The following is a rough transcript which has not been revised by The Jim Rutt Show or Sara Walker. Please check with us before using any quotations from this transcript. Thank you.
Jim: Today’s guest is Sara Walker. Sara is an astrobiologist and theoretical physicist at Arizona State University. She’s the deputy director of the Beyond Center for Fundamental Concepts in Science and a professor in the School of Earth and Space Exploration. She’s also on the external faculty at the Santa Fe Institute, and she’s a Polymath Fellow of the Schmidt Sciences. Now that’s a real fancy thing. Welcome back, Sara.
Sara: Thanks so much for having me. I’m really excited about chatting.
Jim: Yeah, this should be really good. Sara was on back in episode 100 with her colleague Lee Cronin – the topic was “Time as an Object.” Extremely interesting, it’s one of my favorite episodes. So if you like what you hear from Sara today, check out that one too. It’s more focused on a subset of her ideas, but it is amazing. Actually, it blew my mind. I loved it. And of course, regular listeners know what today we’re going to talk about her new book…
Sara: “Life as No One Knows It: The Physics of Life’s Emergence.”
Jim: Indeed. Regular listeners know origin of life is something we do quite a bit of on the show. We’ve had several episodes with Bruce Damer, where he’s laid out his interesting ideas. We’ve had Eric Smith on, we’ve had Stuart Kauffman, and we’ve had some others. So this will be another in our on-and-off series on the origins of life. Sara brings a very interesting perspective and an interesting background. Your original education was in physics and astronomy, as I recall, right?
Sara: Yes, I am formally trained in theoretical physics.
Jim: How does a theoretical physicist find themselves working in the area of origins of life?
Sara: Well, actually, it’s quite funny. I wasn’t very interested in it, but my PhD advisor was. So I went to Dartmouth to do cosmology, and my PhD advisor Marcelo Gleiser was a theoretical physicist, what I aspired to be, and he wanted to work on astrobiology. So I thought, oh, I’ll take the astrobiology project and then I’ll get to work with the cosmology professor and then I’ll get to do some cosmology too and I’ll ease my way into cosmology once he realizes what a great student I am. That’s not exactly what happened. I got more convinced on the origin of life being a great open question from my work with Marcelo. And so trying to work on foundational questions with a faculty member that I looked up to and then gradually getting convinced by the problem that it was so big, I wanted to spend my career on it.
Jim: Unless there’s some amazing breakthrough, there’ll probably be plenty to be done, right?
Sara: Yeah, that’s right.
Jim: As you point out earlier in the book, it is one of the big questions. I sometimes say it’s number two or number three, something like that. I think you laid out your list of the big questions. You put first the hard problem of consciousness, as I recall.
Sara: Yeah. Is that your number one too? I’m wondering if you put life at two or three, what’s number one?
Jim: The number one is the second one on your list. The hard problem of matter.
Sara: Oh, I see. Okay.
Jim: That’s right. I describe it, why is there something and not nothing? Right?
Sara: Yeah.
Jim: And then I put life number two and I would not put hard problem of consciousness because I’m not sure there is a hard problem of consciousness.
Sara: Oh, I see.
Jim: That is actually an area that I work in and know a tremendous amount about. And the discourse around consciousness is a lot of blather. But anyways, another story for another day.
Sara: No, I totally agree. I think usually when we’re identifying these hard problems, I think it’s because we’re not asking the question the right way. And so usually once you figure out the way to ask the question, it gets easier to answer it. But you don’t know what question you’re asking.
Jim: Yeah, in fact, my view is that once we understand the mechanisms of consciousness, it will become obvious why quality is what it is and there won’t be any hard problem. Yes, but I could be wrong. And so could people who also hold that position. But that’s alright. That’s a good way to frame it. And then the third one on my list is the small little question of the unification of quantum mechanics and gravity.
Sara: Yes.
Jim: That one is still like, why? Why do they refuse to be good children and play together?
Sara: Well, it’s interesting when we talk about unifications in physics because I think this is the first time where we’ve really tried to unify theories that are mathematically incompatible, without having a new phenomenon in the universe that we were trying to describe and realize we had to unify things that we thought were different before to actually approach that. So it’s very different from like the unification of space and time or wave and particle or electricity and magnetism. Like there were questions there about experiments or things we couldn’t explain, we were trying to unify. But in this case, it’s like, we know these theories are mathematically incompatible, but we don’t have something we can’t explain that sits at the intersection of them, I think, in quite the same way.
Jim: It’s the case there’s no anomaly, but there is a question, right?
Sara: Yeah, exactly. So I think it’s a little strange and it’s a fairly recent phenomenon in theoretical physics to try to do that. I think it’s only since like the 1970s that all these theories of unification came about that were more mathematically driven than driven by actually confronting phenomena in the real world and asking questions that, you know, like what things do we need to look at differently here in order to answer this question?
Jim: And that can be dangerous. Like for instance, string theory sucked up 60% of the PhDs at one point and string theory is another one of these examples of a question, but not an anomaly, right? It was fundamental.
Sara: Yeah, no, I totally agree. I think that’s a good example. But the other thing that kind of bothers me is like, if you look in sort of popular science, you know, lot of these conversations about theories of everything are about math and have nothing to do with science. So I don’t know where that came from culturally. It’s quite interesting.
Jim: Yeah. But fortunately this one is not one of those.
Sara: By no means, we have a…
Jim: Whole bunch of interesting, real questions to answer. But the other thing I’d like to commend about your book is the other thing we do on our show a fair bit, not as much as origin of life is philosophy of science. And you are very careful to lay out what you think your theory needs to do. You basically pointed out the classic high explanatory power, but then you also brought in the much more subtle perspective of Deutsch on theories have to be hard to vary. Can you talk just briefly about those two topics? Anything else you might want to say about your philosophy of science?
Sara: Yeah, I think it’s been really interesting in my career because I didn’t train as a philosopher, but the deeper I go into theoretical physics, the more philosophy I have to do. And I found that actually very shocking. So now I kind of almost in some ways consider myself as much a philosopher as scientist, but not formally trained as such. So I don’t use that label, but I deeply respect philosophy.
I think the interesting thing is a lot of times when we’re approaching scientific questions, like say the problem of life detection, that problem is viewed in a very small scientific lens typically. It’s like, can we detect amino acids on Mars or can we detect oxygen in the atmosphere of an exoplanet? And separately from that, we have the whole origin of life field, and they’re trying to think about prebiotic chemistry and how they make simple molecules that might be implicated in life on Earth.
These are considered totally separate problems in the field of astrobiology historically. People doing Origins of Life haven’t really talked to people looking for life on other worlds, and they’re not talking to people looking at life in deep time. There’s some interdisciplinary collaboration, but there’s no explanatory framework for this broad phenomena that we want to call life.
And if you look at the major revolutions in science, in particular physics, the theories that work really well, the ones that are very deep, explain a huge amount of phenomena. So it goes back to the problem of unifications. Like if you think about gravity and what Newton came up with, that’s unifying planetary motion with things that happen here on Earth. And so Galileo was doing all these experiments. There’s a whole generation of the first real physicists that came up with this idea of a universal law of nature. And part of the power of that is this idea of broad explanatory power.
So I think for me, the thing you want to do if you’re tackling a problem like life is you don’t want to hold on to any definition of life that might apply to one instance or one particular thing. You want to find this sort of broad explanatory regularity that sits in the realm of laws of physics that allows you to explain more of how the world works. In particular, very important for understanding the properties of alien life because we don’t even know what kind of forms alien life would take.
And that’s hard to vary part is just, it gets into debate about what is the nature of science and what is a scientific idea versus not a scientific idea. But it’s really easy to build theoretical models that can be tweaked in one way or another and still fit your data. Like people do this all the time and they don’t embed their theories in measurement. And the measurement really, it forces your theory to have a particular structure. And so this is something that we’ve done a lot with development of assembly theory that I highlight in the book is that the theory has certain properties that are meant to correspond to the physical world and it has falsifiable hypotheses but also the structure of the theory can’t really be changed and still be the theory. And this gets into some really interesting debates with people in computer science and things who think it looks like measures from computer science, but it has a completely different ontological status and rigidity to the actual mathematical structure that we have because it’s trying to capture something about the physical world.
Jim: Yeah, I’d also say it’s, well, it’s not the same thing. It is speaking to the same thing that the parsimony principles are speaking to, right? When I see theories that are obviously just, this, this, this and this fit the data. Go, okay, any fool could do that, right? Let’s get down to something that’s much more specific and much shorter.
