COMPLEXITY: Physics of Life

Ep 4: The physics of collectives

Episode Summary

How do groups solve problems? Are there conditions that create a pathway to innovation and groundbreaking inventions? In today’s episode, we look at the science of collectives to learn about the patterns that emerge as human societies grow, the importance of a collective structure to foster ideas and create impact, and – from collectives like ants and immune systems – the importance of veering off the beaten path to become better at exploring and discovering.

Episode Notes


Hosts: Abha Eli Phoboo & Chris Kempes

Producer: Katherine Moncure

Podcast theme music by: Mitch Mignano

Follow us on:
TwitterYouTubeFacebookInstagramLinkedIn  • Bluesky

More info:

SFI programs: Education

Complexity Explorer: 

Fractals and Scaling 

Fractals and Scaling: Toward a Theory of Urban Scaling

Introduction to Complexity: Ant Foraging and Task Allocation


Papers & Articles:

Episode Transcription

Melanie Moses: Following around individual ants, you can't help but feel these are the stupidest creatures on Earth. They do just crazy, crazy things. You know, you give them big piles of food and they just wander away from them. In retrospect, I now see their genius. But at the time, I was amazed at their lack of ability to do obvious things to find food. 


Abha: From the Santa Fe Institute, this is Complexity

Chris: I’m Chris Kempes.

Abha: And I’m Abha Eli Phoboo.


Abha: So far in this season, we’ve covered a lot of ground. We’ve looked at the fundamental constraints on evolution, the biosphere, and even life out in the universe. But today, we’re going to turn the focus onto ourselves. 

Chris. Yes! Today we’re looking at human societies with two of my colleagues who study cooperation and collectives. 

Abha: But even though today’s episode is really about cities and societies, only one of our guests is actually going to talk about research with humans. 

Chris: Right, and when you think about it, it’s striking that for researchers like me who are trained in physics, studying humans is one of the hardest things you can do. Physicists are happy to look at space and black holes and theorize about things that we will never, ever see in our own lifetimes. But humans? Humans are too hard. They’re too complicated and unpredictable.

Abha: Right, and we’ll get into why this dynamic exists later on in the season. But I should point out that Chris, you actually do study humans, to some extent. You look at cities, societies, companies and organisms.

Chris: [Laughs] I do, yeah. And it is complicated. One of the researchers I work with is Hyejin Youn. We’ve worked on the laws of regulation. This seems really boring, right!? But if you think about it all complex adaptive systems have to coordinate and manage their functions. This is essential and we’ll talk about this later in the episode. First, let’s hear from Hyejin.

Hyejin: My name is Hyejin Youn. I'm external faculty at SFI.

Abha: Hyejin is a professor at Northwestern University’s Kellogg School of Management and she’s also core faculty of the Northwestern Complexity Science Institute. She considers herself to be a bit of an oddity. She’s a physicist by training, but she works at a business school.

Hyejin: In a conventional sense, my trajectory, the curvature of my trajectory is pretty, yeah, pretty significant from quantitative and like STEM based all the way to qualitative. My department is one of the most qualitative fields even in business school. So I think that would be my fun fact. 

Chris: So you are a strange fact. 

Hyejin: [Laughs] Yeah, myself. Yeah.

Abha: But for Hyejin, this path wasn’t random. In fact, she’s inclined to think that very little about the world is random.

Chris: Do you have a moment in your childhood where you remember first being interested in science?

Hyejin: Oh… Okay. Should I be very honest here? [Laughs]

Chris: Absolutely you should be very honest, yeah.

Abha: As long as you don’t mind hearing it back. [Laughs]

Hyejin: To be honest, a lot of people who have a very like, inspiring career trajectory always say that, “Oh, when I was young, I had this like, incredible vision.” But for me, it was not, like, I came from this small town in Korea. I mean, it's not small town in American sense, but in Korean sense, it's small town. Asian sense is small town. I didn't know like whether I could become academic because, so I'm the first person who graduated four year full education university. I'm still the only one who got PhD in my entire extended family. 

Abha: Hyejin wasn’t really interested in science at first. She actually wanted to draw and paint. But one day, she saw something on TV about a science high school. 

