Transcript of EP 227 – Stuart Kauffman on the Emergence of Life

The following is a rough transcript which has not been revised by The Jim Rutt Show or Stuart Kauffman. Please check with us before using any quotations from this transcript. Thank you.

Jim: Today’s guest is Stuart Kauffman. Stuart was trained as a medical doctor but is best known for his work in developmental genetics, evolutionary theory, theoretical biology, and especially the emergence of order and far from equilibrium complex systems. He was one of the first generation of resident faculty at the Santa Fe Institute, and he’s won a number of awards including a MacArthur fellow and is the author of several interesting and important books. Welcome, Stuart.

Stuart: Well, hi Jim. Thank you for letting me join you in the Shenandoah.

Jim: Yeah, it’s great to have you back. Way back in EP 18, when I barely knew what I was doing, not that I really know what I’m doing now in the podcasting business, I talked with Stuart about his book, A World Beyond Physics, the Emergence and Evolution of Life, a book well worth reading. I should also note that Stuart was more or less one of the two gateway drugs that got me pulled into thinking about complex systems. I went back and looked at my little records of books that I’ve read back in 1998. I read Stuart’s book, origins of Order, a very interesting book, which I can still recommend to this day. It’s not easygoing, but if you really want to get a load of thought about how a order could emerge from non order, this is a wonderful place to start. Though today we’re going to continue on that theme interestingly, in some ways. We’re going to talk about his new paper Co-authored with Andrea Rolly titled, Is The Emergence of Life an Expected Phase Transition in the Evolving Universe? And no doubt we’ll talk about some other things.

So let’s start there. Now, interestingly in the paper, you kind of tip your hand right in the second sentence. You basically say, “Yeah, it is expected sort of.” And that has been consistent with your argument all the way back since the origin of order where you are able to run through the mathematics. And if you make certain assumptions, at least it does seem like we ought to get to at least an auto catalytic set, right?

Stuart: There’s an issue of modern science, and there are the ancient dreams of humanity. The issue of modern science, is the origin of life field has been seriously underway for about 70 years, and it’s really badly fragmented. The RNA world has been dominant for a long time and template replication, RNA and all that. And the alternative view is the metabolism first view. If we had at least, at least one, better more than one. But if we had at least one apparently sensible theory, starting with the Big Bang and leading to life that a fair number of people agreed was sensible, maybe a modest number of origin of life workers could coalesce around a theme and accumulate evidence that might end up saying, yeah, this pathway to the origin of life looks like it might work. So part of the interest in this paper might this article, which will be submitted for publication soon accomplish that.

But more broadly, we have been probably since Neanderthal wondering where did life come from? But that’s a fascinating question. If you were in the year 1620, there was no worry about where life came from. After every rain, there were maggots on wet logs. So the thought was life just emerged spontaneously. So for those of your listeners who don’t know, the problem of the origin of life started with a stir [inaudible 00:03:32] of the milk and a prize was offered to assess whether or not the falling is true. It was known that if you put out a flask with broth in it, after a while it was turbid with bacteria. And the question, where’d the bacteria come from? So pastor did a wonderful experiment. He bent the mouth of the flask into a horizontal S shaped, and in the dip in the S, he filled it with water.

So air from the outside could not get into the broth on the inside, and he cultured it and nothing grew. So he reasoned, whatever it is coming in from the air. And people knew about bacteria and he concluded, life only comes from life. Once you say life only comes from life, you’ve got the problem, how did life start? So pastor started the whole problem, and it was silent for a while. In the early 1900s, Haldane And Oparin came up with the beginning ideas. Oparin thought about coacervates, which were sort of gummy lipid like things. And Haldane invented the idea of organic molecules diluting the ocean, which was called primitive soup. So early in the field you could see cans of primitive soup in lots of talks. So that was around 1920 or 1930. The field was stationary.

And in 1954, Miller and Yuri did the famous Miller Yuri experiments where Sammy Miller took a flask with water in it, and I think three or four kinds of molecules, methane and carbon dioxide and whatever. Put in electric sparks to simulate lightning and three days later, he had a brown scum, and the brown scum had about five or six or seven amino acids. That was thrilling because it said the major molecules of life could form in the prebiotic earth, and that unleashed decades of people trying to make the components of life in weird chemistry experiments. Then the field stayed there. Then in 1954, the same year that the Miller Yuri experiment, Watson Andrick discovered the structure of DNA, which is a template replicating DNA, and people thought, well, I wonder if life started as a double-stranded RNA can replicate like DNA. And it’s a very obvious idea. It’s still a dominant idea, and the field has been worked on since the 1970s.

So people have been trying to get, first of all, just a double-stranded RNA to melt apart and the Watson side to line up the nucleotides for the crick side and link them together. So Watson would make a second crick and crick would make a second Watson, and that never worked. Then people discovered that RNA molecules could also be enzymes, they’re called ribosomes. Then birthed the RNA world, and it has been dominant since 1986. The main idea is still an RNA molecule that can template copy a copy of itself. It hasn’t worked, Jim, it might. A guy in England has an RNA strand that can operate on a different RNA strand and make about 200 nucleotides. So suppose that work, you’d have a nude replicating RNA molecule call that the RNA world.

Quite differently, in 1970 or so, I got to wonder whether or not life was due to the formation of a set of molecules that mutually catalyzed one another’s formation. So I’ll tell you how I got there. There was a book by Melvin Calvin called Chemical Evolution, and in about half a paragraph he said, “Well, what have you had a set of molecules that mutually catalyze one another’s formation out of their building blocks?” I think I got that from him. Maybe I got it by myself. What I wondered at the time was, suppose the concepts of nature were different. So we could have chemical reactions, but we couldn’t make nitrogen or something, so we couldn’t make a double-stranded DNA molecule which would get life. Or would it be ruled out? And I thought, Jim, that’s nuts.

What you need is a set of molecules and a set of reactions, and the molecules in the set catalyze the reactions in the set, which means making them go towards equilibrium faster such that the entire set collectively, mutually catalyzes the formation of everybody in the set, and you have to feed it from the outside with building blocks. So in 1971, I made an initial model of it on a computer and it worked. So here’s the basic idea. It is a phase transition. So let me try to explain the phase transition. It’s pretty lovely. I don’t know if I’ve told you about it before. So in 1959 and 1960, two amazing Hungarians were working on what’s called graph theory. So a graph is just a bunch of dots connected by a bunch of lines. The lines can have errors on them or not. They’re just on directed lines, and I want to get the intuition across. It’s unexpected and it’s beautiful. So imagine you have a hardwood floor and 1000 pairs of buttons on the floor and a big spool of thread.

And the experiment is break off a piece of thread at random, pick up two buttons, tie them together and put them back on the floor. Just keep doing that. Well, we’ve got a thousand buttons and then one thread, then 10 threads, then 100 threads, then 500 threads, then 1000 threads, tying together random pairs of buttons. So here’s their question every now and then, pick up a button and see how many buttons you pick up with it. That thing that you pick up, which we would just call a cluster, it’s called a component of a graph. So here’s the magic, on horizontal X axis, plot the ratio of threads to buttons. On the vertical axis, plot the size of the biggest cluster. So magic happens. You start with 1000 buttons and no threads, then 10 threads. So the ratio is 10 over 1000. Then you have 100 threads. When you get to 500 threads, each thread has two ends.