Sara: Yeah. And it’s actually quite hard because building good theories is not data fitting. And this is something I think gets really confused in the AI of this era of machine learning and artificial intelligence for science, because people think the best model is the one that fits the data best. And often that’s not the case and that’s not been the case in the history of science. The best models are the ones that are most explanatory and allow us to make good predictions about other phenomena we haven’t encountered yet or to anticipate things that we don’t understand yet. And you see this actually, again, in the history of gravitational physics, because when Kepler was coming up with his laws of planetary motion and we were trying to think about spherical orbits and then elliptical orbits, there were versions of epicycle models from Ptolemy that made better predictions.
So I think there is a lot of tendency for the math to lead the thinking or lead the prediction without actually thinking about what these mathematical structures are representing about the physical world and what they’re telling us about how the physical world behaves. And this is actually much more conceptual foundations or philosophical work because then you have to have an interpretation or an explanation that underlies the math. And in some sense, that explanation is deeper than the mathematical description. So when Einstein came in and came up with better theories of gravity, we didn’t throw away the idea of gravity – the conceptual framing of a universal physics where masses behave certain ways that Newton invented is still present in Einstein’s theory. It’s just that he came up with an even deeper explanation than that one. So it drives me nuts because there’s a lot of people that just think, like you just follow the equations and you have understanding. Or you follow the data and you make a prediction, you have understanding. That’s not understanding.
Jim: David Krakauer and I talk about this from time to time, which is…
Sara: Yeah, David’s great on this topic. He’s great on every topic, but yeah.
Jim: Yeah, the danger – data-driven science is certainly useful. But it’s also a dangerous attractor in some ways if it causes us to forego actual understanding. So people need to know how to use the tool, but not to reify its answers. These aren’t necessarily the fundamentals. Anyway, I really commend you for addressing that in your book and framing your work in this style because it’s a good example of how to do science right, at least in my opinion, whatever that’s worth.
Sara: Well, thank you.
Jim: But before we hop into the details of your theory, maybe we should talk a little bit about the histories of theory of origins of life, which you do a little bit of, and they are even sillier than most older theories. So let me give us a little bit of the history of the idea.
Sara: Yeah, so I think it depends on how far back you want to go, but I guess like the modern history of origin of life probably started in the 1920s with the work of Hal Payne and O’Perrin. And they were just basically proposing the idea that some complex chemical mixture on the early Earth could have given rise to life. But that didn’t become an experimental research program until the 1950s with the Miller-Urey experiment. And that experiment was very famous. You put some prebiotically relevant compounds, things that we might have found on the early Earth, in a glass tube, and you submit it to simulated shocks of lightning and you get things like amino acids coming out within a couple days.
When that experiment was first published in 1953, people thought new life forms were gonna be crawling out of these test tubes within a matter of weeks. But that’s not what we saw. What we see is if you keep the system running, you get what we call a tar. It’s just a complex mix of organics. And this is also the kind of structure that we see in meteorites where you get kind of a combinatorial explosion because there’s no selection and evolution in these systems. There’s just chemical combinatorics. And so you get all these molecules and you can’t even identify them anymore. It’s just basically a carbon sludge.
That experiment was so influential that it started a field of research called prebiotic chemistry, which has been ongoing for many decades now since the ’50s, where people have been trying under different conditions that they think are prebiotically relevant to make the molecules that we find in living things. And so I think that research program has had some utility, but I don’t think it’s the way of solving the origin of life. And in fact, I think it’s at this stage not helpful.
And then there have been other attempts in other areas. So you mentioned, you’ve had Stu Kauffman on, and he had some really brilliant ideas about collectively autocatalytic sets, so systems of molecules that could collectively reproduce themselves. This was kind of an early idea in origins of life field. And what he was responding to was a lot of the folks that were doing this prebiotic chemistry were trying to make genetic polymers. So something like an RNA molecule that could replicate itself. So once RNA was discovered, it became the sort of natural candidate for origin of life hypotheses. And so a lot of people were thinking, “Oh, if we just get RNA made under some abiotic condition, the molecules can copy themselves and then they can kickstart an evolutionary process.” And so Stu came along and was kind of arguing that we need more collective autocatalytic sets, self-reproducing systems. But those have been also hard to implement in the lab. They’re really brittle when you actually try to make an experimental system that has these properties. And so there’s been a lot of really great ideas historically, but none of them have borne fruit of totally solving the problem, which is why we’re all still working on this and scratching our heads.
Jim: And you then bring up what I think is the interesting trajectory, which was also surprisingly early, which is Schrödinger’s work on negative entropy. And then you later stitch it together with Prigogine’s dissipative systems. So maybe you could riff on that a little bit.
Sara: Yeah, for sure. Sorry, I was thinking of explicit origin of life, like people that call themselves origin…
Jim: I think this is now where we go to the next step, to the next four.
Sara: So actually, it’s quite funny. As a PhD student, I read Schrödinger’s book, “What is Life?” given to me by my PhD advisor. It’s sort of the physicist’s handbook – if you’ve been trained in physics, here’s what biology is about according to Schrödinger, kind of thing. And obviously there’s a lot in that book that’s interesting, but some of the speculation is quite wrong. What’s cool about that book is it’s an attempt to try to make some arguments about what features of life are very hard to explain with current physics and which ones might be explainable. And I think Schrödinger did a very good job about that.
So a lot of it’s confronted with this problem of the second law of thermodynamics and why don’t living systems just decay into disorder. And he identified that there had to be some kind of information content, which he called an aperiodic crystal. So in physics, we know about crystalline structures, which have exact repeated motifs. And so they don’t contain a lot of information because you can basically specify whatever motif is repeated and you can talk about the structure of almost the entire crystal. But what Schrödinger argued was that there should be some kind of material in what he called an aperiodic crystal that couldn’t be compressed in the way to just a single repeated motif but had all this complex structure, and that would have to be something that underlie the genetic material. And that obviously inspired some of the discovery of DNA, looking for it as, and it does have the properties that Schrödinger set out.
So that was kind of an early attempt to try to understand where physics could explain things and couldn’t explain things. And my favorite quote in that book is toward the end of the book about this idea that other laws of physics might be necessary. So even though Schrödinger was one of the pioneers of quantum physics and knew in quite depth about theories of nonequilibrium thermodynamics, he still thought that we were missing explanation. And Einstein wrote very similarly also that we should be confronted with how primitive physics is when we look at living systems.
But Prigogine came along a bit later, and he was also very fascinated by this idea that maybe we know equilibrium physics well, but if we knew non-equilibrium physics better, that would be a sufficient explanation for life. And so he pioneered a field called dissipative systems, which are systems that become organized because they are held far from equilibrium. And so this seems like a very promising research program for a while, but we know of lots of abiotic dissipative structures. So one I talk about in the book is the Great Red Spot of Jupiter – should we consider that an example of alien life or not? Probably we don’t want to.
So again, it gets into the question of, it’s not that life violates the current laws of physics, but none of these descriptions seem to be adequate. So if we look at the molecules of life like the prebiotic chemists do, we can’t produce the whole thing. If we look at the specificity of the molecules and how we synthesize them, if we try to look at the organizational states of living systems from the lens of more traditional approaches to physics like non-equilibrium thermodynamics or trying to understand information from a classical physicist’s perspective, we can’t explain it. And so we’re kind of still in the dark. It doesn’t seem to be a self-organized state of matter. It doesn’t seem to be the molecules it’s composed of, and we don’t have the right principles to actually explain this transition from nonliving to living matter.
Jim: And then you guys make your leap. I found a single sentence that sums it all up in your book. Do want me to read it to you, do you want to say it?
Sara: No, I’d love you to read it to me. I don’t know what sentence sums it all up, so that’s cool.
Jim: At least I think it does. What we came up with stated most bluntly, these are your words, not mine, directs attention to the question: how is it that information can cause things?
Sara: Yes.
Jim: Did I do a good job of finding the center?
Sara: That was good. It was pretty blunt. It was good. When I was a postdoc, I wrote a paper with Paul Davies, who was my postdoc mentor and still, you know, very long-term collaborator and friend. He’s an excellent human being. But I arrived at ASU and he said, “I don’t think anyone knows what the origin of life is. Why don’t you spend a little bit of time trying to figure out what the actual problem is?” And I think that was like the ultimate gift of my career.
