Why Evolution Isn't Just Survival of the Fittest: Computer Scientist's Alternative Theory

Richard Watson [00:00:00] Whilst biological networks can be adapted by natural induction instead of natural selection, there's always natural selection going on as well, and that makes it difficult to disentangle. What was really responsible here was it, if I, if I turn off the natural selection, does it still do it? And the answer is no, because nothing changes if you didn't have the natural selection there.

Vance Crowe [00:00:21] Hi, I'm Dan Crow, a small business owner living in Central Illinois, and you're listening to the Vance Crowe podcast. Welcome back to the podcast. I'm glad you're here. Today we have Dr. Richard Watson. This is gonna be a fun one. Oftentimes I describe this podcast as being up the graph that is talking about ideas that only a very small number of people know about or think about at all, and that it'll be years, maybe even decades, before these ideas spread out and are adopted by the masses. If you're a long time listener to the podcast, you know that we have characters that come on and talk about things that are a little bit edgy and a little bit different. But today we are going to hear an idea that is actually probably considered heretical, which is that the idea of natural selection may be incomplete in some really important ways. And our guest today, Dr. Watson, talks about his theory that he discovered by using artificial intelligence and writing algorithms about a different kind of intelligence, a different kind of pattern spotting, a different way of knowing that instead of just looking at the individual level of animals trying to eat and mate and find shelter to survive, they're instead behaving as a collective or even the larger biosphere. This is wild, and I gotta tell you that in the beginning, you're gonna have to do a little bit of mental energy to be able to keep up with him. But if you can hang on through the interview, you will get through to the other side having a completely different way of looking at this thing that is so important to our understanding of the world around us that is selection and evolution.

Vance Crowe [00:02:03] So I am so excited to get to this interview and I want to thank Michael Levin for hooking us up with this. This was a rare, rare treat. As we go to the interview, I am now in full scale effort to get our podcast studio built. We are doing Legacy Interviews. We're gonna move where the podcast is being produced. And if you are somebody that has been interested in doing these Legacy Interviews and you're, you're capable of doing it online, now is the time to buy an online interview. We're trying to fill up our time between now and when the studio gets built and we would love for you to sign up if you're interested in having me interview one of your loved ones, maybe a parent or a child, so that that way you can capture their family stories, their values, the things that matter to them, the things that you want your family to be able to remember and pass down, then go to store dot Articulate Ventures and use the promo code Vance to get a 20% off discount because you're a listener of the podcast. Alright, so we're gonna head to this interview with Dr. Watson. I hope you love it and will be back next week. Richard Watson, welcome to the podcast.

Richard Watson [00:03:11] Thank you very much for inviting me. I'm looking forward to this.

Vance Crowe [00:03:14] So a few weeks ago, maybe two weeks ago, right after it was published, a friend of mine sent me your video about this concept that you were pushing back on, which was universal Darwinism. And I'd never heard that before. So we'll get into your background and stuff, but when you start off thinking about universal Darwinism, how do you even explain this to people?

Richard Watson [00:03:34] I think that it's something that people are, are already quite familiar with, but they might not know it by that name. And universal domination, we could describe as the idea that survival of the fittest explains how things improve from the simplest microbes to ourselves, to institutions, to the biospheres. The only way that anything gets better is by survival of the fittest. So more, more, exactly. It was Richard Dawkins who coined the phrase universal Darwinism. And what he was talking about there is he was imagining how could life and evolution work on other planets, right? On this planet. It works by natural selection. The way that Darwin described it, if we find life on another planet, is it possible that evolution could have worked by some other way? And he says it couldn't, right? You, it can't be based on lucky accidents where many things come together simultaneously. It has to be, therefore, based on many small accidents that accumulate over time. It can't be the case that those changes are directed towards things which are better because that would require some hand, which guides it towards things which are better. So they must have been varying at random. So there must be some sort of filtering process which sorts out the things which were good from the things which were not so good. And when you put those things together, many small increments, random variation, a filtering process, it's like, well, it has to be natural selection right there. There's no other way that it could, the the only alternative is that we have some directed process of improvement, and that's not natural, right?

Richard Watson [00:05:10] So yeah,

Vance Crowe [00:05:11] Richard Dawkins well known for not thinking God had some hand in, in pushing the giraffe to have a long neck. He's saying the giraffe has a long neck 'cause all the ones that didn't died,

Richard Watson [00:05:21] Right? So he's proposing that is the way that life on earth works. The way that all biological adaptation on earth, whether we explain the adaptedness of life on earth, is through evolution of natural selection, the way that Darwin described it, but not just for life on earth, but if we found life anywhere in the universe, it would have to work that way because it's the only way, it's the only kind of process that could produce that adaptation, at least that you can think of. But it's, it's more important to us, I think, not because of what it might mean for life on other planets, but because of what it says about the scope and limitations of adaptation for other systems here on earth. Right? So if it's true that evolution by natural selection is the only natural spontaneous mechanism by which things can improve, then that places strip limits on the kinds of things that can show adaptations. For example, famously loose social groups can't be adapted because they're not proper revolutionary units. What does that mean? Right? So you have a sort of, you might have an idea that organisms are well adapted to their environment because natural selection made them so, but loose social groups, whether that's in humans or primates or wolf packs or such, like they can't be adapted as a group because groups don't have offspring groups in the same way that individual organisms have offspring organisms.

Richard Watson [00:06:54] And because individuals can move around from one group to another, that means that selection that's happening at the individual level always undermines what might be good for the group. So it might be, it might be intuitive that a group that has more cooperative or altruistic individuals would do better in the long term, at least than a group that has more selfish or competitive individuals. But that's a group selection argument. And in the sixties that was thrown out as like, well, maybe sometimes under some very special circumstances you could have groups that reproduce as groups, but most of the time it's gonna be undermined by individual level selection. And another example would be ecological communities, right? So an ecological community, which has different plants and different vegetation that different animals are eating and different insects are pollinating, and all of these things are interacting with one another and working, working together well, are they working together or are they all just working for themselves? Right? And universal Darwinism will say, well, they can't be adapted as a unit, they can't be adapted to function well as a whole, each individual part of that community can be adapted to fit in with the other things which are there, but only to serve its own self-interest, only to serve its own short term self-interest. Basically, it's a jungle out there, right? It's dog eat dog. It's all against all, it's a complicated tangled mess. Not a system that's adapted for holistic wellbeing or harmony. Right? And famously, to come back to another topic that Dawkins like to talk about, the biosphere as a whole can't be adapted because the biosphere as a whole isn't an evolutionary unit.