So the number of ends and threads equals the number of buttons, and you get what’s called a first order phase transition. All of a sudden there’s a huge jump, and you get a giant cluster called the giant component. It’s a first order of phase transition. It’s a jump in the state of the system, and the intuition you should take away from this is that roughly if you connect enough things, all of a sudden the whole bunch of things a little connected sort of no matter how you connect them. So I knew that next step, I wanted to find a chemical reaction graph. So I’ve got some chemicals, whatever they are, and they can undergo reactions. So for example, an A and a B combined to make a C. So I’m going to depict this as follows. I’m going to use dots to represent molecules, and boxes represent reactions. So A is a dot, and B is a dot, and an arrow from A and B go into a box, and then an arrow comes out of the box into C.

So this is what’s called a bipartite graph. It’s got dots and boxes, and every box has arrows coming in and out, and every dot has arrows coming in and out of boxes. So next step. So just have a whole bunch of molecules and a whole bunch of reactions. Think about this reaction graph, Jim, okay? And I want to get across a next idea that we need. Molecules are combinatorial objects. Organic molecules are built out of carbon, hydrogen, nitrogen, phosphorus, sulfur, oxygen. So there’s six building blocks. So if you make a molecule with seven of these atoms, you get some molecule. If I make a molecule with 80 of these atoms, it’ll be a more complex molecule. If it’s a more complex molecule, it can be broken into parts in lots of ways. But all the ways that you can break it into parts are ways that it could be built by going backwards. So as the diversity of molecules and the complexity in terms of numbers of atoms and molecules goes up, the ratio of reactions to molecules goes up

Jim: Exponentially or higher even than maybe, than exponential.

Stuart: But in your mind, think that as the ratio of reactions to molecules goes up, it’s kind of like the ratio of threads to buttons going up. But pull that, that’s just fundamental. So the next thing we need, Jim, is the following. If I knew which molecule catalyzed which reaction, I could just assign it, and I can see it’s got a collectively auto catalytic set. I catalyzed the formation of a second copy of view out of two Jim parts. You catalyze the formation of Stuart out of do Stuart parts. So it’s a very broad theorem that I proved first numerically in 1971, that algebraically in 1986. Then with Don Farmer and Norton Packard and our first numerically numerical experience, and it has been proved over and over and over again, roughly, no matter if you just take a bunch of molecules and a bunch of reactions and you have some rule to assign sort of at random what molecule catalyzes what reaction? Call that some probability of catalysis.

As the ratio of reactions to molecules goes up at some point, so many reactions are catalyzed that the giant component forms and it’s a collectively autocatalytic set. So that’s his first order phase transition.

Jim: Let me pause there and ask you a question. I understand the math and it just reminds me when I first started dabbling in this stuff in 2001 when I retired from business. So first three programs I wrote, one of them was to replicate the buttons and strings and it did work, which is interesting. But as I started reading about the auto catalytic set, with the probabilistic catalyst, the first thing I stopped and said is, wait a minute, actual catalysis is constrained within the physical domain by the shapes of the catalysts. And so merely stipulating a probabilistic catalysis may well be missing a very important function of the fact that catalysis works more or less by physically binding two things at the same time to reduce their collective velocity, if you want to call it that, so that they can react at a lower activation energy. So is the assumption of random catalysis actually grounded? Or is it too much of a simplification? Does the physical constraint of catalysis have a meaningful influence on the network that would evolve?

Stuart: Let’s back up and I’ll tell you some mathematical things. So Mike Steel’s a mathematician in New Zealand, and he and Wim Hordijk have worked on this. Mike has shown you could take your molecules and say, each molecule has the same probability, one in a million of catalyzing each reaction. Or you can make it so that some molecules have a high probability of catalyzing reactions and others have none. It more or less doesn’t matter.

Jim: Get the same similar results.

Stuart: Yeah, you get the same results. Okay, so now, meanwhile, concentrations have to be right. They have to actually interact with one another. Let me jump forward to about three years ago. Nobody’s made replicating DNA in the RNA world. People have made collectively auto catalytic sets experimentally for years, GUTR gutowski [inaudible 00:15:31] made a two DNA sequences A makes a copy of B, and B makes it of A. I owed him a bottle of champagne. Then people have made collectively autocatalytic sets of RNA, Niles layman and his colleagues have. That emerge more or less spontaneously. Critically important, you can make autocatalytic sets of proteins. So [inaudible 00:15:53] did it in 1995 and Gonen Ashkenasy, who was his postdoc now has a set of nine peptides and it’s collectively autocatalytic. So hold that. Meanwhile, they’re all made.

The stunning recent result is about three years ago and last year, there’s an amazing woman named Joana Xavier, she’s from Portugal. She worked with Wim Hordijk Mike Steel, Bill Martin, me, and with her lead, we published a paper in 2018. It’s her work dominantly. It’s quite amazing, Jim. So she looked at a bacterium and archaea each from more than 2 billion years ago, and she looked at only the small molecules. So just looking small molecules like iron and sulfur complexes and little organic molecules. Jim, there are collectively autocatalytic sets of small molecules in the archaea and the bacterium. And as of last year, she’s published a paper, and I’m the second author on it. She’s looked at all 6,700 prokaryotes, bacteria and archaea. All of them have small molecule collectively autocatalytic sets in which there’s no DNA, there’s no RNA, there’s no protein, there’s no polymers. So I think she’s found something fundamental.

Jim: Yeah, I was shocked when I read that in this paper. I had not heard about this result.

Stuart: Most important result in the last 10 years.

Jim: Your paper seemed to imply that that has not been demonstrated in vitro.

Stuart: Correct, and Joana would be careful to say this. This is done computationally all of reactions. So you can ask as a collectively autocatalytic and on paper they are one of the things that has to be done to say, okay, can we actually make it in a test tube and it reproduces? So presumably it will. I mean they’re in cells.

Jim: Are the rates of catalysis sufficient?

Stuart: And the concentrations and all that? But meanwhile, all 6,700 prokaryotes have these.

Jim: Now is it the same or similar small molecule networks? Or are they each different?

Stuart: They’re all different, but they overlap in quite magical ways.

Jim: Ah, okay. That’s interesting. Yeah.

Stuart: So meanwhile, Wim Hordijk and Mike Steele showed something that I never thought of. I thought it was this big giant collective component. It’s not. The big collective component, which is also called a raf, which is another name for collectively autocatalytic set is made up of irreducible autocatalytic sets. So the big thing has lots of tiny things, and suppose you’ve got 100 different ones of these irreducible autocatalytic sets, you can make a bigger thing out of any subset of them. So basically what is going on in the 6,700 is, they’re made up of different subsets of these little irreducible ones that could be investigated computationally, but it means something else that’s important. In a paper by troth [inaudible 00:19:02] Maori and VAAs S realized that a little irreducible autocatalytic set is a replicator. It’s like a gene, but it’s not. That means it can be exchanged and it’s heritable. That means that the 6,700 collective autocatalytic sets that Joana’s found, there’s some kind of funny evolution of metabolisms by exchanging these irreversible autocatalytic sets. It’s wide open to be analyzed. It hasn’t been done.

Jim: It’d be very interesting to think about these as composability elements.

Stuart: That’s exactly what they are, Jim. They’re composability elements.

Jim: And that presumably for them to be interesting is they have to feed between each other. Otherwise they’re totally separable and not particularly interesting.

Stuart: And they interact in all kinds of wonderful ways.

