The 10 features of complex systems: Part 2

So, in our last episode, you heard all about the four conditions that we see in complex systems, and you heard about the concept of emergence. In this episode, part two were joined again by Karolina Wiesner, professor of complexity science in the Department of Physics and Astronomy at Potsdam University in Germany. Karolina is going to guide us through the six products that we see in complex systems. These are spontaneous order and self-organization, non-miniarity, robustness, nested structure and modularity, history and memory, and adaptive behavior. And by the time you finish this episode, you'll have the set of underlying principles of complex systems, the set of underlying principles that hold together the wide variety of topics we talk about on this series. This is simplifying complexity. A podcast where we explore the underlying principles of complex systems, systems that seem to defy our rational view of the world, like economies, ecologies, or even you or me. I'm forensic engineer Sean Brady and I'll be your host. Karolina, welcome back on the show. Thank you for having me again. So part two of our sort of deep diving to the conditions and products of complex systems. So in our last episode, we talked about the four conditions, numerosity, disorder and diversity, feedback and non-equilibrium. We also talked about emergence as well. This one is all about the products that come out of complex systems. And they are spontaneous order and self-organization, non-miniarity, robustness, nested structure and modularity, history and memory and adaptive behavior. So let's just jump straight into spontaneous order and self-organization. That is probably what captures our interest most of all these features that you've mentioned, Sean. It's what makes them interesting because it's something that's very visual, right? Self-organization. We've spoken about the lack of a central control. So that's in the word self. The system does it itself without something externally or some central element doing it for it. That's the self. And the organization is that something comes out of nothing. The nothing in this case is what we've spoken about before. So it's the disorders, the feedbacks are out of these things, something arises and that is order, structure. We had several examples already spoken about in the show and I've mentioned chemical reactions because it's so visual, you have initially some liquid which doesn't have any structure to it and then you let it go for a while. And suddenly you see circular forms, circles, colored circles coming out. That is a form of self-organization. Then there are, of course, all the living systems we've spoken about that would be the B-Hive has a form of self-organization, initially a colony starts out small. And then you have the division of labor which at some point is organizing itself. You have the next site that's being built is the form of self-organization. And the key bit there is that the individual bees are ants in and of themselves don't really do much. It's only when you put them all together and make them interact that we get this. Yes, it's a process, right? So the system starts out in some state which is if it really starts out, it's a very disorganized state. In the case of the chemical reactions, initially you have no structure whatsoever, we've just poured the chemicals in. It's at the very beginning. For a B-Hive, at the very beginning you would have a queen and a small set of workers. There is no nest, there is no division in which bees care for the queen and which bees carry in the food and the water and so on. You could even say in an immune system the response to a virus initially, the system is in some state which is already ordered of course, but once the pathogen comes in then the system self-organizes the response. And that is not centrally controlled. It is happening in a very distributed, like we've heard explained, in a very distributed way this response is a form of self-organized behaviour. Okay, so that's order and self-organization and non-linearity. Non-linearity happens in so many ways and some are more simple, more trivial than others. One form of non-linearity is this exponential growth that we've heard about. For example, in growth of a virus when it enters a host, then it replicates. And this replication is not that after a day there will be double demand and after a second day then there will be just a little bit more than double. It's not how it happens. It will be doubling every day, let's say, or every few hours depending on the virus. Which means that the number of viruses that you find is not just the same as the number of days you've waited, but it is way, way more than that. And this exponential growth happens in, well, scaling laws are another example of that form of non-linear growth, so especially in terms of time these systems have a non-linear change over time. Tipping points? They are form of non-linearity. Tipping points are absolutely form of non-linearity. So for one tipping points, the different aspects of tipping points that show this non-linearity, the most obvious one is of course that a tipping point is something where a system is somewhat unstable state and it keeps being nudged from the outside, which we've heard about before. And a small nudge, if it is away from a tipping point, a small nudge will just cause a small reaction of the system, slight adaptation, slight adjustment, whatever. But when it is close to a tipping point a small nudge will cause a huge change in the system. So that is a form of non-linearity in the sense that the response is not a linear function of the input, and that's particularly strong in tipping points that's different. So robustness, and we've talked a lot about robustness back in episode three, but