Richard Watson [00:08:33] In order for the biosphere as a whole to be an evolutionary unit, that would mean that the biosphere has offspring biospheres that the planet gives next generations of other planets with life on that inherit the characteristics of this planet or that we had a parent planet from once we came. And that some planets would be through their ability to survive and reproduce better than other planets would proliferate more, right? We'd need that sort of a process to explain how natural selection could make the biosphere adapted. But since that isn't the case, it would essentially a closed system with respect to the living components on it. The biosphere might be complicated, says kins, but it can't be adapted, right? And this is contrary to Jim Lovelock's notion of Gaia, which thinks of the biosphere, the complex interactions and feedbacks between species on the planet as maintaining conditions which are conducive to the long-term sustaining of life on the planet. You know, even though the sun has got hotter over millennia, the life on the planet has regulated that temperature to keep it within the comfortable zone for life on the planet. And that's, that sort of self-regulation is something that you'd expect from an organism. And Dawkin says, well, maybe it does, but it couldn't have been adapted to do that. It must be dumb luck, right? Because the only mechanism by which anything can be adapted is natural selection. And it doesn't apply here.

Vance Crowe [00:10:03] So what you're saying, or at least what I understand is that evolution is always looked at, or at least right now in its current parlance, is everything happens on the individual level. If the individual survives because they had some mutation, they, they got a little faster, they, they somehow got a claw that allowed them to dig a little deeper. Now that individual goes out and spreads its genes more frequently than than the other ones. And so there are more of them, and that ultimately impacts the herd or the larger group. And you're saying, wait a second, let's look at it from the other direction. What about these larger ecosystems can they pass themselves on? And, and the current thinking is no, they can't because ev when it all comes down to it, evolution's done at the individual level. So if you look at this kind of larger, whether it's a forest or even a, a bigger thing, the entire globe, it doesn't make any sense. There's no ground to build the idea that, that these organisms as a species might evolve together in some way. Is that right?

Richard Watson [00:11:07] Right. That's right, exactly. So the, the wood wide web is another concept if you've heard of that. And then in forest, the, the microrisal networks or all mycelial networks are all connected underneath the ground, and the trees can exchange nutrients and signal one another and all that sort of thing. But could it be the case that the forest is like an organism that it's, it's adapted at that scale to be, to sustain its own survival and reproduction at that level of organization? Well, the, the orthodox view would say, well, no, it can't, unless you have a way in which forests reproduce forests, right? So you have a population of forests, not just a population of trees, and some forests are better at reproducing offspring forests than other forests are, right? And that is not obviously the case. It's not clear that that can work that way. Certainly the individual trees within forests are reproducing and they may sometimes populate new ground that didn't have a forest on it before and the forest spreads. But that's just a selection of the level of individual trees which are doing that. And the fact that they communicate with one another is just, well, the, everything interacts with everything in its environment, but they're doing it for their own individual benefit, not for the, not for the good of the species, not for the good of the forest, not for the good of the group, not for the good of the biosphere. It's all the individual level, which is not to say that there isn't, there aren't some interesting ambiguities in where is the individual, right?

Richard Watson [00:12:38] Sometimes there are cases where a group of organisms that are clearly different species can be selected as a unit. For example, the gut bacteria inside a host organism might be relatively well coapted to one another and coapted to one another in such a way that they create propagules that exit the digestive tract of the organism in question that may be effective in inoculating the digestive tract of another animal. And maybe that whole community is being sort of passed on as a parcel as it were, in which case there, there, there could be some selection operating at that level, but still it would be opposed by selection at the individual level where all of the individual bacteria within the gut, even within one organism, if they find a way to survive and reproduce better than others than they do. And if that causes the community to become sick or the host organism to die, it's like, well, that's not my worry, because that's the future. And natural selection famously doesn't care about the future, right? It's only the immediate and personal consequences of the characteristics of an organism that are the things that drive the change in evolution processes.

Vance Crowe [00:14:02] And so when you are looking at the world, how do you view evolution because it's different than this kind of universal concept?

Richard Watson [00:14:10] Yeah, so I have been looking at other principles of adaptation. So in a, with a computer science hat on, I would call them other algorithms of adaptation. And there are other algorithms of adaptation that we already know and understand in different contexts, right? So the way that our brains learn when we learn something is an individual. It's not because, well, I'm gonna try out these a hundred different brains and I'll see which one works best and I'll make copies of the brain that works best and I'll throw away the brains that don't, right? It's not a natural selection process inside our heads, although there are some interesting side stories about that. And it, it is also not obviously the case that there's a natural selection process either at the neuronal level or the synaptic connection level either, right? And exactly what's going on inside the brains of complex vertebrates like ourselves. As you know, there's all sorts of complicated things going on there. But when we make an abstract computational model of learning in a neural network, affectionately known as an artificial neural network, we can do quite a lot of interesting learning with really, really simple processes. And a, a good place to start is with a principle or mechanism of associative learning, which is just about learning what things go together.

Richard Watson [00:15:42] And you can do that with a really simple mechanism of just strengthening connections between neurons that are stimulated at the same time. So when two neurons are both firing in response to a particular input, you strengthen the connection between those two neurons. And when one of them is firing and one of them isn't, so they're out of sync, you weaken the connection between those two neurons. And you don't need to have a population of neurons, do you need a network? But you don't need there to be a competition between neurons in a population or competition between synapses in a population or competition between whole networks in a population. You can just alter, alter the connection systematically in the right direction, right? When they fire together, you increase it when they're outta phase, you decrease it. That's called associative learning or correlation learning. And a simple name for the rule that we to formalize that is HEBs rule from Donald heb. He was 1940s to sixties who described that method of a, of associative learning. And it's often paraphrased as neurons that fired, again, a wire together. So we know that in artificial neural networks, I mean artificial neural networks are obviously a simplification of what goes on in real brains. But with artificial neural networks, that role is enough to do all sorts of interesting stuff, right? You can form an associative memory of past experiences and recall multiple memories from the same network. You can generalize to produce responses to stimuli you haven't seen before.

Richard Watson [00:17:18] And you can build low dimensional models of high dimensional data sets. You know, if it's like, do a high level representation of something that was complicated. And you can also use that kind of learning principle to learn how to solve problems, learn how to do problem solving better with experience. You have multiple attempts at a problem. And each, with each attempt, you look at, well, in this attempt, these two things went together and in this attempt, these two things went together. So I strengthened those connections. And in this, in all of the attempts I've seen, these two things never occur together. I weakened that connection and that kind of associative learning enables you to get better at solving that problem over time. So that's a different mechanism of adaptation. It's, it is not the same as natural selection, right? There's no proliferation of entities in a population. There's no need for random variation. But you think, well, okay, that's, that might be an adaptive mechanism and it might not be natural selection, but it was selected for the purpose of doing that adaptive mechanism, right? Our brains are a product of natural selection. So it might not be doing natural selection at the time that we are learning, but it was a, the brain, the machinery was a product of natural selection and it couldn't have occurred by accident. That machinery for doing that kind of adaptation, let's call it associative learning, wouldn't have arisen by accident. So the thing that's different in the way that I think about biological systems and biological evolution is that that kind of mechanism of associative learning can occur in all sorts of biological networks that are not neural networks.