It was funded by the Korean government, and if she got in, she could go for free and skip the last year. Once she got to high school, Hyejin gravitated toward physics, but —

Hyejin: It was not like, I had a vision for physics. It's not that I just like follow the trail that is… yeah. And then I came to my university, which is also a fully government-funded university to promote science and technology for the government, to make our country wealthier. Some people just don't have a vision because they didn't have like opportunities to see what's out there to become what. So like I was more of a, this actually connects to my study, one of the study that I try to understand the structure of the world and the adjacent possible is not equally distributed for everyone. Some people just have a different set of adjacent possible to become something.

Abha: In other words, the idea that you can grow up to be anything or do anything, that’s not always true. At least, not for many people. Our circumstances determine which paths are available to us. Hyejin studies the growth of cities and the innovations that arise from them, and she says this outlook applies to her work too. It’s a familiar theme of our show: what often seems random and chaotic is actually, when you look a little closer, following big, fundamental rules. And in Part One, we’ll see how those rules shape the societies we live in.

Abha: Part One: What’s so special about cities?

Chris: Before we get into Hyejin’s work on cities, let’s think back to the biosphere for a moment. And in particular, trees. Even some of the biggest trees, like sequoias, have limits on their size. Because growing upward is a fight against gravity, and at a certain height, the tradeoff just isn’t worth it. So that’s one limit. 

But even if a tree can’t grow up forever, it can still grow out. But if we think about growing out, that comes with tradeoffs too. The tree has to deliver nutrients and water from its trunk all the way to the ends of its branches. And that takes force. Gravity applies here too, but there are other things coming into play like the viscosity of water and how the plant cells absorb everything. And so we don’t have trees that can grow infinitely out either. As you scale from small trees to huge ones each of these factors exerts influence in regular, predictable patterns. And those patterns are, as we’ve said before, not linear. 

Abha: Okay. So let’s look at how this connects to cities.

Hyejin: Society has a structure, physical and cognitive, social, economic structure. The structure is everywhere and this structure is not random. We romanticize the random encounter and the serendipity, but I don't know how much random is really random. So that actually brought me to think that city has some trajectory. 

Chris: Cities can also grow up, and they can grow out. So if the population doubles, that doesn’t necessarily mean that the number of roads will double too. Or the number of libraries or grocery stores. There are scaling laws at play here too. And so far, we’ve only been talking about literal, physical characteristics, like height and density and lifespan. But Hyejin says the scaling laws can also shape more abstract outcomes.

Hyejin: Socioeconomic properties like crime rate, creative productions, and wealth production seems to superlinearly scale with the city size as opposed to infrastructure.

Chris: There are many different ways to define things like innovation, crime, and wealth — more than we could get into in just one podcast episode. For innovation, Hyejin and other SFI researchers have chosen specifically to look at business patents, which imply the creation of some sort of new technology, product, or service. These increase superlinearly — as in, disproportionately fast — as a city’s population grows. 

Hyejin: I would rely on a few studies, regional science and also urban scaling studies, that show there is a super linear production of patent activity as a function of city size. So double the size of the population, the empirics shows that the 15% more patent activity per capita. Actually, this 15% more of the patent activity is exactly same amount that you expect from crime activity. So you have a more, 15% more of the patent activity per capita, also 15% more of crime activity per capita. 

Chris: So, something interesting is happening when communities form. What’s being created is more than the sum of its parts, in both good and bad ways. It’s possible that whatever special sauce is making innovation increase is also influencing the increase in crime. And the rates at which they scale up follow a curve in the same way that plants and animals are bound by the scaling laws. But, like the biosphere, idiosyncrasies in each environment means that some cities don’t adhere perfectly to these curves. 

Hyejin: I try to be very cautious to present this dynamic of the city because people will just say that, “Oh there are cases that don't work for the scaling law. One of the example is Los Alamos.” Like some people will say “Yeah Los Alamos is so small and yet they produce a very much innovative outcome.”

Chris: Yeah, and I think, you know, the thing you mentioned about Los Alamos, which is a famous town that everyone knows about now because of the new Oppenheimer movie, you know, it's not a natural state. It also brings up this thing that you and I worry about all the time, which is how do you actually measure size? And so if you measure Los Alamos by number of people who live there, it's very small. If you measure it by total number of government dollars spent, it's very high, right? I mean, it's a huge city if you just do federal spending in that town and that sort of I think explains some of the innovation because you're that even though there are not so many people there the government is spending huge amounts of money.