Jim: So set A and set B for them to communicate, quote unquote they have to have at least one element in common.

Stuart: And that’s what happens. So meanwhile, I now think there’s a lot to be said here. I now think that the earliest forms of molecular reproduction in the universe must be these small molecule autocatalytic sets, and Joana’s found them.

Jim: That would essentially allow for a form of protolife well below the threshold of what we normally think of as life.

Stuart: Yeah, because it doesn’t even have any proteins.

Jim: Yeah, it doesn’t have DNA, doesn’t have proteins, doesn’t have cytochrome, none of those things, right?

Stuart: None of that fancy stuff. So I really think that Joana has put us on the pathway. This is the earliest form of life. So now with that said, it will arise as exactly the phase transition I’ve talked about. So here’s what’s going on in the paper, is the emergence of life and expected phase transition, Jim. Forget theory, Big Bang, hot quark glue on soup, universe cool. We get some, the quarks and the anti quarks do whatever. We wind up with quarks. Then we get hadrons, then we get some atoms, then we get more and more complex atoms. Then we get all 98 stable atoms. Then the universe starts making simple organic molecules and other molecules. Then more and more and more complex molecules. So if you look on the merchants and meteorite that was formed 5 billion years ago, it’s got hundreds of thousands of kinds of molecules of all kinds of complexity. In short, the universe is cooked up an increasing diversity of organic and other molecules.

So even before we do theory, presumably at some diversity, they’ll be sufficiently diverse that if they can interact with one another, you’ll get the phase transition too. I’ll call them Joana sets. So that’s the intuition. It’s the beginning of a theory that starts with a big bang and ends up with the beginnings of life of the sense of small molecule autocatalytic sets. I’ll go beyond them in a minute, Jim.

Jim: So as you allude to earlier, there are issues of concentration and catalysis rates. So this gets us to substrate questions, right? There’s the lipid school of thought that relatively simple fatty acids that have a hydrophobic and a hydroscopic opposite ends will produce a very simply, a relatively useful membrane. Another one is that there are sort of microstructures in black smoker, bottom of the ocean, hot vents that surface as concentrators.

Stuart: And indeed there are.

Jim: I would suggest that for this small molecule catalytic set theory to move from a hypothesis to something closer to a theory, one would have to be able to stipulate believable mechanisms to achieve concentrations and containment.

Stuart: You’re exactly right. So one option is the hydrothermal vents, and there’s these smokers with little cavities in the forming stones, or David Deemer and Bruce Damer like to think of a terrestrial environment where tiny streams that evaporate and not are flowing slowly into something like a little pond and things can concentrate where those little streams are coming together in the soil. Or for example, on what are called mudstones on Mars 3 billion years ago.

Jim: Yeah, I’ve had Bruce Damer on my show three or four times. He’s a very interesting character, and he’s essentially reviving Darwin’s warm little pond. And his containment in some sense is over time rather than in space. He imagines it’s the work being done over time where the little pool evaporates has a ring of scum around it, then it fills back up, another reaction occurs, then it goes down again.

Stuart: You’re right. The idea is that these lipids, when the thing evaporates forms kind of a bilayer lipid slime on say the clay, and meanwhile there’s a bunch of they like I want peptides or RNA, it’s going to turn out to be the same story, and then when it gets wet, they form a little vesicles. Then when it evaporates to go back to being the sort of multi lumen surface on the clay, or if you talk to Nick Lane, you’re using the energy gradients in the black smokers. The fundamental idea I’m bringing is, if you get an increasing diversity of ever more complex molecules, at some point they’ll hit this phase transition and you get a first order transition to autocatalytic sets. You just will. So the mathematics in this theory is, old mathematics on collectively autocatalytic sets. I did the first one in 1971, Don Farmer, Nor Packard, and I did it in 1986.

We made a model in which we had imaginary polymers that could bind to one another with two strings of A’s and B’s and you could glue them together or break them apart. And then we said that each of these strings was a molecule that had some probability of catalyzing each reaction, and we made, well, Don and Norm did it. I was thrilled. We made a model in which you increase the maximum length of polymers that you could have from five to six to seven to 10. As you do, a polymer length 10 can be made by breaking it in nine places. So the lengths of the polymers goes up, the ratio of reactions to molecules goes up and wham, you get the phase transition. And since then a lot of people have worked on it. In particular Wim Hordijk and Mike Steele with all of these results over roughly 50 years.

That’s nailed. Meanwhile, Jim, a friend Jim Harriet and I wrote down a funny equation in 2017. We weren’t thinking chemistry at all. We were trying to be economists. So imagine that you and I and some of our friends are on some sort of region and wherever, and we have 15 goods and services in our little economy. So at time T, we have MT things. So M at time T is 15, MT is the number thing. And I wanted to know how many things will we have in the next period? Well, if these things work, we’ll keep them. But we could try to jury-rig something out of any single thing. Changing the shape of an axe handle. We could try to jury-rig out of any pair of things like the printing press is a recombination of movable type and a wine press or any four things.

The Wright Brothers airplane is a recombination of a light gas engine, a propeller air foil, and bicycle wheels. So if I’ve got 10 things I can try making something new out of any one of things, any pair of things, any three things, any five things or all 10. We wrote down the simplest equation, Jim, it just says the following. Oh, notice that the number of pairs of things is greater than the number of things. If you’ve got 10 things, there’s 10 single things, but there’s 10 times nine over two or 45 pairs of things. So the number of things goes up, the number of pairs of things really goes up fast. So just to take it, if you have 100 things, there’s 4,500 pairs, and if you’ve got 1000 things, there’s, I don’t know, half a million pairs of things. And then the number of triples goes up.

So you wrote down an equation that just says intuitively, it’s easier to make something new out of a single thing, a little bit harder out of a pair of things, a little harder out of three things. But as you make more things, you get to try more and more possible combinations. So we call this or I call the theory of the adjacent possible. And as magical properties, Jim, if you start with not too many things, at first, over a long time, not much happens. So this is crux. All of a sudden it explodes upward faster than exponential. It’s hyperbolic. It reaches infinity in a finite time. All right? I got to come back to it explains why we’re in the Anthropocene right now. But meanwhile, it says the following, it’s a crude model of the evolution of the chemical complexity of the universe. We started making a few kinds of molecules and wham, an awful lot.

So you put the theory together, this tap theory, and you add to it. Well, what if a thing can catalyze the formation of things? So if you’re with me, when you get to sufficient diversity, you get the first order phase transition through the emergence of autocatalytic sets, and that is, that bulk of the paper. And it says that the emergence of autocatalysis in the universe would’ve got to sufficient chemical diversity is expected. It’s a miracle, but it’s expected. So based on this in Joana’s results, I think small molecule reproduction, anyway, it’s going to be abundant in the universe. There’s something like 10 to the 23 stars. So there’s something like 10 to the 23 planets. If one in a million could do this, there’s 10 to the 17th bios out there.

Jim: Listen, the famous Drake equation, and on my show I often talk about the Fermi Paradox and we end up talking about the Drake equation a little bit.

Stuart: Yeah, okay. The Drake equation was trying to work out what’s the probability of life out there for SETI to find.

Jim: Actually not life, and this is an important point. But life at a technological level that it could use radio.