probably worthwhile just summarizing very quickly before we move on to the next one. And I'll listen if you're interested in more robustness, then absolutely go back to episode three and have a listen to Carolina talk about it there. Yeah, robustness is somehow there in all of these emergent features, but in and of itself, it is emergent again, because of the conditions, the disorder, the feedback, they are actually responsible for something to be robust. And the system as a whole is robust in the sense that, you know, small perturbations from the outside and nudge, slight damage to the system, it does not stop the system from functioning or the system from self-organizing. It is in that sense an emergent feature, so it's a product of the conditions. And part of it from, and we obviously talked about, it's part of it from the sort of the interplay between positive feedback, trying to put loads of order into the system, and you have your noise or disorder, trying to take away that nice order and as the toggle are between the two of them, but the other point you make that I really enjoy is that you can grab a handful of answer to the ant colony and you don't kill the colony. And that presumably comes back to this concept that it's not essentially controlled. It's all dispersed throughout the system, so taking away a little chunk doesn't disable it. The same as we're carrying off the piece of the internet doesn't kill the internet. We've spoken about this order as a nuisance, well, some people would call it a nuisance I wouldn't. And one reason I don't think it's a nuisance, but actually essential, is it makes the system robust in just a sense that you've mentioned. Parts of the system fail, an ant colony's beautiful example, you scoop up a bunch of them, the colony will be fully functional, even in the chemical liquid, you dip in a spoon and you stir it a little bit, that might destroy the structure for a moment, but then it will reform. So it's robust in that sense that the structure is robust. And the same is for the immune system, you will perturb it a little bit in, say, slowing down the TSA response or something, but the system will recover. So it's a form of recovery that is given to the system because of this disorder and feedback. So moving on, then to the next one, which is nested structure and modularity. And I must say this is something I know very little about, so you're gonna have to go on baby steps here. What are we talking about? Many people would think these are the same, but they're not quite the same. So let me start with nested structure. You won't find it in every system, but in most of the big systems that we've spoken about, which is the economy and certainly in other social structures, you definitely find it. I wanted even in the galaxy, which is my favorite example of a non-linear. living system, where, depending on what the resolution of your telescope is, you'll see different structures. At the beginning, you would maybe just look at on the scale of a solar system, and then you see lots of solar systems. When you zoom in, you see that within the solar system, you actually have, say, planets, and then you would zoom in, and then you'd see within the planets, you have more structures, you have, say, moons that are moving around planets. So, nested structure means that you zoom in, and you see no structure, and you zoom in again, and you see no structure. And that is even true on someone more abstract level, say, a social group, democracies, one of my favorite examples here, where you have certain decisions that are being made on a very high level, say, the parliament. But then you zoom in and you ask, well, how do these parliamentarians get to their power? Well, that because they're being elected by a certain subgroup of people, and then you look at that subgroup, and you see your local, you know, city councils, and you look into the city councils, so on and so forth. So, nested structure is that you see at some level a structure, you zoom in, you see another one. Modularity is slightly different. It's more to do with function. So, modularity means that the system is separated into parts, and these parts take on different functions. The main example is the brain, different parts of the brain, they all consist of neurons, and little other bits and bobs, but essentially neurons, and there's parts of the brain that are responsible for the vision, parts of the brain that are responsible for motorics, and motory functions, and so on. So, modularity is to do with function, and it separates the system into subsystems that are somewhat on an equal level. You see that structurally networks, networks have been spoken about here, where parts of the network are, the brain is a network, right, and part of the network are responsible for different things. So, that's a form of modularity as well. And would you say that nested structure has hierarchy, whereas modularity doesn't? Is that the role on the same claim? Yes, that's a good way of putting it actually. Nested structure is a hierarchical organization where, of course, hierarchy we associate something, you know, social with, that the higher level of hierarchy has a, say, higher level of power, but my example from the galaxy or galaxy and solar systems shows that it's, it doesn't have that connotation, but it has that, you can call it a hierarchy in that sense, yes. Yeah, different things happening at different levels. I mean, it's an example that human body is built off of lots of nested structures. Yeah, I was going to say, bodies like metamorph of all these nested structures, but then when you put a parliament of parliamentarians together, that's more modularity. Because it's both. I think the body is both. The body has nested structure because, of course, you can look at it in terms of, you