Abha: And even if a small city isn’t given a huge injection of money, it can still produce innovative outcomes. It’s just less likely to happen in a small city, and more likely in a big one.

Hyejin: There is another fact that I want to emphasize here. Novelty should not be misunderstood with innovation. So innovation requires not only novelty, novel idea, but also impact. So to make an impact requires some social structure, socioeconomic structure to support socioeconomic capacity, the capital to support your idea to propagate and diffuse and implement it in the society and economic systems. And if you are in a small city, even if you may have a very good idea, you may not be able to propagate or you may not be able to align with other people's idea and infrastructure because you are disconnected to the larger system to make a good, big impact. So there are so many things that are going on that give the larger cities premium. 

Abha: So at the moment we currently have Zoom and you know, features like Riverside, which we're currently using. But what if we were to go into the metaverse and you had the possibility of projecting your virtual self into a collaborative environment, which is virtual, would that still hold true? 

Hyejin: Oh, that's a very good question. It always depends on the level of technology to enable us to communicate with many different bandwidth communication channels. So communication is not just to send my verbal statement. Communication involves many different channels for information transfer. So it's not only just me talking, but also like my eye movement, my gestures, and your gestures that actually come together to to restructure the information to be transferred from my mind to your mind. And now coming back to the question, Zoom actually expanded this bandwidth, the communication bandwidth so much that it's actually easier to communicate and the transferred knowledge, but it's not enough because it's enough to transfer the tacit knowledge that is not really codified well. So metaverse, coming to the metaverse, it depends on the like how much metaverse can really capture our information transfer from the listener to the speaker or the speaker to the listener. I think that's the main factor. 

Abha: Until technology manages to bridge physical distance — if it ever does — cities are important places to foster new creations. And as much as American business culture loves to idolize individual visionaries — 

Hyejin: We often celebrate it, and individuals' agentic decisions, we often emphasize in innovation literature and business literature 

Abha: But it’s not really about one person coming up with an incredible idea. It’s about the structure of the communities they’re in, the ideas coming together, and something taking hold because of a particular moment in history.

Hyejin: The empirical evidence is that we find multiple inventors coming up with the exactly same idea. And the examples are abundant. One example, like famous example is Wallace versus Darwin, who came up with the exactly same idea, and yet Darwin had more credit. Newton versus Leibniz who had very similar idea calculus in a different form, but it seems like there are numerous cases to show that the people come up with the same idea at the same time. It seems like time is ripe to come up with a certain idea. And that means individuality becomes less important. It's more like collective... the structure toward a certain dynamic or certain pathway becomes more important when it comes to innovation. Because at the end of the day, you can come up with a really crazy idea, but what idea is picked up by society and make more development and implemented society is always path dependence. So there is some pathway that actually somehow takes on its own life.

Chris: Hyejin is continuing to study what she calls the “innovation pathway,” this overall structure in our world that builds over time and results in important, groundbreaking inventions. Which, in a sense, is a pretty radical idea — it raises questions like whether or not individual people really change the course of history. Collectives push the edges of knowledge in a way that individual people, or organisms, just don’t.

Abha: In Part Two, we’ll look at some other types of collectives — ants and the immune system — and see if they can teach us a thing or two about novelty. 

Abha: Part Two: How to explore

Melanie: My name is Melanie Moses. I am a professor of computer science at the University of New Mexico and also external faculty at the Santa Fe Institute. And I have a secondary appointment in biology at UNM. 

Abha: Melanie also studies how organisms communicate, in particular, collective intelligence.  Much of her work has focused on robotic systems, immune systems, and ants. 

Melanie: One of the things I've studied is how ants forage for seeds, the desert harvester ants that we have here in abundance in New Mexico. And I spent a couple of years during my PhD following these ants around in the desert. Melanie which is a very surreal experience of, you know, you zoom in, you're focused on this tiny little centimeter long ant in a big wide open vast desert. And sometimes you kind of lose track of where you are because you're so focused on these ants. And following them around, I studied them because I was interested in collective intelligence and how the colony would, you know, make excellent collective decisions about foraging for food. 

Abha: Ants communicate with each other in some pretty ingenious ways. 