Richard Watson [00:19:05] And it doesn't need to be the designed or selected for that purpose to do that. It'll do that spontaneously.

Vance Crowe [00:19:12] So what animals or what, what situations do you see this showing up in?

Richard Watson [00:19:18] Okay? So a way to explain how that can happen in other places, and a way to sort of get it away from the idea that it's, that it has to be implemented in neurons, right? Because you'll, from your conversations with Mike Levin, you'll, you'll understand that that, you know, the kinds of things that brains can do can be done in other kinds of tissues in bioelectric networks that aren't neurons, but they can also be done in, in other substrates that are not even biological in principle, right? So in computer science we have this notion of an algorithm being substrate independent that I can implement a, an adding machine on a digital electronic computer, but I can also implement an adding machine on a mechanical computer. Or the mechanical computer could be made out of levers and and cogs, or it could be made out of water pipes and valves and filters. It doesn't matter what you make it out of, it's implementing the same algorithm. And that algorithm that I described of associative learning can is also potentially substrate independent. You could build it out of other stuff. So just to, just to give an example of how general that could be, and, and I think it helps to understand how it could happen by accident. Imagine that you have a system of particles or masses that are connected by springs and the springs represent things like, well, when this particle is up, this particle should be up. Or if the spring is longer than when this particle is up, this particle should be down.

Richard Watson [00:20:50] They represent constraints between the positions of the particles.

Vance Crowe [00:20:54] Can I jump in here real quick? Yeah. One of the things that you're describing for people that aren't into computer science or, or really understand is when you are saying, look, the computer can be built out of any substrate. It could be wires, it could be plumbing, it's because for the most part, all of these computer systems are built on binary, which means one thing is on, it's either on or it's off. And then when you stack up on or off over millions of opportunities, you can make very, very complicated calculations just using that simple thing. Is that where we're

Richard Watson [00:21:25] At? Yeah. So it doesn't have to be binary, but you know what it, the thing that it, the value that it represents could be represented by lots of different things, right? Okay. So you could have, you could have a high and low voltage in a digital computer represents the binary digit zero and one or true and false. And it could be a mechanical lever that's up or a mechanical lever that's down represents true or false. It doesn't matter what it's built out of. What matters is the way in which it interacts with other things, right? So in my algorithm, when this one is up and this one is up, I want the output particularly low. And when this one is up and this one is down, I want the output to be high. You know, that sort of thing, that sort of logic of the, of the algorithm. And in principle, you could do that with a system that wasn't binary, where you had a voltage, could be anywhere in a continuous scale or a lever could be anywhere on a continuous angle. And again, it doesn't matter what you build it out, if it matters what the relationships is between the components, and then I can, I can put the inputs in one side, obviously it'd have to be in the right language, or I put inputs in, in voltages or inputs in, in weights on the levers or something, and I get a reading out the other end in terms of voltages or in terms of positions of levers. But it's done the same calculation no matter what they built it out of. That makes sense?

Vance Crowe [00:22:34] Yes, absolutely. Yeah.

Richard Watson [00:22:36] So these, back to these springs, right? So if the, if the connections between the, the springs between the particles represent the constraints of a problem that say when, and this one is up, this one should be up, or when this one is down, this one should be down. If I let the particles go, the springs will push the particles into a position where they satisfy as many of those constraints as they can. That's what the system would do naturally. But that's only gonna be a locally optimal resolution to those constraints. There'll be some particles which are being pushed by some springs that can't be satisfied that they have a tension in the spring that you can't get rid of. Now that's not a learning system yet, right? It's, it's, it's a network of interactions, a bit like the neural network. But you could set up a system of particles and springs to do the same kind of calculation that a neural network does. But the interesting bit is how do you get it to learn like a neural network learns. And all you need to do to get it to learn is just that the springs are not perfect, that an imperfect spring is not perfectly elastic. So a perfect spring, when you pull it, it'll, and let it go, it'll go back to the same position that it was in before. Or you push it and let it go, it goes back to the same position, therefore is perfectly elastic. But an imperfect spring, a spring that deforms slightly under stress when you pull it too far or you extend it for too long, it doesn't quite go back all the way.

Richard Watson [00:24:08] It changes the natural length of the spring. And that has the effect, even in a mechanical system like that of saying, well, the more these two particles have compressed that spring, the more the natural length of the spring changes in a way that it stops supposing anything else, anything different from that position. And the more they've been apart, the more the spring changes in the direction that makes them more comfortable being apart. It just gives up a bit, right? It just, it just accommodates to whatever the particles are already doing. And that's the same principle of HEBs rule in the neural network. The connection between the variables changes in the direction that makes them more likely to do what they were already doing, right? So the two neurons that fire together get wired together, the two particles which are pressed together, the spring changes to make it easier, just press them together or the two particles which are pulled apart, it makes it easier for them to pull apart. And the point, the reason for saying it in terms of springs like that is to make the point that you didn't have to design or select anything to do that. You didn't have to design a system of particles and springs that did the same thing as a neural network. I didn't have to select it to do the same thing as a neural network. You can't stop it doing the same thing as a neural network. It does it naturally. The reason that the springs change in that direction is because that's the direction that makes them give up a little bit, right? They don't have to fight against what's happening in the system, they just relax. So that means that I describe it as a, a system that can be described as a network of connections that give way slightly under stress.

Richard Watson [00:25:52] And any system which can be described that way is capable of doing the same kind of associative learning that a neural network can do. So we can show a system of masses connected by springs that does the same kind of learning that a neural network does, and it does the same kind of ability to generalize and to find solutions to difficult problems better with experience just like a neural network does. Now, because that's not based on natural selection, it's, I call it natural induction. Induction is the underlying principle of learning rather than of selecting that means that the, it applies to, we're not really interested in masses connected by springs. We're really interested in other kinds of biological networks, including networks between people in a society so long as that's described by a network of interactions which change the behavior of the components in the network. And those interactions give way slightly when they're stressed. That means that the network will do natural induction, it'll do the same kind of associative learning that a neural network does. So that, so

Vance Crowe [00:27:08] This would, this would be an example of like, let's, ima let's apply this to wolves, right? Our, our sense with Darwinism is the wolf that gets sharper teeth or is able to run faster, ends up surviving more, having more children, therefore more wolves of that type. That that's how you selected in this induction method. You're saying as a collection, the behavior of wolves when, when they're doing something over and over again, if it makes it easier for those wolves to go in to, to survive as a group, then it reinforces this behavior. But it's not necessarily about the individual propagating its genes, it's about the fact that the wolves cooperate. Like what, what is the mechanism that makes it so that spring that's been bent in the wolves or, or shifted has some impact on them,

Richard Watson [00:28:02] Right? So we don't need the wolves to be able to, or any, any mechanism that can tell whether the group was doing better. 'cause if we needed that, we would be doing selection at the group level. And this is a different mechanism. So imagine that there's something about the way that the wolves interact, which makes a difference to how happy each of the wolves are, right? That if they work well in a pack, then they get a better share of the of the prey which they hunt. Whereas if they don't work well together, then they're not able to get it and everybody's worse off. Now, from an individual wolves point of view, they're, if they were acting in short term self-interest, right? They're not interested in whether other wolves get another share of the meat. They're only whether be interested in whether they get a share of the meat. But if there's some change that they can do to their way in which they interact with other individuals in the pack, I haven't worked through this example before with falls, so we can be reflect on my feet a little bit.