But what happened is I worked with him on this paper where we suggested that information as a causal category emerges at the origin of life. So the transition from non-life to life is somehow information becoming causal. And that’s a really abstract thing to say, very challenging to formalize. I spent a lot of my career as a postdoc and then into my early faculty life trying to figure out how we could build theory consistent with current physics that meets that property. And I think that it demands new principles, which is why I work on the theories that I do now.
But that was kind of the idea of it. This idea that information has causal efficacy of matter, we kind of coined the term “the hard problem of life,” because now you’re talking about an abstract quantity information, something that feels very disembodied, doesn’t have a material reality to it in any way that we can really understand. Right now, somehow, actually, it matters when you have a thought that things happen in the world, or my words will matter and can entire crowds of people. How is it that these very abstract properties of living things actually end up reshaping the world around us? That’s a very perplexing thing and philosophically has been debated for a very long time.
Jim: And of course, that gets us into the, you know, the whole area of complexity science around emergence.
Sara: Yes.
Jim: In some sense, want to riff on emergence a little bit, another like consciousness, another topic in which much blather is about.
Sara: Oh my god, it’s so much. It’s very funny because my lab at ASU is called the Emergence Lab, and I don’t even think emergence is a real thing anymore. Like, emergent properties are real, but the way people talk about it is exactly what you’re saying – it’s very disembodied. It’s like lots of ideas and they use it to connote something mystical. You mentioned having Eric Smith on, and Eric Smith has this really great quote where he talks about how reductionism makes emergence possible. So people think they’re counter to each other – like reductionism is the antithesis of emergence, where reduction is you study the parts and emergence is you look at how the whole can’t be reduced to the parts. But his whole point with that quote I think is that unless you can reduce the system to parts, you can’t even talk about the parts having collective emergent properties in the first place.
I think it’s kind of weird how we talk about emergence. This idea of information being an emergent property, right? Like it’s supposed to be something about how collections of molecules in a body are alive, but the individual molecules are not alive. And so people talk about life being an emergent property, but that’s very hand-wavy and ambiguous. I started work early in my career working on notions of emergence because that was the best place to think about it. It was something more than what the laws of physics maybe gave us in the individual parts. We get something more out of the system and that’s what we should call life. I was even playing around with other theories that try to deal with these things like integrated information theory and studies of consciousness, but ultimately found them not useful for things I was working on for a variety of reasons.
Long story short, there’s lots of ways that people talk about quantifying emergence, but in the end they end up being very subjective and observer-dependent, and they depend on the way that you want to put a labeling scheme on your system. They don’t have this property of law of nature, measurable quantity that I think we really want to have in things that majorly advance science and give broad explanation. They tend to be very specific to an individual system and have all of these kinds of issues that I was just talking about.
Jim: Yeah, I think that’s very fair. A little homey example, I’m trying to talk to civilians about emergence. And also, you made a very good point about the necessary relationship between reductionism and thinking about emergence or complexity more broadly – because you got to do both. And Harold Morowitz’s book, “The Emergence of Everything,” is really great at that. Because he points out that each emergence is a pruning rule essentially, right? Which then rules out or at least reduces probability of other trajectories.
Sara: That’s right. I think thinking about it more in terms of selection and constraints on future evolution of a system is much better. So I still have a sense of emergence, but for me, emergence is just about how much selection had to go into the construction of this configuration. And that actually becomes a physical quantity in the way I think about life now related to how objects exist in time, like our other conversations. So all of these things are connected.
Jim: Anyway, my little homey example between the relationship between the two is the dancers are the reductionist part and the dance is the emergent part.
Sara: Oh, I like that.
Jim: Yeah. And it’s very specific, right? That’s your point that there may not even be a general theory of emergence, but there are emergences, right?
Sara: Yeah, there’s definitely – I mean, it’s definitely a real thing in nature, but it also tends to be about different levels of description. So one of my favorite examples is like, when does water get the emergent property of being wet? Like a single water molecule is not wet. You have collections of water molecules, they’re wet, but like how many do you need? And a lot of the properties that we associate with water that make it wet are about intermolecular interactions. And the reason that a single molecule is not wet is just those interactions don’t exist for a single molecule. And so it’s not mystical, it’s just there are genuinely new properties that emerge at different scales. There I use the word emerge. I guess I believe in emergence.
Jim: Yeah. There’s a big difference between believing in emergence and believing that there is a simple theory of emergence.
Sara: Yeah, exactly. Yeah.
Jim: All right. So let’s go on to the next thing. And this time I’m gonna take a little deeper dive than I sometimes do, because I also went down this rabbit hole. I just thought it was so interesting. One of your core ideas that you work with is Deutsch and Marletto’s constructor theory of information. Because again, we’re hitting all the slippery topics today – consciousness, emergence, information. Don’t go too far, but feel free to go a ways into telling the audience about what constructor theory tells us about information.
Sara: Yeah, so that came into my thinking. Was part of the same sort of network of grants. Paul and I had a grant on the power of information as it pertains to the origin of life from Templeton World Charity Foundation at the same time that David and Chiara were also a part of this network of researchers. And so I got really into constructor theory and had a lot of conversations with Chiara at the time about it.
I really like the research program that they’re doing. I think it’s deeply interesting and compelling on levels of explanation that we’re talking about as far as like, you know, their first premise of doing science is to build better explanations, which I deeply intrinsically feel about my own work as well. But I like how explicit they are about it.
The idea of constructor theory is that the sort of way we think about physics – cast an initial condition in law of motion – is not the best description of the world and leaves out a lot of explanations that we might otherwise be able to include in physics. And so they prefer this idea of talking about transformations that are possible or impossible and why they’re possible.
That seems kind of abstract, but the sorts of things that come out of it are really interesting philosophical insights into questions like what is information. Information is a hard thing to answer because it has abstract properties, right? So we have this sort of weird way of – well, words are weird. They exist in the firing of the neurons in my brain, they’re communicated over sound waves, then they’re going in my computer and they’re getting to you and everybody else in your audience. And it’s like, what is the property of the words that maintains that they carry the same information through all those media?
What Chiara and David attempted to do with defining information is not to go the traditional route of like the Shannon-esque interpretation of information where you have to assign labels and you have to talk about communication over noisy channels and then assign sort of what is the information content in terms of correlations between states of systems. They said information has this property that it can be copied between different physical media. And if you assume that that’s a fundamental property, what kinds of transformations are possible?
They basically kind of inverted the whole logic. Instead of asking from the traditional perspective of physics, what is this emergent property of information? They say information is a constraint on what kind of things can happen. If we assume there’s a constraint that is interpolable between different physical media, what kind of transformations do we allow?
The interesting thing about that particular paper that they had was that they could derive two categories of information, one that looked like classical information and one that looked like quantum information. So it had some philosophical utility in terms of re-deriving natural categories that we already knew about, but doing it from this very different axiomatic perspective.
I really took their definition of information to heart and it seemed to me that that was a better way of capturing the notion of what we should consider to be information in nature and how we might develop physical regularities. If we assume life is about information having causal power, it says that some of the transformations that life does mediate this property we call information, that it retains these structural elements associated with information. And so that influenced my thinking early on about information and causation and better ways to formalize it.
Jim: Yeah, I had never run across the topic before, so I spent a couple of hours and one of the things I did that was quite helpful was fire up my ChatGPT Pro and had it do deep research on it. I said, explain it to a reasonably intelligent college sophomore. And it did a pretty damn good job. That was pretty long because the deep research does generate a lot of output, but it helped me understand it. I may put a link to that thing on the episode page – you can read ChatGPT’s reasonably plausible take on constructor theory of information.
So now before we actually make the turn to your theories about origin of life, why don’t we do a relatively brief summary of assembly theory?
Sara: Sure.
Jim: We went much, much deeper in the previous episode. But assembly theory is really quite important for your theories of origins of life, so let’s do a few minutes on assembly theory.
Sara: Yeah, so Assembly Theory is pretty important to me as a philosophical and scientific research program, mostly because a lot of the elements that we talked about – I think it’s been built to try to address problems in all of them that we’ve talked about so far today. The main conjecture of Assembly Theory is that complex objects don’t form spontaneously in the world. They have to be constructed by an evolutionary process. And therefore there should be some kind of complexity threshold in nature. And if we see an object more complex than that, we know that life had to produce it.