Stuart: Yeah. So if you do that, you have a whole bunch of things and you assume the following, the Drake equation assumes a bunch of different things have to happen, and Jim, they’re independent of one another. Well, the more things have to happen that are independent of one another. Is if they’ve each got some probability. When you multiply them together, it gets really tiny. So the Drake equation has around seven things in it, and we don’t know the probabilities of those things, but the fundamental problem is, you’re multiplying together independent probabilities. The tap process is not independent at all. The molecules that are around enable the formation of more complex molecules. So the whole thing that’s wrong with the Drake equation, it’s not wrong, but the notion that these probabilities are independent of one another, many of them are just not.

Jim: Well, actually none of them are if you think about it, right? Because the one depends upon the other. You’re not going to get complex life unless you have simple life first.

Stuart: Exactly. So the tap equation exhibits, it’s the complexity now enabled the next complexity.

Jim: Which of course is very interesting. Our former colleague at Santa Fe Institute, Brian Arthur, applied the same thinking to technical entrepreneur hood, basically, right?

Stuart: He absolutely did. And I want to get to Brian too in a moment. So we’re not at the third transition. Before we get to that, Jim, let me back up. Let’s take the Joana set. No polymers. So here’s what I really think happens. The Joana sets, they actually make amino acids. They make at least ATP and they link two major pathways of metabolism, redox, reactions and ATP metabolisms. So they’ve already got the core of an energy metabolism. So here’s part of what I think happened, Jim. I think these Joana sets evolved by exchanging irreducible autocatalytic sets or rafts. That’s how metabolism evolved. We haven’t known how metabolism been devolved over the years. I think Joana solved it or Joana and S and VAAs [inaudible 00:31:39] that metabolism evolved by exchanging these little autocatalytic sets. By the way, what did they eat? So there’s a guy named Philippe Ning.

Philippe Ning in Paris has shown, leave out catalysts. You could have another sense of autocatalytic set in which there’s no catalysts. Let me get the idea across. Suppose that A and B make 2C, C and D make 2E, E and F make 2A. There’s no catalyst, but there’s a cycle, A, C, E. If you start with some A, C, E and you feed it with B, D, E you make a lot of A, C, E. There’s no catalyst. The structure of the reactions is the catalyst. So Philippe and his colleagues a few years ago called these autocatalytic motifs, there’s five different autocatalytic motifs. So part of what can happen, by the way, the TCA cycle is an example. It’s autocatalytic if you run it in reverse is maybe these Ning motifs were the food, those small molecule autocatalytics ate that eat something.

Anyway, then what? So now [inaudible 00:32:54] and I have written papers on this. So we’re imagining the following. People have made peptide autocatalytic sets. People have made RNA autocatalytic sets. And it’s the same phase transition. Whether it’s RNA or peptides. So suppose you’ve got a system that’s now made some peptide autocatalytic sets in some RNA autocatalytic sets, can they help one another reproduce? Well, they can. Here’s the issue. If you have an RNA autocatalytic set, you make a double-stranded RNA at some point. They don’t like to melt. The consequences, you get sub exponential reproduction. If you could have the double strand melt, it could reproduce faster. So imagine you’ve got a double-stranded RNA, you say 20 nucleotides along. And it can make a stem loop on each strand, the RNA, instead of staying double-stranded, it’s a sequence in which it can fold up and make a little thing that sticks out from this strand. And typically it has a little loop at the end.

So suppose remember As for RNA, A, U, C and G. C binds G and A binds U. So real simple, let’s have it sequences, A, A, A, A, A, A, A, U, U, U, U, U, U, U. But it’s on one strand and opposite is a U, U, U, A, A, A so it’s double-stranded. But you’ve got an A, A, A, A, A, and a U, U, U, U, U, on the same strand, so it could buckle out and the A’S and the Us on the same strand to bind and you get a stem, and that’s a STEM and RNA do it all the time. Well, okay, so suppose that you have a peptide or two peptides and the peptides can bind these stems. Well, it’ll help the double-stranded form melt by stabilizing the strands. So the peptides can help the RNA. Now imagine that you have an RNA strand with two of these loops on it. If peptide one binds one and peptide two binds the other, the RNA molecule is acting like a little ligase, keeping the two peptides near one another, it can catalyze their formation.

So it’s easy to imagine an RNA peptide collectively autocatalytic set. And so then the dream is that the small molecule set becomes the metabolism and they’re coupled to this new thing and they’re already producing together. So Jim, I want to jump to an idea due to Immanuel Kant in 1790. He had a marvelous idea that’s lane dormant. I picked it up about 20 years. So it’s really important. I’m going to get across three things that are really important. Kant said, thinking of organisms, an organized being has the property that the parts exist in the universe foreign by means of the whole. So the parts exist, part of the whole. The whole exists because they have the parts. So you’re a Kantian whole. You exist you’ve got a liver and a kidney and a spleen. They exist because they’re part of you.

Jim: And of course that is the sort of paradox of life that it is self-generating essentially, right? It’s obviously not a paradox. There are no paradoxes in the actual universe, but it’s a very difficult concept.

Stuart: Make the step to a Kantian whole. Okay?

Jim: We can stipulate that it’s fair to say we are Kantian whole.

Stuart: Every living thing is a Kantian whole, turns out.

Jim: And very few things or maybe nothing that is not a living thing is a Kantian whole.

Stuart: Exactly. A crystal. The definition is the parts only exist part of the whole and vice.

Jim: Now, I believe in your paper, you guys put forth that therefore a virus is a Kantian whole.

Stuart: In the context of a living cell.

Jim: Okay.

Stuart: A crystal isn’t a Kantian whole. The atoms in the crystal can exist without being part of the crystal. A snowflake’s not a Kantian whole. The water molecules can exist. A pile of bricks is not a Kantian whole, the bricks can exist. Most things aren’t Kantian wholes you are. And of course you’re not equilibrium, you got to eat. So it’s a very important novel structure. Second, we can now define the function of a part, Jim. The function of your heart is to pump blood, not make heart sounds. That means something very specific that we’re going to need. The function of a part is that subset of its causal properties that sustains the whole. For your heart it’s pumping blood, not making heart sounds or jiggling water, right?

Jim: Or burning ATP, which it does.

Stuart: Or burning ATP, right?

Jim: Which it does in huge quantities, right?

Stuart: Right. So there’s really important, we have a non-circular definition of the function of a part. So function is not mysterious any longer. In Gonen Ashkenasy autocatalytic peptide set, peptide one binds two half fragments of peptide two and ligates them. So the function of peptide one is to catalyze, the formation of peptide two, not jiggle water. So hold onto it. The function of a part is that subset of its causal properties that sustains the whole. Next step, Jim selection acts at the level of the whole, not its parts. So why do hearts evolve? Well, there was some organisms with early hearts. Those organisms that had genes for better hearts had more offspring, because they survived. So because of selection at the level of the whole, the parts evolve, selection is acting indirectly on the parts. It’s just crux.

Jim: I mean that’s standard. There’s variation and then there’s selective reproduction. You have to have both, right?

Stuart: Selection is acting at the level of the Kantian whole.

Jim: Correct, at the phenotypical level.

Stuart: And indirectly on its parts. So we need two more concepts. One is catalytic closure. Catalytic closure is just that all the reactions that need to be catalyzed by somebody in the system are. So I’m going to get to a new idea that I’m partially the author of.

Jim: Oh yeah. Let me ask one question about the autocatalytic closure. Of course, in the forms of life, we’re talking about far from equilibrium systems that yet nonetheless maintain something very much like a homeostasis infecting say they do, they are a homeostasis. So would you argue that the autocatalytic set or catalytic closure is the means by which homeostasis is maintained in the far from equilibrium setting of a cell?

