually what's happening is the universe is trying to maximize. Amount of stuff that can exist, or maximize the number of configurations that are actually physical. And that process is what we see in our biosphere. And we don't know where that process is going, but it seems to be one that would forever trend in more complexity, and there will be energy limitations on that. And so that's where the interesting thing comes in. But I really like David Deutsch's idea the beginning of infinity that like, once you have systems like us that can build knowledge like the potential of what can happen in the future is limitless and Freeman Dyson also talks about life infinitely far in the future. In his writing, he had this great paper about analog life forms living in an exponentially expanding universe. It's like you have a metabolism that goes slower and slower and slower and slower and slower. It could just persist infinitely far in the future. So I guess my perspective on it is life is, I think, much deeper than current physics in terms of the structure of the universe we live in. I think it's very deeply embedded, because it's about causation, and it's about the structure of the existence of objects, and there are fundamental laws there that could have a totally different explanation and a different projection for what the future looks like than what our current laws of physics say. And I think there will be some interesting intersections of them. One is this idea of virtualization running faster and faster life forms in computers is kind of one way that you could have life persist infinitely far in the future in smaller volumes of space, because you put all this evolution, it takes billions of years of evolution to get to something like us, to build virtual simulators of the past history of our planet. And so now you can put all of that time in a small volume of space. And I think this is like the physical to digital transition. So that's one way you can think about life constructing more, but it would just become more insular and more virtual in a particular physical object. But I think there's other ways that we can think about this expanding possibility space actually being a real, physical thing, that we see the universe getting more complex into the future. So I have two ways of thinking about it, one very virtual and one very physical, but I think they're kind of the same thing, and I don't really know how to reconcile that, so it's a set of thought experiments I'm playing around with right now, actually.

Emily Silverman  

When I was reading your book, I wrote down a couple examples, and I labeled them "The Magic of life," which I've heard you say that magic isn't a word that you love, but that's just how I labeled it.

Sara Imari Walker  

That's okay. I like, I like the feeling of magic. I don't like magic as a substitute for trying to understand things.

Emily Silverman  

For explanation?

Sara Imari Walker  

 Yeah.

Emily Silverman  

The first example I wrote down is from when I was pregnant, and obviously this baby was forming inside of me, and so I got really excited about embryology, and I went on YouTube, and I started watching these embryology videos and learning more about how the thing organizes itself. And I was reading about something called the primitive streak, which is this axis that develops very early on in the embryo. And they've studied it with chicken embryos, and I've read descriptions of it, and they're really magical. I can't think of a better word to describe it, where they talk about the cells like dancing and twirling, and they migrate to the midline, and then they stop, and then they reverse, and they change direction as the embryo forms, and then it thickens, and then it folds. And I know that you've talked about how a lot of the information or blueprint or memory or knowledge that is needed to construct an animal, like a chicken or like a human being, is stored in the genome is stored as DNA. But there is something else about the way that the matter is animated and comes alive and starts dancing unfolding that is clearly not Newtonian motion. It's something else happening. And so what do you think is happening as that blueprint is then actually being built?

Sara Imari Walker  

I use DNA as an example and genomes because people are familiar with it, but I by no means think that all the information or causation that's relevant in biology is in the genome, and I think very little of it is actually in the genome. So the way that I think about biological structures, or like any object that evolution builds, is you have this underlying assembly processes, recursive construction of these objects, and the object itself is actually not the thing that you observe, like if you think of a molecule, it's not the three dimensional molecule, and the configuration of where the bonds are that's relevant, or atomic composition, or its molecular weight, or any of these things you might think of as physical features. The thing that's physically relevant for the processes that we want to think about as being fundamental to life and evolution is actually that construction history. So that assembly space is what we call it, the space of the way that you can actually make this object from smaller parts by reusing them to make this final object. And so if you think objects have a size and time like they have an assembly space underlying them, when you're looking at all these component parts, these are objects that are deep in time, that are actually interacting across that temporal scale at the same time that they're interacting in the three dimensional space. If you think about the three dimensional space as one layer. But then you also think about time is actually a physical component inside the objects.

Emily Silverman  

This is really trippy. Like, keep going.