This is kind of a very simple statement because it basically just implies biosignatures are complex structures. But the really key innovation of Assembly Theory is the way it formalizes complexity and the way that it ties that to a physical measurement of structure. And then also the interpretation of that physical measurement leads to reinvention of some concepts that, you know, like most new theories of physics, it makes you rethink a lot of fundamental things about the way you think the world works.
The program of Assembly Theory is now developing out, like fleshing out all of the theoretical consequences of that and the experimental consequences of it. But it’s super simple to state the starting point, which is, if I give you a molecule, can you tell me product of life or not? And actually Lee Cronin, who was the original developer of Assembly Theory, is a brilliant chemist, very deep abstract thinker, but does a lot of his abstract thinking by actually building physical experiments. So he’s like in his workshop all the time prototyping stuff. He goes to the lab, actually works with his lab to design and build all these experiments. He’s got a pretty big lab working on origins of life and automating chemical synthesis.
But he was trying to figure out if he had a robotic search engine that could explore different kinds of chemical soups, so like Miller-Urey experiment, but at scale and not stopping them because you got tar, but changing the conditions so you can keep the system evolving. How would he ever know that he got a living system? We can’t necessarily rely on the biomolecules that we know that evolved on Earth because chemistry allows so many possibilities. You might get a totally different chemistry out, but it could be alive.
And so the solution to that was to look for complex molecules. And the question then becomes a measurement question. And the measurement question becomes this invention of the idea of the assembly index, which was actually invented by doing thought experiments with mass spec. So Lee was thinking about how he goes in the lab – a mass spectrometer is a way that people usually try to fingerprint molecules by breaking them apart into fragments and then identifying the fragments and being able to identify the original molecule.
But what Lee thought is, well, I can build this pattern of fragmentation for the molecule. Could I reconstruct something that’s evolutionarily relevant from that fragmentation pattern? And so his thought experiment became one about thinking about the minimal number of steps where you can take a fragment and use fragments that you’ve built before to build up to the original molecule.
And I know most of us aren’t chemists – I usually think about it in terms of Lego when I explain it to people. Like, if you took a Lego Hogwarts and smashed it on the table, it’s a very low probability that if you started shaking my table that those were going to spontaneously assemble into Hogwarts again. And actually, I would say it’s zero. Like, I don’t think if you shook that table, you know, for a thousand trillion million years, it would ever assemble that structure.
But what we do to kind of quantify that Lego Hogwarts is hard to get out of that Lego shaking, but maybe, like, you know, getting two pieces stuck together is pretty easy configuration is to do this minimal path construction, building the assembly index by taking part, taking two Lego, putting them together, and then using parts we’ve built to try to build up to Lego Hogwarts. And we just look at the shortest path as the bound on the complexity of the object.
And then Assembly Theory says, if I find an object with a very large assembly index, a large number of minimal steps in its causal pathway to build it, and I see that object in high copy number, the only way that it could be produced is as a product of evolution. And so with the mass spectrometer experiment, that was actually able to be done in a lab. And so Lee’s lab did a series of very careful experiments looking at non-living and living systems, fossilized systems, blinded samples from NASA that included meteoric material that was supposed to confound the experiment. They didn’t – obviously the experimenters didn’t know what the blinded material was when they got it. They had to classify it correctly. And what they found was all of the living samples were the only ones that had molecules present in abundance that had an assembly index above 15. And so this was, you know, like the key confirmation of the conjecture that life is the only thing that can build complex objects, but rigorously measured and tested empirically in the lab.
Jim: That is a very important and interesting idea. Right? And I wanna slow down here a little bit because I think this is hugely important.
Sara: Yeah. No, it’s… I think this experiment is one of them. If it pans out, it’s one of the most important experiments done in the last century. And I think it really is not, it’s not understood how significant…
Jim: So let me ask you some questions to make sure I understand how significant – maybe I’m wrong, but I might as well, since I’m talking to the person, I can clarify myself and also for the audience. You know, first, this basically essentially is a pivot on the definition of what life is, right? Because you know, you read your science fiction, know, oh there could be silicon life, there could be life inside the sun or whatever. In this statement, life could be anything that produces high complexity in its output or high assembly index number. And that doesn’t necessarily even have to be molecules. Could be something else. Know, if you went to the asteroid Ceres and saw Hogwarts sitting on a boulder, would presume that that was an indication of life somewhere nearby.
Sara: That’s right.
Jim: Or at least had been in the past. And so it makes no claims about the substrate. It makes no claims about the products. It makes a pretty abstract claim about how hard it is for something to have been put together at random where it’s not – I’d say not quite zero but so astronomically small that if the whole universe was made out of computronium you couldn’t calculate a simulation to create it in the history of the universe. It’s effectively impossible. And that’s a pretty strong pivot.
Sara: Yeah, it is. It’s interesting because, you know, like the way the research program goes is you want to solve the origin of life. So this goes back to what we were talking about earlier. The question is not unifying existing theories. It’s like there’s a concrete problem we haven’t solved yet. How do we solve that? There’s a thought experiment about how to do the measurement. And then that measurement gives rise to a new definition of life. And this is actually, I think, the way that it works. You build new theories and your theories tell you how to talk about the natural world when they’re good theories. And so the definition of life that emerges is there are common like some spaces of combinations of objects like chemistry where the numbers of possible configurations are so astronomically large that they can’t all be computed. And if we see objects in this space that’s so large that it’s not computable, basically, and we can tell that by its high assembly index and that we observe it in abundance, we know that that’s a living system. And so living systems are actually, you know, the definition that would emerge – I guess, change my mind on definitions of life all the time, but I’ll just throw one out today for purposes of discussion – is physical systems that traverse these high spaces that are actually not computable, these high combinatorial spaces. So I think that’s cool. And then the other thing I just wanted to make a comment because you were so good on the philosophy and about talking about things hard to vary. And I think you’ll appreciate this because complexity measures are a dime a dozen.
Jim: We did a show on them with Seth Lloyd. We went through 35 of them.
Sara: Oh, Seth’s great. Yeah. He was just at one of my recent workshops and we had a lot of fun chatting about assembly theory. One of his measures, thermodynamic depth, was kind of moving in the direction that I think assembly does a much better job.
There were a lot of debates in the field when we introduced the ideas of assembly theory, with people saying “we already have measures of complexity” and “it’s not any better than any of these other measures” and “why do you say it’s a foundational theory of physics and why do you have to add all this gloss language?” I think they’re really confusing what we’re actually trying to do. Most of it stems from the science of measurement being critically important and also the philosophy underlying the theory.
A lot of these measures of complexity, the 35 varieties that Seth might have talked with you about, have been developed in specific contexts or to develop specific kinds of notions of complexity in particular systems. They’re very label dependent, so it depends on how you encode your data about the description of the system. Whereas Assembly Index is supposed to be derived from measurement, so it has to be something that you can physically measure. It has this structure of being recursive, built on past parts because it’s supposed to capture a minimum amount of causation in an object.
If you want to maintain the features of the theory, you can’t vary the assembly index and use a different complexity measure in its place because that would actually change a lot of the structure of the theory and what abstractions it’s trying to capture about the physical world. This point is really hard for some people to understand because they just say, “Oh, you have an exponential to assembly index when you’re talking about the complexity of configurations of matter. You could just put conglomerate complexity in there or LZ compression or my favorite complexity measure.” Actually, no, you can’t because that equation we talk about in our Nature paper, called the Assembly Equation, has a term related to exponential assembly index, which is capturing this fact that the space of possibilities is exponentially growing, but it’s doing so in a way that you can talk about the structure of the space through this rebuilding and reusing of parts in a way that’s anchored physically and is measurable.
I’ve just gotten into so many interesting debates about philosophy of science where it’s very clear – when we learn about science in textbooks and even when we’re practicing science, unless we go all the way down to the foundations, we forget how much of science is built on certain kinds of structures of assumptions and relations between objects. I mean, mathematical objects and experimental implementation of them. Super interesting. Sorry, that was a side tangent.
Jim: That was good. Very interesting.
Sara: I’m really fascinated by this. Like, it’s just totally blown my mind.
Jim: The other thing that blew my mind in a small way, but it was an interesting one, is the concept of copy number, right? Because lots of things can happen. As I was thinking – I’d never gone down this road before, but when the book took me down this road – I said, copy number, let’s say something interesting. And I came up with the following thought experiment that I could probably program up. One of my backgrounds is in evolutionary computing.