Melanie: So ant communication is often characterized by the way that they lay pheromones. This is not all of how ants communicate. But an ant will, many species of ants, when they discover, an individual ant item of food, it will pick up that item of food, return to the nest, and lay a pheromone trail, just a chemical that evaporates over time. And when another ant discovers that pheromone, it will make a choice, do I follow that pheromone trail or not? And if that ant gets to a pile of seeds and picks up a seed and returns and also lays a pheromone trail, then the pheromone is reinforced by more and more ants. It's a great positive feedback that attracts more and more ants to a big pile of food until the food is gone. And once the food is gone, ants that follow the pheromone trail will get there, there'll be no food to pick up, so it won't reinforce the pheromone trail. It's a chemical signal, so it'll evaporate over time. And that really enables the trail to disappear when it's not needed. And so the pheromone trail can be thought of as sort of a set of trails as a map to the location of seeds. 

Abha: The system as a whole is really clever. But individual ants don’t always behave the way we think they should.

Melanie: Following around individual ants, you can't help but feel these are the stupidest creatures on Earth. They do just crazy, crazy things. You know, you give them big piles of food and they just wander away from them. So observing ant behavior, which was my idea, is, oh, this is the ideal model for intelligence, for collective intelligence. And I spent most of my dissertation feeling frustrated with the stupidity of these ants. In retrospect, I now see their genius. But at the time, I was amazed at their lack of ability to do obvious things to find food. 

Chris: Melanie and her team dyed the backs of individual ants so they could track their movements more easily.

Melanie: So once they were dyed, they were a little bit easier to just literally follow around the desert where this was at the Sevilleta National Wildlife Refuge. There's a long-term ecological research site there.

Chris: As part of their research, the team would place piles of seeds in various amounts and locations and then watch the ants forage. One day, they placed a huge pile —  256 seeds — very close to the nest.

Melanie: And one ant discovered it very early on within a couple of minutes of placing it. And that one ant, we marked and it went back and forth from the seeds to the nest and the seeds to the nest 256 times all by herself. And the other ants just wandered around completely not helping her. And so, you know, I was like, this is my model for, you know, incredible collective behavior. And they're just, they've left her, they've, you know, all on her own to collect these seeds. They weren't finding much. They were wandering. 

Chris: This really seems like the opposite of a functional collective — one ant just did all the work, and no one helped her. When one ant, like this poor ant Melanie observed, finds food, the other ants each need to make a choice: they can either go to that food source too — to exploit what’s known — or to keep searching, to explore.

Melanie: This was what really sort of drove home to me this idea that they are wired to explore for new things, even when there's obvious information that there's something there, right? If the ant decides to exploit known information, it can follow that pheromone trail and get led to the pile of food. If it decides to wander off, then it's just sort of... intentionally in some sense exploring rather than exploiting.

Abha: This exploration is actually far more beneficial for the collective group than we might think. Most of the ants that go off and explore won’t find anything. And if they weren’t part of the group, they wouldn’t survive. But because they’re part of a collective, they can take risks. 

Melanie: One of them will get a big payoff. And so that sort of exploration is a really important part of cooperation, the ability to explore and fail. But the sort of the whole collective continues on. I think over their evolutionary history, that random search was in fact tuned to be the ideal way to really continue to discover new food. And so really it's that choice point where there's a response to the pheromone, that the ant says, do I follow it or do I not? And the likelihood of not following that pheromone was much higher than I would have expected, but I think quite optimal in the long run for these ant colonies. 

Abha: So Melanie has some in-depth knowledge about ant behavior. But when she first started her career, she didn’t plan on studying ants.

Melanie So I took, I would say, maybe the same sort of meandering path that these ants took that I thought was ill-advised at the time, but now I think in retrospect maybe they were a good idea. 

Abha: She studied computer science as an undergrad. But after college, instead of going to work in an office, she moved to Costa Rica for six months. It was there, living in the rainforest, that she became fascinated with ants.