Stuart: Well, we got to go further. Let’s get catalytic closure. That’s clear, right? Going to get to constraint closure in a minute. But meanwhile, that system is a chemical reaction system. It might be wildly chaotic. Well, that’d be useless, right? It could fuck everything up. So we now know that genetic regulatory networks are dynamically critical, which I kind of accidentally discovered when I was 25. So if you make these dynamical systems, they can be ordered or critical or chaotic, and there’s a phase transition into the space of possible networks and so on. And living cells, a genetic regulatory genomes, are critical. And that was actually in the paper that I published in ’71. So if networks are dynamically critical, they behave with the order that you need to survive. So that’s a whole long story.

Jim: Yeah, basically it’s the toolkit necessary to locally reverse the second order, the second law of thermodynamics. Locally, because we know as a whole, of course you can’t. But the actual trick of life is that within the membrane of the homeostatic entity, we could do what appears like a defeat of the second law.

Stuart: It’s what Schrodinger are worried about. How can that be true?

Jim: Let’s say you have catalytic closure and then you have constraint closure.

Stuart: I want to now get to this idea of constraint closure. It’s a major idea. I got half of it and I couldn’t figure out the second half, that’s the crux part. So let’s take our time. I was writing my third book investigations and I was trying to figure out what the work cycle is. So let’s pretend there we’re physicists. So what’s work? It’s force acting through a distance. But there’s a guy named Atkins who wrote a book called The Second Law. So this is really neat. Atkins said, “No, no, no works much more interesting than that. It is the constrained release of energy into a few degrees of freedom.” And you say like what?

So here’s what it is. I’ve got a canon. The canon is the boundary condition of the system. There’s some powder at the base of the canon. The canon ball is next to the powder. When you explode the powder because it’s inside the canon and it doesn’t explode as a spherical wave, the only way the gas can go is down the ball of the canon and it does work on the canon ball and blasts it out of the ball. So the ball of the canon are the remaining possibilities, the remaining degrees of freedom. So Atkins is right, the canon is the boundary condition. And it’s also the constraint on the release of energy.

Jim: Which allows the work to actually do something useful rather than just be an increase in entropy, right? It is an increase in entropy, but it can also do specific work.

Stuart: Well, it blasts the ball out too, so I got it. No constraints, no loose of energy, no work. So I’m not a physicist. So I asked myself a funny question. So where’d the canon come from? Oh-oh, somebody did some work to make the cannon. But we got to have the cannon. Newton doesn’t tell us where the boundary conditions come from. It’s obvious that his work is done. It could make something that could be a constraint. In other words, if you do work, it can build a cannon. So I got as far as if you have something that does some work, that work could construct something that’s boundary condition. And I got totally stuck, Jim, from 2000 until 2015, and I met Maël Montévil who I knew and Matteo Mossio, and they solve the issue. I’m going to tell you a truly transformative idea. It’s theirs, except that I played this little role in it.

So they want to get some work. So you better have some non-equilibrium processes. Let there be three, process one, process two, and process three. So there better be some constraints on the release of energy. We need three A, B, and C. So A, B, and C are constraints or boundary conditions. So just hear it. A, constraints the release of energy in process one, and it constructs a B. B constraints the release of energy in process two, and it constructs a C. C constraints the release of energy in process three, and it constructs an A. This is constraint closure, pause and take it in. A, B, and C are boundary conditions. The three boundary conditions constrain the release of energy in three non [inaudible 00:44:25] process that construct the same boundary conditions. It’s magical Jim.

Jim: Somebody came on this a different way, but I think they got to the same place. You familiar with the work of Terrence Deacon?

Stuart: Yeah, he’s kind of there too.

Jim: Yeah. In fact, I am having a podcast tomorrow with one of his colleagues where we’re going to dig into Deaconism for a second time. I did it with him on his very difficult book and tomorrow’s got an easier book. But yeah, this reminds me, this is the same theory as Deacons’ essentially.

Stuart: We’re all [inaudible 00:44:57], got to it independently. It’s his second order constraint.

Jim: And the fact that essentially when interesting things happen in the universe, they happen not just from positive construction but from the creation of constraint or creation of negative space, which is also another one of our old colleagues, Harold Morowitz and his theory that of all the great emergencies, his 27 or 28 emergencies in the history of the universe, which is pretty arbitrary, but a useful list to read. Each one comes with a pruning rule of things that were not possible. And so all these things are closely related.

Stuart: Terry got there independent. I mean I only got halfway there. What was really funny about it, Jim, is I was looking at my images of autocatalytic sets and they’re exactly what the other two guys said. I was even drawing the same things that they drew and I was drawing them in the pictures of my autocatalytic sets I was too dumb to see it. Now let’s take go Ashkenazis nine peptide autocatalytic sets. It’s obviously got catalytic closure. Let me show you that it has constraint closure. So you’re feeding in half fragments, a peptide one and half fragments of peptide two and a peptide nine from the outside non-equilibrium. Peptide one binds the two half fragments of peptide two.

So those two half fragments aren’t diffusing all over the place. So that lowers the activation barrier to bind the two half fragments together. So the two half fragments are bound together. Work is done because a peptide bond is formed. So that is exactly the constrained release of energy and peptide one is in fact acting as a little ligase. It’s a boundary condition. Peptide two does it to peptide three and so on all the way around. Gonen Ashkenasy set achieves constraint closure.

Jim: Now assuming that peptide three or somewhere else down the stream creates peptide one, right?

Stuart: One makes two, makes three, makes one, makes nine, makes one.

Jim: One. Exactly. That’s the critical part. And again, the cycles of work as you called them long, long time ago, right?

Stuart: Yeah. Now notice, the Joana sets do the same thing. Joana’s small molecule autocatalytic sets, assuming that they work in a test tube, has the same property. Each reaction is catalyzed by somebody. So those are reactions that are the constrained release of energy. So Joana set already is simultaneously a Kantian whole. It achieves catalytic closure and constraint closure. It’d be nice to put it into a liposome or somewhere so it’s bounded. Jim. I think the union of catalytic closure, constraint closure, probably a spatial closure being a Kantian whole is an avital. There’s nothing mysterious, we’ve just unpacked it. It’s an unexpected organization of chemical processes that are far from [inaudible 00:47:53], it builds itself. It builds its own boundaries. So we build automobiles. Automobiles are full of parts that work on one another. The automobile doesn’t build itself. Cells construct themselves. They literally construct themselves. I think that is the heart of life right there.

So we’re now through to the following point, which is going to let me get to this third transition in science. But let’s look at where we’ve got. We have something like this tap process, it combinatory makes more and more kinds of complex molecules out of a bunch of atoms. So the ratio of reactions to molecules goes up. So at some point this [inaudible 00:48:38] phase transition should happen and you should get an autocatalytic set. And meanwhile we’ve shown it numerically. Wim, Roger Koppl and I have put tap and raft and the autocatalytic set and the diversity of things goes up and then wham, you get an autocatalytic set. So I really think this is at least one pathway that gets us to the emergence of Joana sets. And then the Joana sets already have amino acids and nucleotides. So I think you can get to an RNA peptide autocatalytic set.