Sara Imari Walker  

Yeah, it's okay. I mean, it sounds totally crazy, and I hear myself, and I know what I'm saying, and I know it sounds totally- but it's so consistent with what we're doing with how we measure assembly in the lab. This is the physical attribute we look at for molecules. But also when I think about formalizing the origin of life transition and what physics is necessary, this is really the physical description and what gives objects agency, and all of this causation that you're seeing with these things dancing around to assemble something that becomes an embryo. We all start as a single cell, which is just totally mind blowing. Right? Every single living thing on this planet goes through the phase of being one single cell. Development is crazy. How is that even possible? How do we not just, like stick cells together that are already of the type they are, but it's because we're evolved and constructed things, and so all of these things had to be scaffolded on the layers that came before, and so that information still persists in those structures. And so the reason that development can happen, and also evolution can happen so fast, I think, is because of this temporal size of these objects, and that they don't lose their history. It's all bundled up in there, so they're already causally contingent. And the causation is in the structure, and

Emily Silverman  

it's where? -If it's not in the genome, where is it? 

Sara Imari Walker  

The information is in every single physical piece of the cells. I think even the genome we don't actually have right every single piece of the genome of any length is undergoing selection all the time. But we tend to fixate on genes as the functional units and only talk about those. We don't talk about this hierarchical structure where selection is happening across every component of every length. That's easy to think of as a sequence, but now, if you think about the cell, all of these molecules that are reacting all the time, every single scale is undergoing selection from the cell itself, the tissue, higher level pattern the cell, and then the chemistry inside the cell, and then the parts of the molecules inside the molecules. So the way that we think about that is to look at the structure and then build down into the assembly space, and then think about the interactions that are actually happening across the layers of the assembly space, which is like if you could unfold a structure across time, you would have this much larger space, and then you could talk about all the interactions across that space, and that would allow you to explain more of the dynamics that you're actually observing. This is very cutting edge. We haven't really gotten this kind of work out yet, but this is where I imagine the dynamics of assembly spaces are actually going to be able to provide models for what we call agency, or how we view causation more broadly in living systems. 

Emily Silverman  

Yeah, I mean, the second example that I wrote down here is the ATP synthetase machine.

Sara Imari Walker  

Yeah, that one's crazy. 

Emily Silverman  

I remember learning about it in biochemistry, and it's this molecular motor. And I was Googling it, and I came across this article written by this guy, Jonathan McClatchy, who's like a Christian writer slash scientist, and he's trying to puzzle through how this molecular motor could have been built. And so he says one hypothesis is that a proton motor came to be associated with some kind of helicase, and the ATPase of the helicase was driven forcibly in reverse by the proton motor. And that was one option. The other scenario he talks about is maybe there was a translocating protein that got stuck in the protein trans locace, and then subsequently evolved into the central stock of the motor. But then you have to couple that stock to the ring in order for it to rotate and so on. And each example just seems more and more far fetched, and then a lot of story building. He says "it's a marvel of nano technology, an amazing molecular energy turbine. One is led unsurprisingly, to the conclusion that such a machine is best explained by intelligent design." Yeah. So what say you? How does this thing get built? Is it reaching back 3.8 billion years into history, and just like a hand, like puts it together? Or how does that happen? 

Sara Imari Walker  

I think the universe is constructing itself. So I don't find it helpful to explain what exists by assuming that there's some entity outside of our universe, or some explanation that comes from the outside the boundary of our universe, which even standard physics would put because you have to set the because you have to set the initial condition of the universe, and you have to also decide where the laws of physics come from. I don't think any of that's necessary. I think our universe is literally constructing itself. And if you want to understand why certain structures like ATPase come into existence, you actually have to look at the causal contingency about how such a structure could be built, which is like the structure of the assembly space, and you have to look for what objects could co construct with it. This is sort of, I think, the issue why everything looks intelligent in design is, if you look at any object in isolation, of course, it's like, where did that thing come from? But it had to be built by the things in its environment. And so really, what we need to do, and what we're trying to do with assembly theory is look at objects that were co constructed and actually construct what we call the joint assembly space of the construction histories, and use that as a model for how we think about explaining why these objects come into existence. And one of the key features also this goes back in the long history literature of the origin of life, is like this idea of auto catalytic sets that you need systems that one molecule can produce the next molecule can produce the next molecule, and they form closed cycles. We don't exactly think about it that way in assembly theory, but there is this idea that, like when you're moving into this combinatorial space of these incredibly intricate, complex objects, a lot of it involves trial and error and making things that just can't persist. They're not able to persist in time because there's no selection mechanism to reinforce them. What happens is there's a random search, and you get a collapse on the space of a set of objects or molecules that can reinforce their own existence. And so ATPase cannot come into existence on its own. It has to be part of this other structure where it's all self reinforcing. One way I think about it is like we're all hanging on to being able to exist by coexisting with a bunch of things that help reinforce our existence, right? You imagine the universe is like random out there, and there's this insanely large combinatorial space of possibilities. The only things that can be structured are things that can be co built and actually exist along these same lineages. So usually I describe life as lineage, as a propagating information or these sort of construction histories or trajectories we were talking about. So design emerges along those because it starts to look like things have function because they're reinforcing the structure and existence of other objects. And so I think very much a function is an emergent property of this kind of dynamics, not as fundamental, but you would expect to see things over time that look more and more designed because they have more and more history and more and more selection and more and more constraints that were ruled out in order to make that specific object. The only reason that object exists and not some other object is just because of the particular contingency in the historical pathway to get there. So we're ruling out a whole bunch of objects that will never exist, like the transistor example, like all these technologies, that will never exist because they're just not along our particular pathway. But it doesn't mean that anything in that path was designed by something outside of it was all internally constructed, and all has to be self consistent along that particular trajectory.