Melanie: There are lots of leaf cutters in the rainforest there. And so I became interested in understanding how they built networked societies. And I was interested in the parallel for how we did the same. I initially was very interested in studying sort of human cooperation and it was clear that that's a very hard thing to study. One, just logistically, right? You you can't put dyed powder on humans and follow them around, although people do maybe analogous things with cell phone data now. But humans are so much more complex and so much more complicated and they have so many conflicting motivations. I was actually very interested in finding simpler systems where we might sort of uncover what are the fundamental constraints on cooperating in ways that are beneficial for all of the cooperators. I was particularly interested in search questions. I think a lot of the things that I study now in robotics, ant colonies, immune systems, are really questions of how do collections of agents effectively search over long time periods so that they're leveraging, having a large population that's able to both explore for new things and exploit known information. 

Abha: And Melanie thinks these ant studies do apply to human cooperation too, and potentially all cooperative groups. As we learned in Part One, communication and cooperation are really important for human innovation. Hyejin, and many other researchers, would say that the structures underlying cities and societies are not random — and the innovations that come out of cities are not necessarily random either. But even if the overall structure is set, Melanie still sees random exploration as a crucial ingredient in the path to novel discoveries and ideas. 

Melanie: The apparently stupid aunts were actually a little smarter than I might have given them credit for. That randomness is incredibly important. You have a large group of cooperating agents. One of the great benefits is that every agent can go off in a different direction. With ants, that's literal, right? They can go off in a different direction on the landscape to search for that next pile of food that no one has yet discovered. I think that's also true in a more abstract sense, which if you think about humans, exploring the frontiers of knowledge. And I think that the concept that I've come to is that humans are very good at exploiting other human information. We're very good at copying what others have done. And I think we actually maybe socially are less good at exploring and finding brand new things. 

Abha: Ants aren’t the only ones who are better at exploring than humans. Within our own bodies, our immune systems are fine-tuned to explore in order to find pathogens.

Melanie: Particularly T cells, which I think post-COVID people are now much more familiar with T cells than they used to be. They behave very much like ants foraging for food. They are searchers who are able to kill virally infected cells. And so we thought of them as sort of this kind of swarm of searchers that were looking for infected cells to kill. Much of their search is also random. And by analogy with these ant systems, we sort of are exploring the possibility that randomness is in fact adaptive, right? If the T cells followed paths directly to sources of virally infected cells, they would miss virus that had spread far away. So it's again sort of the same principle that the explore-exploit tradeoff is maybe much more tilted toward explore than you might think for any given search problem. 

Chris: Yeah, and I think it's really interesting that you're saying for the most critical problem, discovering pathogens and dealing with them, if you get that wrong, you're dead, right? I mean, it's exponential growth, everything's happening very fast. And if you miss something, it's a real problem. And so you're saying even in the most critical situation, you still are more tuned on explore than exploit. So even when it's life or death, you use randomness to explore, and I think that really illustrates that there's something deep about how essential it is to explore. 

Melanie: Yeah, that's very well put and very much why we thought that T cell search is so interesting, right? Because it is so time critical. You need to get it right.  If there's a large infected area, you need lots of T cells there. T cells can kill a handful of cells. And so you need more T cells in places where there's more infection. So there is some aspect of recruitment, but this randomness to go look for… you know, that little spot of infection that's far away from the other infection is also critically important. 

Chris: So are you saying that we should spend more time, spend more resources, encouraging people to go off and do random, seemingly unrelated, crazy things just to sort of speed up discovery? 

Melanie: Yeah, I think that would be my advice to funding agencies like NSF. I think that over time, the funding agencies have become more conservative. There are obvious wins for doing the next step, for mapping the next genome, for discovering the function of some particular protein. But radical new things, I think, are quite risky. Most of them won't work. And so I do think that that's an important lesson that we should learn from these ants, Melanie there's a lot of value in exploration, even when most of that exploration fails. And so, yeah, I think that that's important. I think we give ourselves credit for being extremely innovative, and maybe we don't quite deserve all that credit. 

Abha: If we think back to cities and societies, and some of the work Hyejin’s done, it makes sense that when you get more people together, there are more opportunities to explore and fail — or to explore and get a huge payout that benefits everyone. And remember what Hyejin said earlier? Innovation isn’t just about novelty.

Hyejin: So innovation requires not only novelty, novel idea, but also impact. So to make an impact, it requires some social structure, socioeconomic structure to support socioeconomic capacity, the capital to support your idea to propagate and diffuse and implement it in the society and economic systems. 

Chris: But growing comes with a price. In Part Three, we’ll learn what that price is.