Then it’s actually Nile Layman and I published a paper on the emergence of coating. We think it’s almost natural to get to coating. But I want to point out to something selection exit the level of the Kantian whole and indirectly selects the parts. So once you have Joana set as the metabolism, that is feeding a collectively autocatalytic set of peptides and RNA, the RNA set by itself as a Kantian whole, the peptide set is a Kantian whole. When the two unite, it’s a new Kantian whole within it, the metabolism was a Kantian whole. It’s now a more complicated nested Kantian whole. So selection always acts at the level of the whole, and that is downward causation. We haven’t gone to strong emergence yet.

Jim: Now of course there is, I don’t know if this is the right place to insert this, but it’s one of my favorite topics when we’re starting to think about the road up from what we would currently call protolife. But this is the issue of information conservation versus the error catastrophe in evolution. My academic work or my most intense work is on evolutionary programming. And there we’re always looking at the air catastrophe. You got to make sure you keep your mutation rates low enough that you don’t fall into the air catastrophe. And that to my mind is one of the big puzzles actually around the Fermi paradox. Yes, one can see how we get these autocatalytic sets and. The math is pretty compelling that with the right set of substrates, you’re going to get an autocatalytic set of some sort eventually, right? As you showed in origins of order, a probability is very good.

But how do you preserve information between the generations quote unquote to give the evolutionary ratchet the ability to work and not be defeated by the air catastrophe?

Stuart: So I’m hearing your question. Let me state it that I want to show you something that puzzled me very much in the paper that you just read. Then I want to get to the third transition. I realized something very odd. So let’s think of Gonen Ashkenasy’s autocatalytic set. There’s nine peptides, each peptide achieves catalytic closure and constraint closure. So peptide one specifically makes peptide two. Peptide two specifically catalyzes the formation of peptide three. And around the cycle till nine catalyzes the formation of peptide one. It’s specifically built exactly itself. Hold that. Now let’s think of Von Neumann and his universal reproducing machine. So Von Neumann in 55 or 56 has the idea of a universal constructor, you. Well, the universal construction is a real machine and it can build anything. Because it can build anything to build anything specific. It needs some instructions.

So the instructions are made out of some steel beams that are welded together and put inside of the universal constructor. And at this point something magic happens. The instructions have dual use. As the instructions, they are used to direct the building of something else, and in fact, they’re used to build a second copy of a universal constructor. Then the instructions are copied by the first universal constructor and the copied instructions are stuck into the second universal constructor. And everybody said later when DNA was discovered, Von Neumann foresaw template replicating DNA and instructions in Mountford Iegan’s Air catastrophe. That’s nothing at all like what Gonen Ashkenazy said he’s doing. It is specifically constructing itself,

Jim: But it’s also probably constructing some side branches as well. That’s where the noise can come in. If the side branches start to dominate the cycle, it’s like a river being captured by another river.

Stuart: Jim you’re right. But what I want to point out to be puzzled about is inside of Gonen Ashkenasy set, which surely reproduces, there’s no separable thing, which is the instructions of dual use. The distinction between hardware and software just isn’t there, just not.

Jim: Yeah, that makes sense. That actually is a fair distinction between the pre RNA DNA world where there is no information substrate, it’s essentially all dynamics. Essentially all the information is stored in the dynamics.

Stuart: Yeah. So where does the instruction stuff come from? Well, it came from the development of template replication and the development of a genetic code. So once you’ve got a genetic code and coding in a cell, the DNA sequence and Paul Davies points it out, that is a universal constructor in the sense that if you’ve got the coding apparatus and so on, you can construct any definable sequence of proteins. Just stick the DNA and you get it, right?

Jim: Yep. That’s the theory at least. Obviously we know it’s not quite that simple, but yeah, that’s the simplified model.

Stuart: Yeah. So the point of it is the technique replicating DNA or RNA plus the encoding translation apparatus is a universal constructor for a vast class of polypeptides. That’s true. And that’s coding and that is information. And that’s part of what Niles and I got to in our pipeline on the evolution of the code. Now let’s ask whether it cell, a real cell is a universal constructor? Well, it’s not. So here’s a cell with a bunch of genes. Cut out all of the genes and substitute for each gene a random sequence of DNA. Well, the cell will make all of these random peptides and it’s just going to be cell lethal.

Jim: Will not work at all, right?

Stuart: No. Cells are not universal constructors cells specifically construct themselves. In it, we now have encoded protein synthesis, which is marvelous, but there’s something fundamentally wrong or there’s some funny distinction between encoded information as in the genetic code and what Gonen set’s doing, and it’s reproducing just fine. And presumably so is it Joana set, let’s just let that hang there. Meanwhile, all of the stuff about the air catastrophe is there. So can I take the last 20 minutes now and tell you about this third transition?

Jim: Yes, let’s do it.

Stuart: Okay. So we define the function of your heart. The function of your heart is to pump blood, but your heart does make heart sounds and your heart does jiggle water in the pericardial sac. Darwin pointed out in Origin of the Order that an organ used for one function could come to be used for another function. Meaning precisely that a different subset of causal properties of the same physical thing could come to be of use.

Jim: It’s what we call exaptation, right?

Stuart: It’s exactly exaptation. I got fascinated with it’s jury-rigging, Jim. A trivial example of jury-rigging is my daughter dropped her purse in a puddle. I took a broom handle, I slid down a wire coat hanger and I fished out her purse with look on the wire coat hanger. Jury-rigging is using things not for the reasons they were designed. And evolution does it all the time. Evolution is all the time making use of the same thing for a novel function. So I’ll give two examples. Dinosaurs had spikes that were good for thermal regulation. They were co-opted become flight feathers. My favorite example is the evolution of swim bladders. Some fish have a bladder. The ratio of water to air tunes neutral buoyancy in the water column. And so it could detect buoyancy in the water level. They evolved from the lungs of lungfish, water got into some lungs. The same object could come to be used for something else.

All right, now here’s the next step I’m going to make. It’s going to be that you cannot deduce the different uses of the same thing. So see if you agree with me, if you do, it means that evolution cannot be deduced because it’s full of these exaptations. So let me give you the case of an engine block. What can you use an engine block for? Well, you can drill eight holes in it and make an engine. But my colleague Andrea Rolly says, “You can drill eight holes and you use it to store wine.” But I knew for years that it’s rigid and you can use it as a chassis to make a tractor, which our tractors were made. Now it’s also true, you can use an engine block as a great paperweight.

And Jim, it took me about a year to get to this. The engine block is rigid and its corners are sharp. You can use an engine block to crack open a coconut. It’s a triumph. Right? Now if you’re using an engine block as a paperweight, can you deduce, I can use this thing to crack open coconuts. It might be an orange peel.

Jim: Yep. Sensible uses exaptations are to some degree constrained by the co-evolutionary context that they’re in.

Stuart: Absolutely.

Jim: Right. So you can’t make a living using the engine block for a paperweight probably. No one’s going to pay you to use your engine block for a paperweight. But one could maybe imagine on somebody might make a living using their engine block to crack coconuts. And so that yes, there is an unlimited amount of arbitrary adaptations that we can’t even envision. Or it can use it as a pendulum and a rude Goldberg contraption. Yeah, you could. So you have to constrain the set of possible uses with those that in some sense can make a living in the current co-evolutionary context. But to your point about it being open-ended. Every time you do that, you’ve just extended the co-evolutionary context.

Stuart: You got it Jim. So evolution is absolutely full of these, their exaptations. And that’s what Darwin said. And then they got named exaptation by Gouldian Verba. You cannot deduce the new use from the old use.