Sara: That’s a great field.
Jim: Strong background in evolution. One could come up with some a-life thingy or other and play the game of writing useful DNA or something, right? And then take the evolved DNA and send it to one of these services that will print DNA for you, right? And I could produce a string of DNA that would have a very high assembly number, right? However, there’s only one of them or small number as much as I can afford to have them print. And very importantly, it doesn’t reproduce. So your copy number concept trims that loophole essentially and says that it’s got to be found – at one point you say more than two, but I would suggest that if you really want to avoid the evolutionary computing hack, you have to say in large number, right?
Sara: Yeah, we do say in large number. Actually, it depends on the limits of your measuring device. So in the real physical world, it’s actually really hard to measure duplicate structures. This was actually a point that was lost on the early experiments with the life detection using assembly theory because in Lee’s mind, copy number was very obvious, but he didn’t explicitly account for it in the way he talked about the experiments because the experiment of measuring a molecule in a mass spectrometer already needs 10,000 copies of the molecule to even get a measurement. And so it was just always there. So he didn’t worry about making it explicit to people. And then all these people really didn’t understand the argument, and then it just became very apparent that you have to be very explicit about copy number. So he always understood it, but I think it took a little while for the rest of us to really get it.
But I think the issue of copy number is incredibly subtle and very deep. And I think one of the reasons it’s really hard for us to recognize that it’s difficult to get a complex object in an abundance is because we live on a living world, which has tons of evolutionary infrastructure constructing all of these objects in our environment all the time. And so it might be easy to think that these things can happen spontaneously or they can happen repeatedly, but pretty much everything in your environment is built by another evolved system and it has to be kept rebuilt in order to persist in time. And that’s one of the key arguments of assembly theory – that these structures do not exist permanently and they don’t come into existence unless they have this evolutionary architecture. And I think that’s a really radical shift. And it’s also something else that the history of complexity science, and you’ll know this well given working on these A-life systems, has had a challenge determining whether something’s random or complex. Right? Because most measures of complexity from computer science, not all of them because people have recognized this problem and tried to address it, but most of them inherently cannot tell the difference between a random structure and a complex structure. And assembly theory has a pretty easy time doing that because of this feature of copy number and thinking about the physicality of these things, not just thinking about them as computer programs.
Jim: Yeah, I got to say I automatically reject all theories of complexity that can’t tell the difference between complexity and static.
Sara: I know it’s terrible.
Jim: They’re just not talking about complexity in an interesting fashion.
Sara: Yeah. And actually, I think there’s a big misconception. I confront this all the time. So I think when you work on life as foundationally as I do, and you really have to drop a lot of the assumptions you have about what we are and what we imprint on the world, it’s shocking to me how many things we take for granted that we don’t realize are like granted being just a universal thing that happens in physical reality. We don’t realize how much of our cognitive architecture and representation in our mind we actually impose on the world.
And one of the places I think this is very apparent to me right now, or at least I’m experimenting with, is the notion of randomness. I think randomness in some sense is a human invention because it’s actually a mathematical structure. It comes out of the theory of computation to define it precisely. And somehow we think random processes happen in nature, but randomness is very different than undetermined. And undetermined dynamics are what happens in quantum mechanics. It’s not even that they’re random, it’s just they’re completely not deterministic until a measurement is taken. So there’s all these kind of weird places where we impose these things. So I actually think randomness is a techno signature. Like it requires a certain level of sophistication of mathematical representation and algorithmic development in order to build a truly random structure.
Jim: Well, yet another rabbit hole we could spend two days on. I know. But hey, we’ve spent a lot of time setting the foundation here.
Sara: Sure. Yeah. Alright.
Jim: So I think I now turn to have Sara apply her ideas about information and her ideas about assembly theory and bring them to bear on the question of abiogenesis or the emergence of life from non-life.
Sara: Yeah, so we are working very actively on this. The first part of your question I’ll address is this philosophy of information being causal and then this idea that assembly index captures minimal causation in objects. They might seem very different, but I think they’re actually the same idea. When you look at it from the perspective of assembly theory, suddenly you’re talking about something physical and measurable, so you’ve built a theory of physics that actually accounts for the properties that we call informational.
The way that I think about that structure and philosophy for the theory and its conceptual interpretation is to say we can go in the lab and we can measure assembly index of molecules. The assembly index is capturing something about the minimum causation, minimum amount of informational constraints to construct the object. Because it’s measurable and we think it’s intrinsic to the objects – like the same molecule will have the same assembly index if it’s on Earth or if it’s on Enceladus. The Hogwarts LEGO castle will have the same assembly index if it’s here or if it’s in some alien spaceship somewhere. It’s a feature of the object just like electric charge is a feature of an electron.
If you adopt that property, then it means that complex objects actually have a size in time. They have a minimum causal size that’s measurable and exists with them as a feature that you can measure from the outside independent of where they are in the universe. This basically implies that time or information extended over time becomes a physical attribute. I think this is one of the most radical suggestions of assembly theory – to think that time is a material property.
For me it explains a lot about the kind of structures evolution builds and also why by the time you get to things like us, the world looks so abstract because we’re so deep in time, we’re so deep in this level of causation. Most of the things we interact with in our environment are also deep in time. Yet we see everything only in sort of the spatial dimensions, so we don’t see how complex all these objects are in time. But we’re kind of existing in this very large causal structure that is our biosphere.
That leads to a lot of really interesting reframings. My philosophy is objects look more informational if they’re bigger in time than we are and they look more physical if they’re smaller in time. Because basically if you can bound the amount of causation in an object, it can look very physical to you. But human societies are very large – they seem very abstract, but they’re nonetheless physical. It’s just about scales of time. And that’s how I account for emergence also – emergent properties are really just how much evolutionary time and selection are in the structure.
When you get to the origin of life, what we want to be able to define is a transition, and we’re thinking about it like a phase transition in the assembly space. The assembly space is just the way we talk about the physical space of all these assembled objects. You talk about their coordinates of where objects or configurations of matter are located in assembly space in terms of their assembly index and copy number. We expect there to be a phase transition where if you don’t have any selection, there’s no constraints on the system, no object is constraining the formation of other objects. You just get a combinatorial explosion, like what we described in the Miller-Urey experiment or in the meteorite samples.
If you have molecules that have molecular recognition and they basically can allow certain molecules to persist over time longer than others, and they might have some causal feedback and actually allow persistence of lower assembly structures, you can get an abrupt phase transition to more complex states of matter. We expect this to create a cascade of complexification where you’re increasing the assembly of structures over time. We have that almost worked out theoretically – I’m working on it with Lee and a really talented student in my lab. But it’s a bit tricky to work out all the mathematical details because these are really high dimensional theories.
That’s the conjecture – that this idea of this abrupt transition where this is complex enough to be life and this is not complex enough, there’s a boundary that we measured at being 15 in the experiments, we think we can predict that. And not only can we predict it for molecules, but we’ll be able to predict it for any kind of physical material. By physical material, I mean quite broadly, any material. So it could be human language and algorithms – if you want to know if an artificial algorithm is alive or not, or you want to know if a mineral is a product of evolution or not, or you want to know about some organizational state of some material on the surface of a planet.
The basic program we’re undergoing right now is to build a generalized theory of where this complexity threshold is and why it arises. I think we have it pretty well worked out also for exoplanet atmospheres and thinking about atmospheric physics. I think we can identify biosignatures pretty robustly. The planetary organization of atmospheric chemistry from an assembly theory perspective can pick out Earth as compared to other planets very easily.
Jim: Some questions for you. Just a thought when you mentioned minerals, have you guys run the numbers on limestone? You know, limestone doesn’t exist unless there’s life, right? Because the materials get processed through mollusks, etcetera. You lay out huge piles of mollusk shells on the bottom, they get covered over by rock, they get compressed. And then they eventually become marble or dolomite if they’re compressed more. I’m just curious. I don’t have any idea. Something else?
Sara: We haven’t looked at limestone specifically, but we do have a paper on assembly theory applied to minerals that’s in review right now. I can tell you about how it works for minerals. It’s actually super interesting because it’s a bit different than molecules. And I think it gets into some interesting features of the generality of theory.