know, limbs as a head, they're in the arms and they have different functions, but there's also nested structure. I can zoom in and I can, you know, I can zoom all the way up to the immune system or I can zoom out a little bit and just think about where is blood and where is flesh and so on. So, it's a mix, which is why it's probably difficult. Modularity is really function. Modularity is about function. Nested structure is about zooming in and zooming out. And is that where fractals fit in and nested structure or are we talking about something different? No, that fits in quite well. And that's why the question of fractals and, I mean, it gets you back to scaling laws, of course. It's all linked up in a way. Fractals came about time in the history of the science. They are quite parallel fractals and complex systems, which maybe is not a coincidence. I don't know. Fractals are a form of nested structure. You zoom in, you see new structure and in the case of fractals, it's the exact same structure you see again and again and again. So, to try and take an example and tie a few things together, modularity is that a bit like saying that you've got a circular tree system in your body and that's what it does. So that's that module. But then when you get into that, you've got those nested structures. But as you zoom in from your Aorta and everything spreading out and you're getting the branching like Jeffrey talked about, is that a bit like nested structure? So the nested structure in the body, yes, you can look at the nesting of blood vessels, which is part of what Jeffrey West is talking about. And you see the large blood vessels and they are branching off into smaller ones that are branching off into smaller ones and so on. So you could think of that as a nested structure because it really is a physical structure that has different levels, different sizes, if you like. In the body, the modularity comes in terms of different functions. So the blood vessel system as a whole has one function whereas the nervous system has another function and the nervous system is also has also a nested structure in its physical form and then the body as a whole has its modular distribution of tasks, if you like. And would we say that? I mean, the example of the blood stream, oh, we zoom in like in the nested structure, they're all self-similar. But you can also have nested structures that are not self-similar and accurate. You move to a new level in the hierarchy and it looks quite different what's going on there. Correct. We could go back to the planetary structures in space that they're not self-similar in the strict sense. But we could also think of networks, which happens spoken about before. So networks often we find clusters. So that's groups of nodes in a network that are sort of more closely connected to one another than other groups. And then you can have clusters of clusters. And they are generally not self-similar. They can be quite different and that reveals something about the inner workings and the dynamics that are happening on that network. So history and memory, what's history and memory? We like to distinguish between history and memory because history is, you can think of the history of a complex system. How did it come about? That is, you know, how did, if you like, any, the history of any living complex system goes back to origin of life, which goes back to the origin of the universe and so on. So the history is really in a way as far back as we know at the moment. And they all do have history if nothing else because they come out from these conditions that we've spoken about, disorder feedback and so on. The memory is within the complex system. Of course, the brain has memory. That's what it's for. Remember, things we store them in neural structures in the brain. The immune system also has memory. It remembers pathogens it's seen before, even if it was years and years ago. And it stores that memory, you know, in a biological, physical form. And it reactivates it whenever it's needed. And colonies can have memory. So the thing about memory, what makes it so interesting is that the memory of a system can be older than the elements of the system, which means in an immune system, if it has seen a pathogen 10 years ago, there might not be a single T cell that's the same anymore compared to 10 years ago, but the system has remembered that memory and that's fascinating. And the same, in fact, can be true about an colonies. They can remember sources of good food from, you know, this is being sort of passed on from generation to generation. So even though the food source was discovered many years ago, and none of the answers that old, but the an colony as a whole has kept that memory. So memories stored within the system. And it is something that the system uses to its benefits. What's the relationship between or the distinction or is the one between memory and information? I say there isn't a distinction because I've just spoken about memory as something that the system keeps. And it is, in many cases, older than the elements of the system, which means the system has to store it in some form. And in some, I call it an abstract form, but it has to encode it, otherwise it can't store it. This means you are talking about information, because if you talk about encoding, you talk about information. And I mean, this is in our genes, presumably, we come even before we're born with all this information encoded inside us that help us be a better system, or help us be a system at all. That's right. DNA is probably the prime example of such abstracted information. It's encoded in the sequence of proteins. And it's information, it's being stored, it's being read out again, and it's acted upon. The DNA is an example. The immune system is an example. Of course, the brain is an example. I