Jim: And because each new use produces a new co-evolutionary context, it opens a new horizon.

Stuart: And I call it the new adjacent possible. What is actual now enables the novel uses of the same things. So increased creating an adjacent possible. So there’s going to be a bunch of consequences among other things for both evolution and economics. There’s no deductive theory of jury-rigging. So given what’s actual now in the economy or in the biosphere, there is an adjacent possible, but the number of uses anyone thing alone or with a bunch of other things is indefinite. Can’t be deduced. But that means evolution cannot be deduced. That means evolution cannot be entailed by a law that says this is what’s going to happen next. It means evolution is unendingly creative.

Jim: And the only way you can find out what it’ll do is to run the algorithm.

Stuart: Except it’s not an algorithm.

Jim: By analogy, you have to let it do its thing.

Stuart: Yes. And it’s critical that it’s not an algorithm. Algorithms are syntactic computations. The bias is not a syntactic computation, Jim, it’s a propagating construction, not an entailed deduction. Now I want to show you a next thing that we realized in this paper. The third transition in science. Then I want to get to economics. So think of a protein in a cell, what can it do? Well, it can conduct an electron. It combined ligand, it can catalyze a reaction. It can absorb a photon, it can carry a tension load or a compression load or molecular motor can walk on it. The same goddamn protein can be used for any of those things. So here’s our cell and it’s got thousands of kinds of proteins and we sort of say, well wait a minute, any protein can be used for indefinitely many functions by different subsets of causal processes. But those novel functions better all coordinate because selection acts at the level of the whole cell. So all of the new functions have to always be coordinated with one another so that the cell stays alive even as the functions of the parts change.

Jim: Well, not quite. I mean, those that don’t, don’t survive to reproduce. So those that do, yeah so.

Stuart: Because selection acts at the level of the whole. And meanwhile it’s got constraint closure and catalytic closure. Jim, this is both [foreign language 01:02:17] selection is downward it’s acting at the level of the whole, and the parts are evolving indirectly. This is downward causation. It’s strong emergence, and it cannot be deduced. We cannot deduce the new uses of this protein. Now what this does is it says, and this is the third transition in science, we cannot deduce what the bio search is going to become. It will do it faster than we can say it. The molecules will find novel uses that turn out to be useful, and that’s it. That means if the final theory in physics is supposed to unite quantum mechanics and general relativity and deduce the evolution of the universe, including biospheres, there’s no final theory. There is no [inaudible 01:03:04] equation to be on somebody’s T-shirt. There’s not.

The third transition in science is, Newton’s the first, quantum mechanics is the second is critical that in Newton you have the things, the objects, and you have the relevant variables, position and momentum. Then you have laws Newton’s three laws of motion. Then you define the boundary conditions, say of the billiard table. The boundary conditions define the phase base of all possible combinations, a position and momentum given that you specify the initial conditions and then you integrate Newton’s equations and you get an entailed trajectory in a fixed appreciated face base. That’s the Newtonian paradigm. But it’s the same thing in quantum mechanics. You have the Schrodinger equation, fixed boundary conditions, and you get the entailed evolution of a probability distribution and then whatever measurement is.

So All of science, essentially all of physical science is a Newtonian paradigm. What I just told you is that the evolution of the biosphere is beyond the Newtonian paradigm. It turns out we can’t use any mathematics based on set theory to deduce what the biosphere is going to become.

Jim: Now I read that in the paper and my eyebrows went up because as you and I both know, a lot of mathematical physics is grounded in set theory. That’s a strong claim. Expand on this a little bit.

Stuart: Well, our proof goes as follows. Can you list all the uses of an engine block alone or with other things? No. Now let’s take a screwdriver. What can you do with it? I can screw and it screw. You can scrape putty off the window. You could tie it to a stick and spear of fish. You can rent out the spear. You list all the uses of a screwdriver, alone with other things, you can’t. Here’s the first axiom of set theory that, and Andrea and I saw, it’s the axiom of extensity. Two sets are identical if and only if they contain the same members. But we cannot prove that the uses of an engine block are identical to the uses of the screwdriver because we can’t complete the list. That means we can’t use set theory.

Jim: Well, I suppose we could prove there’s a union, but we can’t prove what the union is because you can’t enumerate the set.

Stuart: Well, yeah, you could say there’s some union, you have no idea what it is. And there’s an intersection, but you don’t know what the intersection is. Because the number of uses of a thing is indefinite or unknown. Because new uses will come up in the future.

Jim: And even if you use my pruning rule of not just all uses but good uses, uses, that can make a living, that also will be open-ended as the system evolves.

Stuart: Exactly. Well, Jim, now let’s turn to economics. The growing global economy is the same thing as the biosphere. Bugs are making livings with one another, and we do, we invent new things. Once you’ve got a bow across bows in the adjacent possible, we cannot deduce the new goods and surfaces that will arise. There is an adjacent possible, Jim, we don’t know what’s in the adjacent possible. That means the following thing. Consider flipping a coin 100 times and you want to know does it come up head 60% percent of the time you don’t know. But you know all the possibilities, it’s two to the 100th, the sample space of the process, so you have a probability measure. But we do not know what is in the adjacent possible of the biosphere or the evolving economy. We don’t know the sample space. We don’t have a probability measure. None. We don’t have a probability measure.

If you’re an economist and you’re trying to figure out what’s going to happen, you cannot assess risk. But this also goes to something fundamental for the economists. The fundamental founding theory of microeconomics is the Arrow-Debreu theory of competitive general equilibrium that Ken Arrow explained to me one day with Phil Anderson, it was wonderful. So here’s the issue. Supply and demand curve while raw. So as cost goes up, as the price goes up, the baker makes more bread, but people don’t want to buy it. So the supply curve and the demand curve cross, that’s the point where the bakers make the amount of bread that people buy. So market clearing is equilibrium for the economist. Problem, suppose that you want butter on your bread. Well, the price of butter is going to affect your demand for bread. Does there exist a pair of prices so that the market for bread and butter clear?

What if it’s thousands of goods? Arrow-Debreu in 1954 solved the problem and they got the Nobel Prize and it’s the foundation of all microeconomics. Arrow-Debreu competitive general equilibrium. It’s brilliant. Here’s the move. They invent something called dated contingent goods. So a dated contingent, good for you, Jim, is a bushel of wheat delivered to your doorstep if and only if it rained in Nebraska the Friday before for an hour and a half. You’re supposed to imagine all possible dated contingent goods. So since they’re all possible, there’s a continuum of goods, that just a continuum of goods. Then you’re supposed to imagine that all of the agents have, well-founded expectations about what’s going to happen. And we gathered the beginning of time and the auctioneer auctions off contracts for all possible dated contingent goods and we buy the contracts. If that dated contingent, good comes then the theorem, oh, what they prove is a fixed point theorem in this continuum of all possible dated contingent goods.

If you comb your hair, there’s a colic at the back of your hair. So it’s a theorem that there’s a fixed point. They use a fixed point theorem to say there is a list of prices of vector of price where markets clear. It’s an exquisite theorem and it’s the foundation of all microeconomics. Then people have gone and said, “But what if there’s more than one equilibrium? And what if there’s not complete markets and blah, blah, blah.” Notice it among other things, this has the notion of well-founded expectations. But given this third transition in science, Jim, we don’t have well-founded expectations. So this is going to mean something that ties into Brian Arthur. Over the past 100,000 years, the number of goods have gone from a couple hundred to billion. We have no idea what’s in the adjacent possible. We cannot have well-founded expectations. We cannot deduce what’s going to happen.