Vance Crowe [00:29:02] We can go in the direction that you want. That's if it's easier, no problem

Richard Watson [00:29:07] If that's, if there's a way in which they, so it's obviously in the interest of an individual to modify their relationships with other individuals in the pack that makes their life easier. And what this mechanism shows is that by changing relationships in a way that makes their life easier, just like if this relationship is causing me frustration, just pay less attention to it. And if this relationship is working for me and it's easy for me, pay more attention to it. Right? And that's the heavy in principle, that's the associative learning principle at the PAC level. And the wolf doesn't need to know what the consequence of that will be for the pack. It's just changing the relationships in a way that makes its own life easier. But the consequence of the system level at the pack level is that the pack is doing associative learning. The pack is doing the same kind of distributed associative learning that a neural network can do, where each wolf is acting like a neuron in this case, and their relationships with each other are like the synaptic connections.

Vance Crowe [00:30:07] So what are the examples that you typically use when you're describing this?

Richard Watson [00:30:12] So one that I've done a lot of work on is interactions between genes and a gene regulation network. So in the genes within an organism, you know, decode into the proteins and that's important to determine the phenotype of the cell and the organism. But it's not just a bag or a beanbag as it's sometimes referred to of proteins, of genetic material. But it's a network of interactions between those things. So when this gene is upregulated, it downregulates some other gene which upregulates some other gene, which down regulates the first one. And there's a dynamical network of interactions between them. Those interactions are going to change over evolutionary time. And if they were to change in a way which followed HEBs rule, then the gene regulation network would be able to do the same kind of tricks that a neural network can do that it can learn multiple phenotypes, multiple cell types, and store them as memories in that network. One network could store and recall multiple phenotypes and they could do that with generalization. It could show an ability to anticipate novel cell types, which are well adapted to new environments, even though it's the process which has been changing the individual connections with natural selection. What's happening at the network level is natural induction that it's doing that kind of associative learning that the neural network does. The other kind of system that I've worked out some details of with my student, Daniel Power is an ecological network, a community network where we have a whole bunch of different species that have competitive relationships with one another that they're competing over resources.

Richard Watson [00:31:58] And there are constraints that you know, well if I'm eating this resource, you can't be eating that resource. We are competing over it. And if I'm paying more attention to this resource, I'm paying less attention to that resource. And if we just let the dynamics of the system run forward to what we call a climax community, which ones, which species do we end up having a lot of, and which species do we end up having? Little of those constraints will will shape what combination of species we arrive at. But Dawkins would say that ecosystem as a whole can't be adapted. It's only the individual species which are adapting themselves and their relationships with others to increase their own growth rate. They're not interested in anybody else's there. If I could increase my growth rate in a way which decreased your growth rate, I wouldn't care, right? We're not, we're not in this together, we're in this as individuals, right? But if it's the case that the changes over evolutionary time that happen in the relationships between those species have that same principle of giving way under stress, right, of being imperfectly, elastic, then that that means that the ecosystem can be adapted, not adapted by natural selection, but it's adapted by natural induction. Natural selection doesn't apply at the system level. It applies only at the individuals within each species, but each individual species and its interaction with other species are just trying to change those relationships in a way that make it better for itself.

Richard Watson [00:33:34] If I'm competing with you and you are in high density right now, there's a lot of you around then it's really important that I reduce my competitive relationship with you. If I could, for example, eat something else, eat something that you are not eating right, that would reduce our competitive relationship. But if you were not in high density and somebody else was in high density, then it would be more important that I change that competitive relationship. So, which relationships change over revolutionary time? It's the one where two species, which were in high density at the same time change their relationship more quickly than when one species is high and one species is low or above low. And that's the heavy in principle again, right? It's where the stress is in the system where the most competition is being experienced is between two species, which are both in high density and both competing over the same resource at the same time. And when that gives way because it's a, well, I'm gonna go do something else, right? I'm not gonna fight with you anymore. I'm gonna go and eat something else that changes that strength of the connection in the same way that HEBs rule would, which means that the ecological community can learn, have associative memories and get better at solving problems over time in the same way that a neural network does that same kind of learning. So Dan was more, how should we put it, had more confidence in the ability of this system to do cool things than I did at first. And I thought, well we, we'll we'll do some problems like the ones that we did on the gene networks first.

Richard Watson [00:35:06] And he said, ah, I'm gonna do Sudoku. What do you mean you're gonna do Sudoku? Right? I'm gonna set up the competitive interaction coefficient between species to represent the rules of Sudoku and see if the ecosystem can solve a Sudoku problem. Ah, that's never gonna work. That's too difficult, right? I'm gonna do it anyway. You do it anyway. So you have a species that represents a number one in the first cell and a species that represents number two in the first cell, and a species that represents number three in the first cell. And they're all competing with each other because they can't all be high at the same time. You have to pick a number and they have competitive, the species number one in this cell has a competitive interaction with species number one in all the other cells on the same row and on the same column and in the same box, but it's less competitive with the other number ones and that are in different rows and different columns, right? So you set up that to mean the rules and then you fix particular species to be high density. Those are the numbers that are filled in at the beginning, right? There's a seven over here and a four over there and you say those species can't change, they're always at high density and the other species have to figure out what the solution to the puzzle is. But the way that they do that is not by, right? If you did this in a selection way, you'd say, okay, give me a total community which solves this problem, right? You do selection at the community level, but in this model there's no selection at the community level. All that's happening is that species that find themselves in a competitive situation with other species change that relationship quickly. And species that are in a competitive relationship with other species that are not in high density at the same time, don't bother changing that relationship.

Richard Watson [00:36:43] 'cause it doesn't matter if we're not in high density at the same time. And that differential easing of frustrations in the network is enough for the network to learn how to solve Sudoku problems better with experience, right? So in, at first, the species just find a sort of locally optimal solution where there's still lots of violations, right? There's two sevens on this row and there's two threes and there's box and there's, there's violations all over the place, but that's the sort of local equilibrium. And then you pull back from there, you randomize the species st tendencies and let them have another go and they find another local equilibrium that's equally bad. And then you pull back and have another go and they have a another equilibrium that's equally bad, but over time, they're changing the strengths of those relationships with one another. And the changing nature of those relationships means that the dynamics of the system changes over time. So that now I, I don't pay so much attention to that guy anymore, I pay more attention to this guy. And the way that they change makes it better at finding higher quality solutions. And he showed that, you know, the kind of Sudoku that you get in the newspaper, the ecosystem can solve that quite easily. And you had to go and buy the book of the world's 10 most difficult Sudoku problems to find something harder. And it did eight outta 10 of those as well, right? So it, it's not perfect, but it's good at doing Sudoku by human standards. It can solve puzzles that you would find difficult.