mean, social organizations have memory in terms of culture, in terms of norms, things we adhere to, we pass on to the next generations. And we pass them on in, you know, we encode them in a way, pass them on through behavior, or we pass them on through words or writing, those are all encoded forms of information to pass on which is a form of memory. And you've got a lot of line in your book where you say any robust order that exists in a system can be thought of as memory. I really like that line. We like to state things as general as possible. So memory is something, you know, the way I've spoken about it until now is something just more intuitive. I guess I remember a face or remember a song or whatever. But memory is in a way a persistence of structure. And a song, if I remember a song, it's a persistence of structure in my brain. But persistence of structure I can also find in non-living systems. So if you have, for example, drastic coast and south of England comes to mind for some reason, I don't know, it's a beautiful structure. And it has come about through this interaction between Christy, the ocean and the coastline. And it's been there for a long time. It's, you know, slowly changing over time. If you like, it's a way of, it is the memory of these past interactions between the ocean and the coast. And there's no living system involved, not in the way I'm speaking about it. Now it is really just interaction between physical elements. And it is a form of memory because it's persistence structure. It's fascinating. When we talked to Dewey Brian Arthur in episode seven, he was talking about his model that they built of a stock market with a single stock and they put the agency in and they give the agents a variety of strategies they could use to decide whether to buy or sell. And he certainly found that if you reduce the memory of the agents, you got back to a much more Newtonian view, an equilibrium view of how the stock market worked. But once you give the agents memory and once you give them the ability to try new things, they essentially started to behave in a more, shall we say, realistic version than the way they did before. They get more complex essentially, which is a lovely idea. I mean, I think many of us when we think of systems, or at least I find that they're one of the hardest concepts in complexity science when you come at it is information and is this concept of memory. This is a really nice example. In particular, you know, this interplay between the individual and the system as a whole because the individual needs a tiny bit amount of memory. If it has no memory, that means there can't be any feedback. So it needs a little bit of memory. But that memory is then when many agents together have a little bit of memory, the system as a whole can get a lot longer memory. And this is, you know, where increasing returns comes in again. I just need to remember a little bit about what agents have done a day ago, other agents. And then suddenly the system, a year from now, will remember how it started off, for example, which brings us to the final one. Adaptive behavior was that. Adaptive behavior is certainly something which we would only associate with complex systems that are alive or functional. You often see people talk about complex adaptive behavior, which I think is saying the same thing twice. Because if a system has adaptive behavior, it's complex. And it means behavior is there and it's adaptive. If it is changing according to changes in circumstances or according to changes in memory, a system is adapting to what's an example. Immune system is a functional system. And it is adapting to new pathogens that come in. And it's, you know, adaptation in the form of building up a memory in the form of sending out T cells or not sending them out. And the living system is adapting, a flock of birds would adapt if the predator comes in, they would, you know, the flock would, for example, dissolve and then form back again afterwards. Which also means that adaptive behavior is a form of robustness and resilience because if a system does not adapt, then it's much more vulnerable to perturbation. So if a flock of birds or a shell of fish does not adapt to, you know, shark coming in and just, I'll be eating. So, I mean, this is where you really start to see all of these, these products sort of pull in the same direction, don't you? That, you know, as you say, you can't have adaptive behavior with a memory. And this is when all of this starts to sort of come together. And that's why, as you say, it's kind of hard to unpick some of these products. You put them together, but they set a blend a little bit at times or they depend on each other. They do. They do. And I guess it would be strange if they didn't. But once you, once you see them, well, separate as well as together, you get a sense of how these systems can come about and how they can survive or maintain their function. The modularity that we've, we're struggling a little bit because it's a complicated concept. But here, modularity of the brain is also a form of robustness and a form of adaptive behavior. So if parts of the brain are damaged, then other parts of the brain can take over that are usually not originally not in terms of their function made to perform this particular task. But because the heart of the brain that's supposed to do it has gone down, it's not there anymore. The brain can adapt. And because of the modularity, other parts of the brain are not damaged because they're separated enough. But then the adaptation happens on the level of these modules of the brain, which is an extreme form of resilience, really, to damage. So we've talked about four conditions over the two episodes and we've talked about the sixth product and we've talked about emergence. And these are all covered