That means microeconomics mathematically is exquisite as long as the economy doesn’t, as Brian Arthur says, build out appreciatable new goods where we can have no well-founded expectations. So the brilliance of microeconomics mathematically is superb. It cannot talk about the [inaudible 01:10:12] build out of the economy.

Jim: It’s only really and Arrow. I did talk to him about this once at SFI. Did confess that, and of course this is the fundamental flaw in some sense of microeconomics, that it only works statically at one point in time. It’s not a theory of change over time.

Stuart: Yeah. And Ken and I talked about this a long time ago, he’s just a lovely guy. And he said, “But there must be some kind of probability distribution.” And this is 18, 20 years ago. I said, “Ken, I don’t think so.” So what has now happened all these years later is that Andrea Rowley and I have proven there is no mathematics that we can use to deduce the evolution of the biosphere, Jim, and that’s the last part of the paper is the emergence of life and expected phase transition. Because what I just told you is already true for a Joana set. Those molecules can turn out to be useful in all kinds of ways. And it’s certainly true once you’ve got a DNA RNA peptide autocatalystic set where any molecule can be useful in all these wacky ways.

Jim: And also think about the intermediate case. If the Joana set is constructed of a bunch of smaller subsets that are composable, there’s a whole bunch of combinatorics there as well.

Stuart: Absolutely. And it is constructed of some combination of the irreducible sets. So we have here the overall theory. The overall thing in that paper is, the emergence of molecular reproduction seems more or less inevitable. How many places among 10 of the 23 planets? Who knows? But it’s expected as a phase transition and once it happens, the thing is the Kantian Whole with catalytic closure and constraint closure, hopefully bounded, and the way that it gets to come to exist, the ever new uses of the parts that sustain the whole cannot be said ahead of time. Its strong emergence. It cannot be deduced. It’s a radical creativity in the universe and it means so many things, Jim. We have in the Jewish Bible, there’s two readings of the first sentence. God created heaven and earth and gave Adam mastery. Or the continuous present God creates. Well, if God creates, he doesn’t give man dominion over what hasn’t been created before. Out of God created comes man has dominion, that’s also western science.

Science is to have dominion, command and control over nature. If Andrea and I are right, we don’t have any dominion over the ongoing biosphere. We have no idea what it’s going to go do. We become participants with the rest of nature. I think it’s just a profound transition. And what does it mean for science? I mean, I begin to know what it means. I want to ask how does the evolving biosphere keep creating novel possible ways of things getting to exist together? It’s been doing it for 4 billion years. At a further point that I’m just getting to with my friends, catalysis isn’t what’s cruxed, it’s speeding things up. So if I’m a bacteria and you’re a bacterium and I have a flat tummy, I afford you a highway, you can crawl over me to get food. We have no idea what the fungi and bacteria are doing one another.

And I’m beginning to think that it’s not an autocatalytic set, it’s a mutually consistent set of affordances or opportunities by which we make a living. So I’m trying to get together with my friends and write a purpose saying, life will find a way. Meaning that if you have a sufficient number of species and the species got a sufficient number of processes and enough molecules that could be used in enough or wacky ways, what’s the chance that there’s some mutually consistent way that they can make a living together? And I bet it’s the same phase transition and that life will find a way is fundamental to life. So we’re going to work on that paper.

Jim: Oh, very interesting. Look forward to it. And you can think about it as every diad has some form of interaction. They’re very much like your buttons and your threads. And then the question is, how do those webs of diadic interactions add up to something greater than the individual pieces?

Stuart: Some kind of self-consistent whole. The name of the game is getting to exist in the [inaudible 01:14:34] universe. Same thing true for the global economy. You’ve got all of these things and processes. How many ways are they able to be recombined so that they can manage to make a living together? I mean, they might screw up, but roughly suppose you gathered together 10,000 different technologies all over the place, could they do something useful with one another or not? Well, we’ve never asked. But it’s entirely new, Jim. It’s that anything can be used for indefinitely many functions if you have enough things, each of which can have indefinitely, many functions. That’s like having indefinitely many reactions that could be catalyzed. What does it take for there to be a mutually consistent way of doing it? So I’m involved with a wonderful guy named Jan Dyksterhouse in Holland.

We’re doing the 140 species experiment with Jan and some other people in Holland. We’re taking 70 genotype, DNA sequence bacteria and 70 DNA identified fungi. And we’re going to mix them together into 140 species and plate them out on 50 plots of sterilized sand and watch them for a year or two. We have no idea what they’re going to do. They’re going to create new opportunities to exist together. Jan says, “It’s a search engine that will create bubbles of new possible ways to exist together.” We’re not going to be able to deduce ahead of time what they’ll do. Afterwards we can look at the mutations that have accumulated and say, so what happened? And if we have 50 samples, do they accumulate the same mutations in the same order?

Jim: Probably not. I’m going to predict they will not because the co-evolutionary context will be different in each location.

Stuart: That’s exactly what we expect, Jim. Nobody’s ever done an experiment. It’s just wide open.

Jim: That’s interesting. That’s a lovely elegant experiment. I really look forward to hearing the results.

Stuart: I do too. I’d have thought somebody would’ve done it. Nobody has. And another thing I want to look at is, suppose you culture one bacterium together alone or two or three or 10 or 50 or make it fungi bacteria. For a given period of time, say a year. How does the total number of mutations that accumulate scale with the number of co-evolving species? Is it linear in the number of species? Or does it go up as the number of pairs of species quadratic or higher order? If it’s going more than linearly, Jim, they are creating niches together.

Jim: Exactly.

Stuart: Co-evolution. Yep. And I bet we look at the genetics of that and the statistics of that. It may be tap and it’s the same thing for the global economy. Goods and services are niches for new goods and services that are compliments and substitutes of the goods that are around. So the thing that’s coming out of this third transition in science is to ask how does the living world keep creating new possible ways that things get to exist together in all of these wacky ways? And we don’t know. We haven’t asked the question. It’s so enthralling.

Jim: And as you say, life will find a way. It has actually far more possibilities and will actually work. But all that to do is find a few that work each time.

Stuart: It has to find at least one way that a bunch of things get to exist together.

Jim: That can do their work and close their cycle and then save up enough resources to be able to reproduce.

Stuart: Yeah. So I’m really thrilled by this. We give up the Newtonian paradigm, the evolving bias here is beyond the Newtonian paradigm, and it can only be beyond the Newtonian. It’s non-deducible. We can’t say what it’s going to do. We can start to study how it does it, and it’s just enthralling.

Jim: It’s very exciting. Well, Stuart, I want to thank you for one of the most amazing conversations we’ve ever had on the Jim Rutt Show. This is going to rack my head for a few days as I think about it.

Stuart: Please tell your audience the paper is Stuart Kaufman, Andrea Rowley, a Third Transition in Science, The Journal of the Royal Society Interface, April 14th, 2023.

Jim: And as always, we will have the link to that and the other paper and the books we’ve discussed on the episode page at So they’ll be easy to get to.

Stuart: I hope some economists think about what this means.

Jim: I expect at least a few economists I know listen to the show. So maybe we will catalyze a reaction.

Stuart: I hope so Jim.