But minerals, we talked about also when we were talking about Schrödinger, we’re pretty good at talking about periodic structures, right? So you have a unit cell in a mineral, which is like the configuration of atoms in one of the units, and then you have a repeated structure in the mineral, and that defines the properties of the mineral. It turns out, if you want to do assembly theory on minerals, you have to look at the structure of defects in the material because the defects determine the copy number of the macroscale organization. So if you want to go to the unit cell, you might have five unit cells repeat defect, three unit cells defect, four unit cells defect, but it’s kind of random.
And so you can talk about assembly in these materials by looking at assembly index of the repeated units and the copy number. Now, if you go to a technologically engineered material, like a silicon chip, the defects are no longer random, they’re precision laid on the surface of the material. And therefore you have a very high copy number at a macro scale of organization and the assembly jumps up. So the same abrupt transition we see in chemical space where you have abiotic systems, low assembly, when you account for assembly index and copy number, living systems, high assembly.
You can see that in minerals, but you can use it to detect technologically engineered minerals. And so we haven’t really thought about minerals. I’m working on a project to actually look at mineral evolution over the course of life on Earth, looking at things like limestone, but also just the mineral diversity that’s been created by our planet because we have a biosphere and trying to actually build a phylogeny of minerals using assembly theory. But that project’s still in early stages of development, but we have been able to work out this issue of technosignatures and actually how to quantify assembly of minerals. But you actually have to look at the bulk material again, and it depends on what you measure. And, you know, like this issue of defects comes from the measurement, like about the diffraction patterns you actually get from the material. So it’s a very nontrivial thing to embed these questions in measurement and the actual physicality material.
Jim: I’d be very interested in it. I’d like to read that paper actually on-
Sara: I’ll send it to you.
Jim: Yeah. Minerals. That’s just interesting. I don’t know why it treats me, but it does. So now let’s focus a little bit more sharply on applying your ideas to some of the more traditional ideas about origins of life. Does your lens give you any insights into some of the competing theories from the more traditional side? For instance, the warm pond that evaporates and refills, you know, the Damer-type theories versus the blue smoker, black smoker theories that life evolved at the bottom of the ocean around hydrogen sulfide vents.
Then Harold Morowitz, who was my first mentor in complexity science – I used to drop in once a week and Harold and I would just shoot the breeze and he’d write on the whiteboard and I’d go, “Damn, that’s some interesting stuff!” He has a conjecture, I guess is a better word, that because we’ve been unable to explore the chemistry (and he’s a physical chemist by background), when unable to explore chemistry very extensively at high temperature and pressure, he suggests that maybe life could only evolve deep under the earth under very high temperature and pressure and then permeated back up from below to the surface. And we do know that life exists as far as a mile underground, maybe further. So there may be something to that story. Anyway, does your lens – is it any way helpful to think about these other theories that came from different directions?
Sara: Yeah, I think so. But my perspective on it has always been that once we solve the mechanism of the origin of life, then we can ask where it happened. So I’m not quite ready to say I favor a particular environment. In fact, I’ve always been kind of on the side that origins of life is like a planetary-scale phenomenon. It requires a lot of intermixing of microenvironments. I usually think about origin of life like: how did a geosphere lead to a biosphere, not how did this particular environment make the first cell? I think that question is not framed exactly the right way.
But that being said, I think we will be able to say some things about these different environments. For example, a lot of the work that comes from assembly theory might be related to bond stability in different environments. This is not a new argument – people have thought about this – but I think it gives a little more formal weight to it in terms of how this actually leads to complexification. You don’t want an environment that’s like a plasma where you can’t make molecules, but you don’t want to be frozen, where all your bonds are locked in permanent molecular structures. You want some turnover and exploration of the diversity of chemistry.
I think a lot of these environments might provide it. One of the things that we’re doing here – I have one research scientist that works with me, Kurt Robinson – he’s doing some geochemical experiments exploring high temperature and pressure. This would be along the lines of what Morowitz was thinking, looking at how assembly increases for very simple chemical systems. You can predict things on kinetics and thermodynamics, but the complexity regime that you get from assembly doesn’t exactly map to the thermodynamic or kinetic picture of what’s going on. So it suggests these kind of windows of complexity, as he’s calling it, that you might see further exploration of more complex molecules emerge from.
But we’re still in the early stages of those kinds of things. On Lee’s side, he’s got all these experiments running in his lab with these computers where he’s doing these kind of batch explorations of lots of different chemical mixtures, lots of different minerals and automating the exploration of different chemical diversity. So I think our main thinking is it takes a lot of microenvironments to explore the combinatorial space of chemistry to actually get something interesting. I’ve never been a big fan of a single environmental hypothesis for the origin of life, because I think it’s not broad enough to actually understand the mechanism.
Jim: So I’m going to rephrase it. Tell me if I’m wrong, but that you see a prebiotic geo-evolutionary period that has to build up an index number, assembly index, high enough to have the pieces for the phase change to occur and then create life.
Sara: Yes. I think it’s a planetary scale phenomenon, which is why I really think you need to have a sufficient – I almost think of it as like a geochemical combinatorial search engine. And what we’re trying to build in the lab, what Lee’s trying to build, is a chemical search engine. I think people really underestimate how big chemical space is. Actually, one that’s really fun – Michael Lachmann, who is external faculty at the Santa Fe Institute and was there for a number of years, works on assembly theory with us. He did this back-of-the-envelope calculation that if you tried to exhaust the combinatorial space at the assembly depth of the RNA nucleobases, just one molecule of each molecular formula, and you put them on a volume the size of the Earth, the planet would collapse to a black hole.
So there’s just so many molecules that a planet can generate. It’s impossible to really think about that search space. I think what you need is a very – I’m going to use the word “fertile” just because we’re thinking about the origin of life – but a very fertile geochemical environment that’s just exploring a lot, but it allows things to persist long enough to interact. And if you have enough of that going on, I think the origin of life will happen. But the questions we don’t know are how long and how much volume of material. And this is one of the reasons that Lee and I are really keen on building large-scale experiments, a little bit like the international collaborations for particle physics, where you have to build a really large experiment, explore a large volume of energy conditions and particle configurations to potentially discover new physics. We think the same thing is true in chemistry, that we just need to build a large experiment and search chemical space and see when it gets complex.
Jim: One of the problems with prebiotic evolution is the error catastrophe, right? And we know from evolutionary computing and lots of other places that when you don’t have the kind of high-fidelity information preservation between rounds that biology found – and in fact how biology actually evolved to that point without having it is one of the most, to my mind, mind-boggling questions, right at the origin of life boundary essentially. Have you guys done any evolutionary simulations to see if the error catastrophe kind of rules out geo-evolutionary preposure?
Sara: It’s much more interesting than that. So I have a really talented exchange student from Spain right now. She’s a PhD student working in my lab last couple months. We’ve been working on exactly that problem. And I think we’re starting to write a paper on it. Basically, the error catastrophes have a really interesting structure in assembly theory. And I think it’s not so much of a problem. I think the thing that I really like about the assembly theory framework is once you embed some of these insights that we had like the error catastrophe from Eigen or the idea of autocatalytic sets from Stuart Kauffman but you embed them in the structure of the assembly space, and you really try to think about the causal structure underlying the formation of these molecules, I think you solve a lot of the problems that these things had. So with the error threshold, because of this sort of structure of causal feedback, it’s not, I used the word brittle before. It’s just like, it doesn’t, like the systems aren’t necessarily going to collapse just like they would in what Eigen was seeing, or like the way that we see it. And it doesn’t necessarily require template information either. So you can talk about information in a much broader way. Like all molecules carry information and specificity in them about what they interact with. And that was also one of the problems with Kauffman’s picture is the idea of autocatalytic sets is brilliant, but you have to basically define all your molecules reactions they’re catalyzing. And then you’ll get these closed networks that, you know, like if you have enough interactions, you’ll have closed loops. But if you try to actually implement that in the lab, you don’t actually end up having enough exploration of the space to get those closed loops. And these systems aren’t very stable unless you fine tune and engineer them. And assembly theory allows having that kind of structure form spontaneously by allowing an error basically. So the error threshold helps autocatalytic sets form, which I think is super interesting.
Jim: Ponder. I’d love to see that paper.
Sara: We’re working on it right now and I’m really excited about it, but I’m not like – I want like, it’s one thing. Like, I’m always like, so like, I’m gonna tell you everything. But like, I think we’re just gonna flesh out a couple details and then but I’m happy to send it to you when it’s ready, but I’m super excited about it.
Jim: This is so close to my own core interest, it’s like scary. If you could actually find a way to understand how we bootstrap through error catastrophe to get to life, that would be like, whoa.