Vance Crowe [00:38:07] So when you started putting this idea together, it, you're running into space that is pretty solid, right? There's, there's very few people out there saying, I don't believe natural selection explains everything in the world. How has the reaction been to you putting forward this idea?

Richard Watson [00:38:22] Well, you have to tread gently, right? So one of the things that's difficult is that as soon as you start saying, I've got a different mechanism of adaptation that's complimentary to natural selection, there's gonna be some opposition there. And the thing that makes it particularly difficult is that whilst biological networks can be adapted by natural induction instead of natural selection, there's always natural selection going on as well. And that makes it difficult to disentangle. What was really responsible here, was it, if I, if I turn off the natural selection, does it still do it? And the answer is no, because nothing changes if you didn't have the natural selection there. But the natural selection only explains what's happening at the individual level. It doesn't explain what's happening at the system level that's explained by, and it's like the difference between HBS rule versus associative learning, right? You know, HBS rule tells you how to change the individual connections, but it doesn't really explain how the network got better at solving the problem over time. So, you know, there are treading carefully, right? One is to say, one way to phrase it, instead of saying this is a different process is to say, look, there are some circumstances under which natural selection can behave like a learning system. And when that's within an evolutionary unit, like a gene regulation network is within an organism within an individual cell, that feels okay, right? Because it sort of gives you this flavor of, well, I already knew natural selection was clever and now I know it's cleverer than I thought.

Richard Watson [00:39:56] So that's, I'm okay with that, right? But when you start moving to the ecosystem model is a wait a minute, ecosystems aren't supposed to be able to solve problems, right? Ecosystems are supposed to be, it's a jungle out there. It's all, it's all against all. They don't work together to find solutions to complicated problems. And then that's a, that's a more controversial claim to talk about, but it, it enables us, you know, the, the way that Darwin provided an argument to explain biological adaptation wasn't by experimentally showing that natural selection could produce adaptation, right? What he did was he said, look logically if we have heritable variation and reproductive success, so we have organisms that reproduce the offspring are like the parents, they're not all exactly the same. And some of those differences matter to the survival and reproduction of the individuals. We have all of those things, then natural selection must occur. It must be the case that the things which are better able to survive and reproduce will leave more copies in the future, right? It's a logical argument about how things could work. And we see all this amazing biological adaptation and this is the first and only theory we've ever had about how that could occur. So it must be the theory that explains it. So the things that you need for natural induction to occur are similarly ubiquitous in biological systems.

Richard Watson [00:41:28] Biological systems are full of networks of components that are interacting with each other. It's nearly always the case that those interactions are not fixed or constant, but they change over time and they're gonna change over time in a way that gives way under stress, right? They don't push back harder against stress, they just give way. That's the natural way for things to change. And the other thing that you need in order for natural induction to work is that the system, I mentioned it before with the species that you sort of pull back from an equilibrium and then have another go that the species densities get occasionally perturbed or randomized. And the system has another approach to the climax community and it might be a different climax community the second time than the first time. If you don't do that, then it just rushes towards one equilibrium and then all the connections change in such a way that it reinforces that equilibrium. And then from then on, even if you do perturb it, that's the only state configuration you're ever gonna see. It's like training a neural network on one training pattern, right? You only, you'll only ever see that pattern if you only train it on one patent. So the system has to have a diverse distribution of experiences in order for it to learn and generalize over that distribution. So to the extent that connections, to the extent that systems are described by networks of connections and those connections are not perfectly elastic and the stresses the system experiences are not constant natural induction must follow.

Richard Watson [00:43:03] What's needs to be shown empirically is to say, okay, well now show me whether it was natural selection or natural induction which was responsible for the adaptations that we saw in a particular instance, right? That's more difficult. But in the case where natural selection didn't apply, because it doesn't apply to ecosystems, it doesn't apply to the biosphere that's already ruled out. So then the ular is the, what remains is either, well, they're just not adapted then, or we would need a different theory to explain how they were adapted. And there are lots of people looking at ecological systems and the biosphere and saying, wow, that's kind of cool the way that that works. Right? Look at all those feedbacks on how those feedbacks self-regulate one another and create conditions for the sustaining of one another. The ecological functions, cycling of materials, recycling of resources, different roles playing different, different functional roles within the ecosystem. Look how well organized that appears to be. And they just get a slap, right? It just isn't, it is not well organized because there isn't a mechanism for it to be adapted because it's not an evolutionary unit, right? They, oh yeah, yeah, yeah. Sorry about that. I shouldn't have, I shouldn't have said that. It looked well organized. I should have just said it was complicated, right? But if there is a mechanism that could explain that could apply in systems like that, when then we are entitled to look again and say, well is that organi? Is that organization better adapted than we would've expected by chance?

Richard Watson [00:44:34] Is that what you would expect from a everything against everything, all against all It's a jungle out there sort of dynamic, or is it possible that a mature ecosystem is better organized than a immature ecosystem that has had more experience with how to deal with the stresses that it experiences and it knows something about those interactions which enables it to find equilibrium which have a higher total biomass and where more of the species are happy more of the time.

Vance Crowe [00:45:05] How on earth did you find this as a question to pursue?

Richard Watson [00:45:16] I have always been inspired by the awesome adaptive complexity of biological systems. And I naturally enough was exposed to the theory of evolution by natural selection quite early on. And that's the orthodox explanation for the biologic complexity we observe. So I set my career path on studying evolutionary biology, but I did it from within a computer science department. In a computer science department, you can do things like take inspiration from that biological algorithm, the algorithm that Darwin described, and you can use it to solve engineering problems. To optimize engineering problems, right? And sure enough, you know, that field is called evolutionary algorithms or sometimes it's called genetic algorithms. Sure enough, evolutionary algorithms can show an ability to improve solutions to engineering problems over time to optimization problems over generations. And they show some pretty quick improvement at first, and then they tend to sort of slow down a bit and then they sort of pop out and then after a while they don't show any more improvement, sort of finished. The opportunity for innovation is exhausted. And then what should we think about that? Should we think, okay, great, my evolutionary algorithm worked because it started off down here and it ended up up there.