Chris: This is Part Three: The cost of the collective

Chris: On this show, we’ve talked about many potential rules of life, like the scaling laws, reproduction, and the presence of parasites, just to name a few. Hyejin’s adding another one to the list: regulatory functions. As in, some sort of management system that keeps everything in check.

Hyejin: Regulation is an interesting term, and can be perceived differently and reacted differently from social scientists all the way the physics and natural scientists, in a sense that regulation for natural scientists and engineers is kind of a good thing to regulate system, to make healthy function. But regulation also has some connotation that restrict and constrain your free will and your individuality. 

Abha: Depending on your views, regulation could be your ideal, or it could be something close to a dirty word. But no matter how you look at it, one thing that’s clear about regulation is every system needs it. 

Hyejin: I could use the word regulation or I could use the word any kind of like mechanism that put individual or cells or a nervous system or people in the city or people in employees in a company together to function. And to make a coherent function requires regulation, we call it regulation. 

Abha: Hyejin and a team of researchers from SFI looked at the way people collaborate on Wikipedia.

Hyejin: We look at Wikipedia because individuals somehow come together to contribute the coherent knowledge structure without knowing each other that much, through this medium with a goodwill across different spectrum of knowledge background.

Abha: And in order to create coherent structures and articles on Wikipedia, regulation naturally emerges.

Hyejin: What are the costs involved in this becoming coherent output? They have to talk to each other, they have to check with each other, some people have to punish other people who misbehave, and even if it is a really flat platform. You also need the administrator's activity. You also need the rules in the end. You also need to get some bureaucratic systems at the end of the day, even if it is a voluntary system. So what are the mechanisms that make individual human being or individual components membership become sort of organization. And this regulation seems to be really fundamental property for complex system, for that matter, complex system with all this nested structure. 

Chris: As people come together, they can create more, and they can discover more. But there’s also more overhead. And even in mammals and cardiovascular networks, the larger an animal gets, the harder it is to pump blood out from the heart to the rest of the body. The size of the aorta has to get bigger and bigger to keep up. And on top of that, there are opportunities for things to go wrong — errors, miscommunications, misleading information, damage. Melanie does research in this area too, trying to find ways to expand systems without slowing them down. This is probably a familiar scenario for anyone who’s worked at a growing company or in government. Or if you’re an academic like me, and you find yourself trying to avoid spending your entire day on email.

Melanie: Over time, more and more of your email seems useless, but there's a few critical things that are really important in your email. And I think of those in a scaling sense, right? We're building up a social world where we're more and more able to communicate with each other. Some of that communication is extraordinarily valuable, right? Somebody is going to tell you, maybe in an email, that key idea that you really needed to solve the next problem. But a lot of what's in your email it's just overhead. So I think of that sort of in the same way as the aorta. It's the price we pay to be in a system where communication is possible. And that overhead is actually in some sense important and while I think there are probably computational ways to minimize it. It's really important to keep eight billion people on this planet connected to each other. So when we look at the misinformation problems in social media and all of that, in some sense I now look at those as the price we pay for being in an interconnected world. And I hope somewhere in there there's a pathway to think about how we can make that overhead sort of less painful. But I don't think we can make it go away. 

Chris: So until we figure out how to make this less painful, I will be forced to keep checking my emails and dealing with all this overhead, no matter how useless it feels.

Abha: And so will I. But the upside of that overhead — the reason you and I are here at SFI — is that we get to be part of this collective group of people taking shots in the dark with new ideas, theories, and creations. 

Chris: Many of which will fail, just like the ants who go wandering around in the desert and find nothing. But hopefully, some of what we do here will get that big payoff, and it’ll stick and contribute to the world in a meaningful way.

Abha: Chris, do you think your work is heading toward that big payoff?

Chris: [Laughs] We’ll have to see, I guess. Time will tell.


Abha: Coming up on Complexity, we’ll pull the frame all the way back and ask: -  how do we know what we know?

David Krakauer: Facts are just sort of accidents of history in some sense.

Abha: That’s next time, on Complexity. Complexity is the official podcast of the Santa Fe Institute. This episode was produced by Katherine Moncure, and our theme song is by Mitch Mignano. Additional music from Blue Dot Sessions, and the rest of our sound credits are in the show notes for this episode. I’m Abha, thanks for listening.