Sara: I think the mechanism is pretty clear in assembly theory. I’ve been thinking about it for a number of years, I think like I finally Lee and I have been going back and forth about it, lots of conversations on this topic, and I think we’re pretty close to having it fleshed out. We have some really promising toy models that explain some of the ideas, but we’ll see. And then it has to work experimentally is always the thing.
Jim: That’s cool about you teaming up with Lee. You’re a theorist, and he is about the most practical guy you can imagine.
Sara: Well, he is. He’s a really good theorist, though. I think people under appreciate that, but he is extremely pragmatic. So I think one of the biggest transformations in my career was when I started working with him. And he will not let me bullshit at all. I am not allowed to say anything unless I have a concrete idea about what I’m talking about and I think that’s phenomenal. I know very few people like him that are just like, it has to work. It can’t just be an idea in your head. It has to work in reality. And I’m just like, “Oh, well that’s hard.” So it’s really fun working with him. And he’s also a really deep thinker. Yeah, he’s very, very pragmatic and it’s painful sometimes.
Jim: That kind of pain is good.
Sara: It is good pain.
Jim: Especially for theoretical physicists who sometimes fly off into the sky.
Sara: No, that’s true. And I thought, when I was starting as a PhD student working at a cosmology group, I just thought I would never have ideas that were testable. And I think that was, you know, like I’m really glad that I met the people I did in my career and that things have gone the way they did because there are a lot of scientists that never get that opportunity. And I think that’s kind of a loss.
Jim: You mentioned Stuart Kauffman and his autocatalytic networks a few times. And, you know, again, another one of my gateways into this field was the second book I read in what I could broadly call complexity science was Stuart’s “Origins of Order,” which very few people have actually read. But it is a book well worth reading even today. What’s interesting is then he not only talks about the autocatalytic networks, I don’t remember if it was in “Origins of Order” or his later book. He also then adds, I think, the very important constraint that in biological systems, we’re also talking about cycles of work that have to be able to produce enough energy and something or other that keeps the autocatalytic networks running.
Sara: Yeah. In his more recent work, he’s talked about that in terms of constraint closure. And I think this is also a very powerful idea. But I think that’s actually what is like if you think about the assembly space, every time that you do one of these operations, sticking two pieces together to assemble objects, you have potential for error. So you have to have constraint closure to get any kind of structure being built in assembly theory. I actually have thought a lot about Stu’s more recent writing on that topic. I think he brought up a really important issue that needs to be addressed, but I think it comes from causal closure. And so once you have a theory about causation molecules, it gets easier to talk about these kind of properties. I think without the right abstraction, you know, you can identify that this needs to be solved, but you don’t have the math to do it. So that was cool. But, yeah, that book was amazing. I love that book. I used to read it all the time when I was in grad school. Was carrying around with me for quite a while.
Jim: That’s what you mentioned because I carried around me for a very long time. And I was carrying around the Santa Fe Institute one time, Mr. Said, “Oh, look, you got that book.” And I came up with an off-the-top comment. And I said, whenever a book is open, it’s still alive because somebody is interacting with it. If it’s just on the library shelf…
Sara: It’s dead.
Jim: It’s dead. So so that, at least at the moment, your book is still alive. You have at least one reader.
Sara: So funny. I love that idea, though.
Jim: So let’s take another turn here. You talked about a sharp phase transition. What can you tell us about what you guys think so far about what the nature of that phase transition might be?
Sara: So that’s related to the work I was talking about with the error threshold, but also we’re working on formalizing the phase transition. And I think it is very abrupt in aqueous organic chemistry because the space of molecules is so large. The idea being that this combinatorial explosion of exploring different configurations of matter in the form of molecules – basically you can saturate the space really easily and you won’t make molecules in a detectable abundance. That’s what we get in meteorites or Miller-Urey. But we think that if you have some kind of, I’ll call it causal closure or constraint closure, like Stu would say, you can pass that threshold where you wouldn’t observe any objects in high abundance and suddenly see much more complex objects in high abundance.
So that’s what we mean by the phase transition. Most of that’s worked out but not out in papers yet because there’s some details that are a little bit hard. But I’m really excited about that. Assembly theory basically has three really well-defined substrates worked out so far. They’re all kind of chemistry related, and I’m trying to move into spaces that are not so chemistry related also, but I think getting the foundations clear first is super important. One of them is aqueous chemistry on Earth, which was the life detection experiment. The second one is the minerals we talked about. And the third one is planetary atmospheres, which I mentioned briefly.
Planetary atmospheres are super interesting complex because molecules that are volatile and can persist in an atmosphere are not complex. This is like the whole issue with this major announcement of life detection on an exoplanet. DMS has an assembly index of three if you count the hydrogens in the molecule. So it’s not complex and it’s definitely not a definitive biosignature because we know abiotic processes that can make that molecule. It just happens that on Earth the dominant mechanism is a metabolism. So it’s by no means a good biosignature in my book on any level, and I think that it’s not a verifiable hypothesis that life would be persisting on this planet creating DMS. I don’t think there’s any way in the future to test that. In part it goes from the combinatorial space of abiotic systems being so large, you actually can’t computationally explore all the hypotheses for this planet. And you need a really good robust hypothesis about why only life can make this on this planet, and there is no such hypothesis. So sorry, I had to go on my rant about exoplanet biosignatures.
Jim: That was very interesting, very important, and very timely.
Sara: Yeah. Because I’ve worked on exoplanet atmospheres for a long time. There’s a lot of excellent people in that field. It’s a very hard problem. But anyway, atmospheres can’t have complex molecules in them. There’s probably about 14,000 molecules that – actually Sara Seager did this work – trying to identify what molecules could be in planetary atmospheres. So it’s not a huge number. If you’re thinking about the combinatorial explosion of chemical space, 14,000 is very small, right? Like there’s estimates of like 10 to the 60 molecules that are about the size of amino acid, but like 14,000 can be in an atmosphere. Okay, so it’s a pretty small number.
And so what we’ve been doing is actually looking at the assembly of the atmosphere as a whole. So if you think about all the molecules in the atmosphere and you say, how hard is it for evolution to co-construct this set of molecules in the atmosphere? And we only do ones that are detectable in high abundance, so they have to be above a certain parts per million threshold even for Earth. And then you compare all the exoplanet types we know and you compare atmospheres in the solar system, you compare Earth, it turns out Earth sticks out a lot in terms of having a very high molecular diversity and a particular structure to the relationship between the molecules as codified in the assembly space.
But it’s weird because it’s not like a very – the plot is kind of like you have all of the sort of solar system planets on a line and then Earth is like a little deeper in the assembly space than the other planets. But I think it’s past a phase boundary, but it’s really hard because I think it’s not going to be a sharp, abrupt transition in planetary atmospheres because there’s only 14,000 molecules. The phase transition’s kind of flat and it can’t just have like an explosive transition to complexity. It’s kind of – there’s just not a lot of room to be complex, so to speak.
Jim: Does that basically say that exoatmosphere research is probably a dead end?
Sara: No, I don’t. I actually – so it’s very funny. Two of my former PhD students, Cole Mathis and Harrison Smith, wrote this paper about how exoplanet biosignatures are futile, and I was very proud of them. I think I mentioned it in the book actually. Because of course what you want your PhD students when they go off and do their careers to do is to go against the grain and totally question the foundations of their own fields, right? So I was very proud of them, but their point was exactly that they don’t think that in the near term there’s going to be any hope of exoplanet biosignatures because the challenges are so hard.
I’m much more optimistic, and my current PhD student, Estelle Jeanine, who’s working on the exoplanet atmospheres, is also more optimistic. But I think her results are actually showing you can tell the difference between Earth and other planetary atmospheres we know. And so whether that means that we have a generalized biosignature or not is another question. But what we did not do in this work was put Earth in. We got Earth out. And this is very different than all exoplanet biosignature work I know to date because everyone basically tries to take the pale blue dot, build a biosignature off of it, and then try to look for that biosignature elsewhere. We said, let’s look at assembly theory, take the principles of the theory, and apply it to exoplanet atmospheres or atmospheres generally and see what we get. And it does turn out that Earth is different in the ways we would expect a living world to be different from the foundational principles of assembly theory.
Jim: And do we think that we can detect a sufficient level of precision in exoplanet atmosphere?