Richard Watson [00:46:48] And you know, that's adaptation. That's, you know, it's got better over time. That's what evolutionary natural selection is supposed to do. That's how we understand it to work. So it worked from an engineering point of view and it, it in a loose sense represents a model of how biological evolution works. And I guess the, the truth of it is, I didn't buy it, right? If you run that same algorithm longer, let's let's run it for a month instead of running it overnight like nothing, right? It doesn't show any further improvement. You think, okay, well fair enough because I just ran it over for a month. Whereas biological evolution had 4 billion years. So it's not surprising that there, those results of that look more impressive than the results that I did on my machine overnight. And also the incredible massive parallelism of natural selection is, is implemented in the natural world could also be part of the story to recount for how the outputs of that are so much more extraordinary than the outputs that we can do in an evolutionary algorithm. But I still don't buy it In computer science, we have this feeling of either your algorithm is intelligent, meaning that it knows which direction to go to find improvements, or you're guessing that's a little bit too sweeping because natural selection is sort of in between where you produce variation at random and then you select the things which are good, but it still has this sort of gradient following process, right?

Richard Watson [00:48:32] You, you follow the gradients to the, to the nearest peak, and then after that, when you're stuck at the peak, there's nowhere to go. There's, there's no way that you can get to a solution that's better than that except by guessing.

Vance Crowe [00:48:45] Yeah, it's a local optima. Right?

Richard Watson [00:48:47] Exactly. So an evolutionary algorithm can't get off local optima. And if that's a good model for how biological evolution works, then biological evolution can't get off local optima either. Do we really think that the extraordinary biological complexity that we observe in the natural world is just the nearest local optimum that out of all of that incredible functional interactions between all of those components, you just sort of run up the obvious hill and when you get to the top, you've got a squirrel it's like, or wolf or whatever organism you're thinking about. It's like, it doesn't look like a locally optimal solution. To me it looks like something that's much better than a locally optimal solution. And in computer science we appreciate that that kind of an algorithm, just a sort of random variation and selection algorithm. It's like sure, that's an algorithm, but it's only like the first page of the algorithms textbook, right? There's lots of other algorithms in artificial intelligence and machine learning textbooks right now. Most of those other algorithms wouldn't apply to the natural world, right? Like what, like a branch and brand tree search algorithm or something like that is like what that maybe that maybe that doesn't apply to biology, can't do that spontaneously. But the neural network algorithms that I was describing there, they're known as connectionist algorithms.

Richard Watson [00:50:19] They're really, really simple. All you need is that the connections reinforce the correlations that are already there and they, by giving way slightly under stress or with HEBs rule, and you can do some amazingly clever stuff with that really simple mechanism. And because it's so simple, it is plausible that that mechanism could operate spontaneously in other kinds of systems. Whether that's interactions between chemicals in the primorial soup or interactions between genes and the gene regulation network or interactions between cells in a tissue that's creating an organ like Mike, Mike Levi studies with the electrical networks or connections between neurons in our brain or connections between people in society or between species in an ecosystem or types of species in a biosphere. All of those are networks at very many different scales of biological organization. And they all have potential to modify connections in that same sort of way that a, that a simple neural network associative learning algorithm does. And with that kind of algorithm, the kinds of problems you can solve and the kinds of, the quality of solutions that you can find is much greater than just climb up the local slope until you find a local optimum. So that, that I think provides a better explanation for why biological evolution provides or discovers such amazing functional adaptation that it discovers.

Vance Crowe [00:52:00] So if you're putting forward an idea that kind of shakes the, the permanent position of, of Darwinism, right? This, this would be like adding a a, a very, a very different lens to the overall view. What does it take to get this sort of view brought into the general consciousness down the graph as we say on, on the podcast so that more people not only see and understand it, but agree and, and add it into the thinking about evolution.

Richard Watson [00:52:31] So I've done that very carefully by publishing purely within a framework that is non challenging so far, right? So by talking about the action of natural selection operating on connections between genes in a network, I can describe that as, there's nothing funny going on here. There's nothing up my sleeves, right? There's no, there's nothing funny going on. But it better explains the adaptations that we can produce with under those circumstances than we could under the, under the previous model. Because a gene regulation network enables evolution to learn how to produce a set of phenotypes, not just one and how to generalize to produce phenotypes which are adapted for environments that hasn't been selected for in the past and stuff like that. But it's sort still sort of playing within the rules. Where I am heading with this now is to be older at saying what's really going on here is a different process. And that because it's a different process, we can then explain adaptations in systems where natural selection doesn't apply. I mean, natural induction doesn't apply everywhere either, right? Natural selection doesn't apply to systems that aren't self reproducing. It doesn't apply to systems that don't have heredity or systems that don't have any variation or operator or wouldn't apply. And natural induction doesn't apply to systems that aren't networks or doesn't apply to systems that aren't perturbed so that they have a distribution of experiences or doesn't apply to networks where the springs are perfect, where they don't give way under stress, then natural induction wouldn't apply.

Richard Watson [00:54:11] But because they're different, that means that you know, that the, you know, they're not fully overlapping in their, in what they can explain. And the, the example of an ecosystem solving Sudoku puzzles takes us, you know, into, around the ecosystems aren't supposed to be able to do that, right? They're not supposed to be able to solve puzzles 'cause they don't do adaptation at the network level. But by explaining how that works in terms of a, a selection process that individual species, individuals within each species and showing that the consequence of that at the system level is that the system is doing something you wouldn't expect the system to do. Whether I give it a new name or not, it's expands are the, the remit of systems that we can talk about as being adapted.

Vance Crowe [00:55:03] You know, I'm struck that your idea requires a level of abstract thinking about this because, you know, we, we, and and I find myself taking everything you're saying and filtering it through what I know about evolution. So whether that's, you know, from reading Dawkins or I'm a big fan, Matt Ridley's been on the show several times and we've talked about evolution and I'm struck by the idea that you have to conceptualize a form of cooperation that is hidden underneath the surface. Do you think of this in terms of pictures or in terms of the words that you're using to describe this? How do you build this model and how do you look at it from different angles?

Richard Watson [00:55:47] The way that my mind works is more pictorial graph, like, geometric than verbal, propositional sort of ways of thinking. But I suppose that it's my, it's it's my computer science background that makes me think about things like what kind of an algorithm are they? And that's, you know, that notion of an algorithm being substrate independent, that it doesn't matter what you make it out of, it's the same algorithm that colors the way that I look at a lot of things, right? And physicists do the same, right? That they have, you know, they have a way of understanding a, a distribution in statistical mechanics and then they see that distribution in economics and in populations and in parcel deliveries, right? It's everywhere, right? You have that same way of you, that process is analogous over multiple different systems and you can recognize it in multiple different systems. It's that sort of way of thinking, analogous way of thinking.

Vance Crowe [00:56:50] Yeah. You've added a new dimension. I, I play a game with people oftentimes, like if I'm going out to give a workshop or even just my friends, I'll say, okay, let's pick a room that we are both familiar with but we're not in right now. And how would you draw that? And, and what I've come to the conclusion is there's really two types of people. There's the type of person that can see it from above like a blueprint. And then there's people like me that I actually have to imagine myself walking into the space and I have to imagine like a flashlight looking at this wall and then over to the front and then over to the side. But you're right, this is a a third dimension, which a substantial number of my friends are in computer science or programming, that if you have that number algorithm language, there's a third kind of conceptualization way that is hidden if you've never been exposed to it.