Emily Silverman  

I once heard you ask in an interview, and I think you were asking this question to yourself, "is there a problem that the universe has that life solves?" I'm wondering where you've landed on that you know, we have these exploding phylogenetic trees of increasing complexity and novelty. What is the problem in the universe that life is solving? 

Sara Imari Walker  

Well, one part of it is, I was told by a very prominent senior physicist, you're not supposed to ask "why" questions after I gave a lecture once, and I thought that was really hilarious, and I actually was doubly bad, because he was like, the last person to ask a question, like, this was Boltzmann, and you know, it happened to him. Oh, my God, are you kidding me? But also, you know, people have been asking, what is life for so long? This is a famous Schrodinger book, and the what is life question is just like over and over again, we're not asking, Why does life exist, or what does life do that nothing else in the universe could do? And I think those questions are actually more scientifically productive, because the what usually follows the why and the what does it do, kind of question. So I thought a lot about the nature of life and like, what are the things about reality we couldn't explain without a fundamental mechanism for life. And like, what is life doing? And the thing I've always landed on, and I had ways of talking about it before assembly theory, and I have ways of talking about it after assembly theory that I think are much more precise and concrete, but the way that I've always thought about it is life is sort of maximizing the number of possibilities that could exist. That was an early thought I had, because there's a lot of stuff that can be created on our planet that cannot be created anywhere else in the universe. And some of those things you might really like, like the Library of Babel or Shakespeare's works, and some of them you might not like Tiktok or pick your favorite political campaign. Like there's lots of good and bad things that people love or hate. I'm pretty agnostic on everything on this planet, but they can't exist anywhere else. They can only exist here. And I think this is really profound. So that was one part of it. But then as I was thinking more deeply and thinking about really, just how large the space of possibilities is, which is when I really started working on assembly and really thinking about the size of chemical space, which is just like, ridiculously huge. I mean, I already talked about this pharmaceutical drug design problem. It's like, epically huge, like, you can't train AI on things that are that big, that you don't even know how to make them think about that. You know, it really struck me that what life is doing is maximizing the number of things that get to exist. And so you think about Darwin's in the struggle for existence, every form is out competing each other. But what assembly theory gives you, that's, I think is a deeper story is there's also a competition to get to exist at all. It's not even like once you exist, you get to keep existing. You want to keep existing. Also, there's a drive to try to make more things exist. And I think this is also one of the reasons that evolved objects like us are so efficient for the universe to maximize existence, because all of those layers of evolution and history and contingency now get packed in a small volume of space, like I'm five foot three, so it's kind of tiny volume, but I represent 4 billion years of evolution packed into one physical structure. And there's a huge amount of recursive layers of structure in me, and all of that gets to coexist in one physical object. So that feature, I think is really important that the universe is actually increasing the recursivity in objects as they get deeper and deeper in time. So they're becoming much larger temporal structure, small volume of space maximize existence, but also being able to use those structures to generate more complexity and more stuff. And so I guess I would think about the universe as kind of like the universe is what exists in a local volume that we can observe. There's the Hubble volume, which is like observational boundary of our universe, and it's just trying to pack as much stuff as possible in there, because everything wants to exist. Because the other option is we don't know what non existence, but non existence is a concept that exists, so I don't actually know what it is to not exist.

Emily Silverman  

Well, you said something once about how a lot of people look up at the night sky and they feel so small, they feel like a spec and so on. But you're really flipping that idea on its head and saying, Actually, we are some of the largest structures that exist in the universe. And I really loved that. In these last couple minutes. I just wanted to ask you about aliens, because we have to people think of aliens as life that evolved off Earth, but you have a slightly different way of thinking about it. So tell us how you think about what an alien is, and then how you're going about trying to discover/ find aliens.