in really excellent detail in your book with James Letterman called What is a Complex System? Why do you write the book? What was important for you in discussing these aspects of complex systems? Why, what drove that? The short answer is that one day James came into my office and he, he asked me, so Karolina, what is a complex system? And I didn't have a good answer. That was the beginning of a long friendship. Ten years later, the book was finally written. So it was really a long process. And partly what drove it was, for me, it was this, I wasn't satisfied by, for example, measuring complexity as a single number. I couldn't quite see what exactly are you measuring when you measure complexity. And while talking with James, I realized, oh, well, it doesn't make sense because there's so many things that are within this concept of complexity. So only if you unpack it, can you really understand what it means for a system to be complex? And you cannot measure it with a single number. I like to unpack things and that's the way I understand them. Which is, you know, the scientific method of unpicking things, understanding the parts and then putting it back together to understand the whole. That's how we do science. And this is also how we do complexity science. You read the opposite at times. And it's just not true. It is still the reductionist method that we need to use to understand a complex system. So explain that. But I think that's terribly important. How can we use the reductionist method to explain a complex system? Well, we do it. All the examples we've seen say the historical example of the Medici, we unpick them by looking at, well, what makes up that social system in back in those stages, which people are involved. How do the people interact with one another? That's a way of unpicking, you know, zooming in, unpicking the elements of the system, unpicking how they interact. And then we put it back together again to explain how can it be that the Medici were so powerful in those days. And that this is the family remember, whereas other families we've forgotten about. The same is true for scaling laws. So we see them in the statistic and we want to understand how they come about. Well, we do that by unpicking the system, by zooming in, by looking at the structure of say the blood vessel system and then putting it back together again to understand how the scaling law came about. We have to do that unpicking to understand the mechanism underlying the non-linearity, underlying the tipping points, underlying the adaptive behavior. It's the scientific method that we use to do that. And power on this topic, is it fair to say that so much of science, it still is this focus on this sort of top down explanation shall we say, you're trying to get down that one equation that describes the system. And I think for me, the revelation in complexity science is the sort of the giving up on the need for that, the giving up for the desire for that simple explanation or that elegant explanation shall we say. And instead saying, let's get into the messiness and let's what comes out of the messiness. Let's not try and tie a bow in that. Rather let's understand the interactions and then work our way up to try and build an understanding of the emergent behavior we're getting out of it. What you've just described as stochasticity in the system is really the essence. And even when you come from physics where you'd initially learn about Newtonian mechanics and things are all deterministic, stochasticity comes a lot later if at all. And I make a parallel here which might make sense to you, which is the discovery of quantum mechanics which through people off because it meant things were not deterministic. On a very fundamental level, much more fundamental than we're talking about here. But it's the same confusion about, well, how come that I can't perfectly predict things. And this is true for everything we see in everyday life. It can't perfectly predict them. But if we understand them including this noise, then that actually makes them more predictable, funnily enough. Really enjoyed in Dobie Breiner's episodes. The way he talked about the quest to the economics was for equilibrium and the quest in physics was somewhat the same. But the biologists were totally cool with not equilibrium for a long, long time. In their world, they started off going, there is no one equation we're going to get here that's going to make this. And for them, he said it was by processes and understanding the dynamic processes that are taking place. And I mean, that is a lot of what we've talked about in this episode on the last. Yes, it's funny, isn't it? We're used to talking with something and for someone from a different discipline, it's their daily bread and butter. And that would be stochasticity for biologists. And I suppose that's the beauty of, as David Cracker said in episode one, we push the disciplines into the background and we pull to the foreground the things that sort of unite the disciplines, concepts of energy and information and all those sorts of things. Yeah, we pull to the forward unite them and we translate concepts from one field to another. And that actually advances both fields, which is part of the goal of complexity science, I would say. Carolina, thank you very much for being on the show. It's been a pleasure. Thanks for listening to Simplifying Complexity. When we look at the key concepts of complexity science, what expert minds from across the world, concepts like emergence, self-organization, adaptation, networks, scaling, tipping points and much more. This podcast was produced by Brady Hayward and Wavelin Creative. To make sure you don't miss an episode, be sure to subscribe to or follow the show in your podcast app. I'm Sean Brady and I'll see you in our next episode. [Music]