Richard Watson [00:57:42] Yeah. For things which, you know, have moving parts and interactions. It's about, you know, well this happens and then that happens and it causes this to happen and it causes that to happen. And so, you know, when something is static, you, you know, the different ways that you can view it from the inside or from the outside. But when something has moving parts and interactions, that's another way in which you can think about what it is and how it works. And, and it doesn't really, you know, it doesn't really matter. Well what color was the wallpaper or was the, was the door handle on the left or the right? So long as it gets you one room to another, it still had the same functional purpose. Right? And it still have the same kind of interaction between that.

Vance Crowe [00:58:16] Yeah, that's, I mean, that's a whole nother level on this experiment that I've, I've run this thing probably, I don't know, 150 times. I ask people this question all the time, and really at its root, it's the, it's the question, do you think in words or picture or, and it's one of those ones that unless somebody asks you about it, it doesn't even cross your mind to think about

Richard Watson [00:58:36] You. You've done the experiment of getting people to draw a bicycle as well, right?

Vance Crowe [00:58:40] I've heard of this one. Explain that one to the people that have not done it.

Richard Watson [00:58:43] So, you know, you, you know what, you know what a bicycle is. Yeah. I know what a bicycle is. You know what a bicycle looks like. Yeah, sure. You show me a bicycle. I recognize a bicycle. Okay. Draw me a picture of a bicycle. Right. And oh, well, I know it's got two wheels. You know, another wheel there, it's got a saddle here. Oh. And it's got pedals down here. And then, let's see, this bit must be, Hmm. I don't really know how these bits connect. Where does the chain go around? You look at the mechanical details. Some people, you know, they can recognize a bike, but the, the understanding of the functional interactions between the parts and how they're connected is not there. And other people that, that perhaps people that have owned a bike and done more to the point done repairs on bikes have that a different level of sort of functional understanding. It's not just, it's not just that you are remembering an image of a bike better than somebody else. It's that, you know, that the chain has to connect the pedals to the back wheel and you know, that the front forks have to be able to turn, so there can't be a bar across from the axle. Right. You know, that's, that's sort of how things are connected and how they work. And people think the short answer is people think they know how to draw a bike and some people can't, is the short. Yeah.

Vance Crowe [00:59:53] And I think that goes to my, my belief that most of what we think of as thinking is actually just an amalgamation of emotions. Right. It's, it's like, it's mostly how are we feeling until you do the act of having a linear listing of things. Right. Until you say, how does that chain moving around from the pedals actually impact that wheel? Because I think our, I think our brain is doing, you know, to your point, kind of this flow state, it always wants to, to go to the easiest one that it can. And feeling and, and thinking feels the same from the inside until you go to write it all the way down. And only when you write it down, or only when you go to say it out loud, do are you forced to grapple with the fact that you didn't actually know. And it's mostly gaps. It's like beginning of thought, end of thought, and a giant set of gaps in there.

Richard Watson [01:00:47] Yeah. It's, it's a, in artificial intelligence, it's conventionally understood that one of the things that makes human intelligence special and different from that of other animals is our ability to do that abstract logical reasoning, forward thinking over time about possibilities that are not in front of us right now. The kind of abstract reasoning that you'd need in order to play chess well, for example, that you can reason about, well, I make this move, then he can make this move, or he could make that one, or she could do this one and then, well, I'll do this one. And you are reasoning forward through time over a space of abstract possibilities. And, you know, other animals can do that to a certain extent, but humans seem to be one of the things that's special about our intelligence is we can do that better than others, but actually we're not very good at it. Right. You know, you, you ask somebody to reason forward through more than a couple of possibilities and they're lost already and they can't hold those in their head. And in fact, grand masters at chess don't even do that anyway. They just look at the board and it's immediately obvious to them what they should do. It's more intuitive. It's pattern recognition. It's more, oh, it's just obvious. It's in the way that they see the board is more like the person that knows how to draw a bicycle properly. Right. And they, they see the functional interactions. It can't be like that. This can't be a good move. It's just obvious that you would do this. Or the sometimes they think, well, there's a couple of possibilities, I might need to think about that. And they spend some time thinking about it, but apparently often they usually do the first thing they thought of anyway. And that was probably the right thing to do.

Richard Watson [01:02:19] And our, the way that our brains actually come up with solutions is much more instantaneous. And gestalt, you know, it's all like a holistic than it is a sort of linear chain of reasoning. Right. But we are special in our ability to do that, but we're still not that great at it. And, and it takes practice to, to exercise that.

Vance Crowe [01:02:40] Yeah. When you began this whole thing and you started talking about patterns, I mean, you know, I I fully believe that basically intelligence is the ability to spot patterns. And it does make sense to me that if pattern recognition is actually more like flowing, right? It's the thing where the, the least amount of effort has to be done in order for you to be able to, to put two things together. Then it would make sense to me that a collection of individuals that are operating themselves by survival of the fittest, I need to eat, I need to mate, I need to do, you know, find shelter. But that the collection of those individuals has a form of intelligence that is just a reaction to a pattern that, a reaction to a stimuli that is a form of a pattern, right? Like we, when we do it this way, it just flows easier as opposed to having some form of will necessarily that makes this, I don't know what you wanna call it, it's not really a challenge to, to evolution, but the, the, the added layer that you're trying to propose make a lot more sense to me. That, that, that, yeah,

Richard Watson [01:03:48] Pattern

Vance Crowe [01:03:48] Recognition is a form of flow.

Richard Watson [01:03:51] So the, the algorithms that I'm talking about here are just a small departure from the really obvious flow, right? So when something flows in a completely sort of natural or physical way, like a ball rolling downhill or a,

Vance Crowe [01:04:10] An ox boat created as a, as water runs downhill. Yeah.

Richard Watson [01:04:13] Yeah. When it's, it is just flowing in that just reacting to the immediate local forces that are acting on it. It's like it couldn't do anything else, then that doesn't require any special explanation, right? You, you pour a bucket of scrap metal into another bucket and it takes a particular arrangement and you say, oh, why did it do that? It's like, it doesn't need any special arrangement. It doesn't need any special explanation, it just fell there. Right? Whereas in, we have this idea that in order to do something that's clever, in order to do something intelligent, you have to do this deliberative reasoning about future possible states and considering this and considering that. And that seems like something that physical systems don't do naturally. Right? A bucket of scrap metal can't do that. Right. But in between those two ways of thinking about intelligence, that, that, you know, there's, there's stuff that's unintelligent and doesn't require any explanation, and there's stuff that is intelligent and it's requires a special arrangement of components to make it do that. Right? How on earth do you build a machine that does the same thing that we do when we play chess, right? In between, there are other states of, there are other possibilities systems, which are almost just falling to the bottom of the bucket. But as they fall, they are adjusting their trajectories a little bit. So the species, let's do the example of the species solving Sudoku, right?