Sara Imari Walker  

I never liked the idea of defining life. A lot of people my field try to do it. I think assembly theory has a natural definition for alien life, and alien life is something that emerged from the random froth of chemistry along a different causal history than we did. And so the origin of life as an experimental paradigm becomes an interesting problem, because now you can talk about this huge possibility space, like this huge volume of things that the universe could create, and you can treat it like a search problem. If you could build a chemical search engine that could actually iterate through different chemical possibilities, how long would it take that search to actually find an alien life form? And so this is actually not my idea. This is Lee Cronin idea, and he's been working on building a technology for that for more than 10 years. He's a pioneer in digital chemistry, and the whole reason that he even got into that field was because he wanted to build robots that could solve the origin of life by trying to scale the technology to do chemical searches. So right now, state of the art in the chemistry lab is you have somebody pipetting individual reactions and things. And so he's basically built a technology that automates that process, and the idea is that we'll be able to scale it. And so we're collaborating on this idea of trying to build large scale, origin life experiments guided by theories. So we have Assembly theories, the abstract, fundamental formalism, and we want to build these very large, robotically driven experiments to look for aliens in the lab. It's a little bit like how particle physics has the theory coupled with massive experiments. And we started a company called chemify, which is trying to scale up the technology. It's actually a drug design company. They build the technology to build molecules, and I'm writing a book to try to get people excited about it. So that's what we're trying to do to get there.

Emily Silverman  

Should we be scared?

Sara Imari Walker  

 So it's like, no, I don't know. I'm terrified, but I don't know how to be scared of understanding things. I love this Marie Curie quote, it's like, now is the time to understand more so we can fear less. And I really feel that very deeply. And so I guess for me, when I talk about the things that we're envisioning doing and the ideas that we're excited about, a lot of people have this kind of fear response, oh my god. Like, what kind of synthetic life forms are gonna come out of these labs, and what kind of technology are you building? But really the motivation is the fundamental understanding. And I think we're so existentially traumatized by these living technologies and all these other things that we don't understand, that unless we really understand fundamentally what life is, we're not going to be able to understand how we're talking about aliens or what they are. We're not gonna be able to understand how we're talking about artificial intelligence as an artificial life or what they are. This is a really existential problem we have to solve, and I would be more afraid of walking into the future not knowing what it is that we're talking about than doing this kind of fundamental science to really understand the core of this question. And so I think there probably are absolutely technological repercussions that would be scary, but I also think that there's a lot of them that are incredibly optimistic. And ultimately, I think as humans, we build our own future. And so I want to keep choosing the optimistic one and the one that gives us the most possibilities for the future, because I think that's what life does. And so I'm going to keep trying to work in that direction. And I hope other people do too, because I think our future could be insanely bright if we want to build it that way. 

Emily Silverman  

Boom, yeah. Well, this has been so much fun for me. I've been looking forward to this for a long time, and I'm just so happy to have you on the show. And wanted to encourage the audience to please consider going out and buying Sara's amazing new book, "Life As No One Knows It." You'll learn a ton. I also recommend just checking her out on YouTube and watching some of her talks, because she's a great writer, but as you can tell from this interview, she is also just so skilled at communicating complex scientific topics to mortals like me. So thank you again, so much, Sara for coming onto the show, and I can't wait to keep tracking your work and see what you do next. 

Sara Imari Walker  

Thanks, it was a lovely conversation. Really enjoyed it.

Emily Silverman  

This episode of The Nocturnists was produced by me and Jon Oliver. Jon also edited and mixed. Our executive producer is Ali Block. Our head of story development is Molly Rose-Williams and Ashley Pettit is our program manager. Original theme music was composed by Yosef Munro, and additional music comes from Blue Dot sessions. The Nocturnist is made possible by the California Medical Association, a physician led organization that works tirelessly to make sure that the doctor patient relationship remains at the center of medicine. To learn more about the CMA, visit CMAdocs.org, The Nocturnists is also made possible by donations from listeners like you. Thank you so much for supporting our work in storytelling. If you enjoyed this episode, please, Like, Share, Subscribe and help others find us by giving us a rating and review in your favorite podcast app, to contribute your voice to an upcoming project or to make a donation, visit our website at the nocturnist.org. I'm your host, Emily Silverman, see you next week. 

0:00/1:34