Sara: We can actually. So this is the other thing. There’s been a lot of interest in doing complexity science for exoplanet atmospheres. So I, for a while, was doing network science and atmospheres, but I knew that was a stepping stone, not a permanent solution to the problem because you can’t detect an exoplanet network. Like you can’t detect an atmosphere network. We put so many assumptions into our models about the kinetics of these atmospheres and everything else. And you can’t be confident that you have a surface flux of a certain amount and all these kinds of issues that people have in the field. And then there are some other people that are trying to apply information theoretic approaches or using even Jim Crutchfield’s causal state theory on exoplanets. None of these things are detectable with a telescope. Assembly theory is detectable with a telescope, so another feature that’s going to be in Astell’s paper is actually we’ve done a lot of work on infrared spectroscopy and how you can actually infer this property using infrared spectroscopy of exoplanet atmospheres. So it is a complexity measure that we can detect remotely, and then the question becomes about the observational limits and how much we can infer, but we’re working on it. I’m actually optimistic that there is a path to measuring life on exoplanets using assembly theory.
Jim: That will be very cool.
Sara: I know. I’d be really excited. We’ll see what happens, but we’re working on it.
Jim: We’ve been pretty damn grounded so far. This last one, very common for me to ask people working in related fields on the show, Fermi Paradox.
Sara: Oh, sure.
Jim: What are your guesses? And feel free to speculate maddeningly if you are so inclined.
Sara: Okay.
Jim: Alright. To remind the audience, we talk about this all the time, so most will know Fermi paradox comes from a story from Los Alamos or maybe not, but that’s the legend that a bunch of physicists were sitting around speculating about how many tens of thousands of intelligent species are there out there in the Milky Way. And Fermi walked by and said, “Where are they?” And everyone got quiet for a moment and that was the paradox. If there are a bunch, why don’t we see any sign of them? And of course, now we’ve thought about it more deeply. Maybe they’re not there. Maybe they’re there, but we can’t see them for all kinds of odd reasons. I had a very cool conversation with Krakauer on the podcast once about his views on that. So anyway, Sara Walker, your take on the Fermi Paradox.
Sara: Yeah. So I did write about this in the book, too, and I don’t think my opinion has radically changed since then. But as far as what I think – what’s nice about the Fermi Paradox is you take what you think is interesting from the question, right? So my answer is kind of like the most obvious one that people already cite, but I think the reason behind it is a little bit different, which is that we haven’t seen alien life just because we don’t know what we’re looking for. But I mean that in a pretty deep way in the sense that we’re an evolutionary system that evolved on one planet and we have certain sense perception of the world. So like we can see and we can hear. But we’ve also built technologies that extend our sensory perception. So like the LIGO experiment allows us to detect gravitational waves. We have telescopes that allow us to see deep in the universe. We have microscopes that allow us to see all of the microbial diversity that we didn’t know was covering our planet for most of human history.
And so these technologies allow us to see the world in new ways and essentially open our perceptual horizon or our observational horizon to interact with the world. And I think, going back to the conversation on philosophy of science, I think good explanations are some of the best technologies that we build to see the world better. And I think because we don’t have a theory of life that allows us to understand what alien life looks like, we can’t build the technology to allow us to see it. And so I see sort of technology and science as kind of these observational horizon expansions where we can understand and interact with more reality. And we just actually haven’t expanded our ability to do that in a way that we could see alien life.
I think it’s promising that we might get there. And that’s also a very assembly theoretic concept. So a lot of my philosophy of science is also informed by how I think assembly theory frames what an evolving system is. So I have this kind of weird way of, like physicists are usually used to looking at the world from the outside, but to study life, you have to be inside the world, but like act like you can see it from the outside, but be looking at yourself inside the world doing the science. I guess I sort of very viscerally feel and see these sort of horizon extensions and like why the Fermi Paradox is just a natural consequence of the fact that we don’t actually have the technology.
Jim: So you fall to the theory of there’s two big branches in Fermi Paradox opinions, and they’re all just opinions so far, which is one, they ain’t there. The other, they’re there and we can’t see them. So you fall more on they’re there, but we can’t see them.
Sara: We’re there, but we can’t see them. And the fact that we can’t see them is actually part of the structure of what I think life is. So it’s sort of weird connection that, like for me, all of the theories and philosophies I build have to be logically consistent with one another. And so if you think about evolutionary systems becoming more and more complex and being structured deeper and deeper into the assembly space, being deeper and deeper in time, they also build these observational horizons of like sensory perception, maybe for a biological organism or technological perception. If you’re a technological system and also like these explanatory theories that allow putting those sensory perceptions together to actually have an understanding. And that whole system is evolving on our planet at the level of our biosphere and our technosphere. And we’re just not deep enough into our own evolutionary history to see another structure like us.
Jim: Very interesting. Okay, now the final question for me, paradox. Origins of life questions actually are important terms in the famous Drake equation, and then further transitions – one of my favorites is the transition from prokaryotic to eukaryotic. You know, that’s like, how the hell that happened, right? How likely is that? So pure speculation, but you’re a very informed person to do the speculation because you think about it from the bottom and the top, having thought about exo-atmospheres and the phase transition between abiotic and biotic life. How common do you think technological civilizations are in the Milky Way galaxy?
Sara: I think I’m not in a position to actually put a number on that. The most uncertain term in the Drake equation is the origin of life parameter, like the probability of life emerging on a planet. And so I see the whole enterprise of origin of life research constraining that probability. This is one of the reasons I think these kind of large-scale experiments exploring chemical space in the way that Lee originally envisioned it are the most promising experiments to be done. Because even if they’re not successful, they actually allow us to constrain the probability of life forming on planets. It’s the only experimental program I know of that has that capacity.
I think about it a little bit like the Super-Kamiokande experiment in Japan that’s looking for proton decay – they do it by having these huge volumes of water with protons in them and they’re looking for the signatures of the decay process. But the longer they go without observing an event, the longer they know the lifetime of the proton is. And so if you imagine we had like a very large volume of a chemical soup that we were turning over and keeping out of equilibrium and introducing all these minerals and all these geochemical micro-environments and exploring to look for a complexification process measured using assembly, if you didn’t observe any events for five years and you searched a certain volume of space, you know something about the probability – actually constrains it.
And so this unknown parameter in the Drake equation, I’m hoping we’ll be able to solve the origin of life and identify the mechanism and observe an event in the lab. But even if we don’t and we design the experiments right way, we could constrain it. I think all the terms after that in the Drake equation make no sense to talk about in terms of generalities because they’re so specific to life on Earth. I think the idea of technology and civilization, fine, but we don’t actually really know what alien life looks like or how it traverses the space of the complex. And so I would hold off anything about those terms as being meaningful probabilities until we actually understood something about the origin of life and then what kinds of evolutionary processes emerge out of that. Because maybe there is something like multicellularity, but maybe it’s not entirely cellular. It’s like, I don’t even know if these are the right probabilities to impose on alien biospheres. I think something like technology and something like evolution of complexity for sure will be there. But it’s hard to say it’s going to be the same kinds of transitions that we had on Earth.
Jim: Yeah, it kind of does make sense, especially if it turns out, you know, Harold actually conjectured that there was only one mechanism for life, period. Right? It was the reverse citric acid cycle. If you didn’t have that, you didn’t have life now.
Sara: Yeah, spent a lot of time thinking about his work on that actually. I’m trying to do some things with rTCA inspired by his work.
Jim: And if it’s not true, there’s lots of make much more plural architectures than your statement. Certainly true. Anyway, that’s a good hardcore scientist’s answer, must say.
Sara: Well, even I have to say, actually, even if the reverse citric acid cycle was like geochemically convergent on planets in the way that Harold wanted to argue and Eric Smith also argues, at least for Earth, it seems like it’s the one thing that geochemistry could have organized around. I think even the scaffolds that were built out of that into the more complex chemical space of amino acids and RNA and all this other stuff, not very deterministic. So I think even that, and then you have to think there’s these layers of complexity that are built and everyone is moving into an exponentially larger space. So expecting the same path to be taken on every planet, I think suggests a level of biological determinism that I am not ready to prescribe to without evidence, but it could be there.
Jim: Actually, that’s a very nice lens. Sara Walker, author of the very interesting book – you should really read this book for these topics – “Life as No One Knows It.” I got to say this is going to be a classic Jim Rutt Show episode. Thank you very much, Sara, for an amazing conversation.
Sara: It was so fun. Thank you, Jim. Appreciate it.