Richard Watson [01:05:45] The species solving Sudoku at first, they're just falling, right? The, the way in which they fall is determined only by the rules of the Sudoku puzzle, which is a bit like the shape of the bucket, right? That they just fall into it and that's it. But as they change their relationships with one another, or the more they fall in particular configurations, the more those relationships between them change, that changes the way that they fall. And now you can chuck them any old harrow, and instead of just falling to the model, the bucket in a haphazard way, they relax into the configuration where they fit together rather nicely. And that's, it's a, it's sort of, oh, it's like falling with style. I buzz light. You would say that, that it's, or is it woody that said that? That it's, you know, mostly just doing what's obvious, but the trajectories are modified slightly by past experience. And neural networks show us an algorithm that enables a system to get better with experience just by modifying connections slightly. But when it's finished doing that, when it's a mature ecosystem that's had lots of experience, you then, you know, you tip your species into the puzzle and they just find their way to a good answer. They fit together really nicely because they've accommodated one another over that past experience.

Vance Crowe [01:07:14] Yeah. I feel, so I've gotten a chance to speak with different turf management groups, so the people doing grass, studying grass, and it's amazing to me the way they talk about grass is that it learns and that it shares information and that it spreads that information around. And, and for the first time I've ever seen this in, you know, the, to your point, the kind of evolution on a, on another level, because it doesn't do any single blade of grass any good to share information with other blades of grass, right? It's gonna live and die and propagate and it doesn't matter. And yet it still finds a way to signal information through should we keep growing? Should we, you know, lose water? Should we try and hold onto more water? Should our root systems work in some different way? And that information sharing doesn't, I, I've, I've only read limited on, on evolution, but it doesn't seem to be addressed in just having survival of the fittest.

Richard Watson [01:08:12] Hmm. So that's, that's something that I think is important for us to understand as, you know, as, as a society and as you know, part of a planet in which is in crisis in all sorts of different ways. What's that? What's the right way to think about this? Right? So if you, if you think that survival of the fitness is the only way in which anything improves, then it's not, you could take an attitude of, well, you might not like it, but that's the way it is. That's the way of the world. You may as well just get on with it. And by the way, you, whatever you can, whatever you can find for yourself, hold onto it because other people are gonna try and take it off you. And you, you know, the world is competition based and self-interest based and short-termism and other things, and other people, other species in the world are just resources to serve your short-term self-interests. They're resources to be exploited, to compete over with other individuals. And, you know, you might not like it, mate, but that's the way of the world, right? That's one way to think about it. But even worse than that is the, that that is actually how it should be. That because natural selection or the survival of the fittest is the only way in which anything can improve. If you do anything which hinders competition, you're making things worse, right? You have to foster competition in order to make things better, right? You need to deregulate the market so that firms and companies can make the most profit they can, because that's how things improve, right?

Richard Watson [01:09:52] That sort of, that sort of way of thinking about the world trans translates or transfers that survival of the fittest mentality from just being a theory about how biology works to being, you know, rules for life, right? This is, this is how we should treat one another and how we should treat other species and how we should treat the, the biosphere. And that's not just the way it is, but it's the way it should be. You know, that's, you know, don't mess with that. That's competition is how it's supposed to work, right? And if you see any, you know, working together and concluding, you should break that up, right? Because it's messing with competition, right? But when you think about systems adapting by this modification of their relationships between one another, then you see it very differently. And you see that world view of, you know, competition is good and self-interest is good, greed is good, as Gordon Gecko would say, right? But we don't need a, a fictitious character to say it for us. We can, you know, any, any advocate of profit maximization or cost sustained cost reduction or performance indicator maximization, you know, sustain sort of pressure for a system to improve that way is the same kind of mindset as survivor of the fittest. And it makes things worse, right? It leads us into a situation where we've exploited the planet to a, an extent where we're gonna cause another mass extinction, where we're gonna drive climate change, the very resources that we need to depend on being depleted and destroyed.

Richard Watson [01:11:32] And our compassion for one another as individuals within our local communities and more globally is just, you know, at, at the very least an uphill battle to remain any, to retain any compassion for one another as individuals, right? But if you think about a different source of adaptation that it is, the way that adaptation works is not through this thing survives and this one doesn't. But the way that adaptation works is through changing the network of relationships between components. And that by easing the most frustrated relationships, you're not just making life easier for yourself. You're making a network which is smarter. You're making a network. You are, you are, and you are on, you are, you're an individual connection node in a network which is contributing to the wisdom of that network, the collective intelligence. And I prefer systemic intelligence that the system of interactions that we have with one another and with other living things, to the extent that it's been exposed to a history that was representative of the situation that we're in right now, has distributed knowledge that gives its intelligence about how to respond. And it's not an intelligence of a single individual or even just the consensus of many individuals. It's a distributed network that computes stuff. The network that we create with each other, and the network that we create between one species and another has knowledge in it.

Richard Watson [01:13:05] And then when you think about, but what should we do with these natural resources? Should we eat them? Should we consume them? Should we exploit them? Should we dig them up and burn them? It's like, what are you doing right? You are, you are destroying the neural material of the planet when you do that. You are not just using up resources that we're gonna need in the future, but you are eating the wisdom of the biosphere. Those relationships that we have with those species are where the systemic knowledge of living things resides. And I think that that offers us a, a different sort of worldview, a different way of thinking about ourselves and our role in our relationships with one another and the biosphere as being part of something bigger than ourselves. That the systemic intelligence resides at a level of organization that's bigger than the individual. And that's important for what we view as our role and our place in the world. That's very different from the survival of the fittest view, which says everything else is just there as a resource that you can use and exploit to make yourself win and somebody else lose. And that's all there is to it.

Vance Crowe [01:14:16] Well, to me, your message of, there's another layer here is certainly hopeful because I remember the, the darkness gathering as you really do learn about survival of the fittest, and there's something powerful and important in order to understand those things. Like understanding natural selection helps you understand the world. But to your point, I think it's probably not been enough. And I think most people intuitively know that. So finding a way out of this Yeah. I think is a really important thing. Yeah.

Richard Watson [01:14:46] If people

Vance Crowe [01:14:47] Wanted to learn more about your, your work, where, where would they go? How would they get more?

Richard Watson [01:14:54] You can go to my website, which I guess you can put a link on the, can't you, and that will point you to my publications, which are grouped into various different categories. You want to know more about the algorithms or you want to know more about gene regulation networks or the ecosystem model. You can find links to those papers there.

Vance Crowe [01:15:14] Well, Richard Watson, this has been a fantastic conversation. Thank you so much for, for accommodating me and sitting down and explaining your, your wild ideas. And I, I hope they go somewhere important.

Richard Watson [01:15:28] I've really enjoyed it. Thank you so much, fans for the invitation. It's been really great.