COMPLEXITY: Physics of Life

Carlos Gershenson on Balance, Criticality, Antifragility, and The Philosophy of Complex Systems

Episode Notes

How do we get a handle on complex systems thinking? What are the implications of this science for philosophy, and where does philosophical tradition foreshadow findings from the scientific frontier?

Welcome to COMPLEXITY, the official podcast of the Santa Fe Institute. I’m your host, Michael Garfield, and every other week we’ll bring you with us for far-ranging conversations with our worldwide network of rigorous researchers developing new frameworks to explain the deepest mysteries of the universe.

In this episode we speak with Carlos Gershenson (UNAM website, Google Scholar, Wikipedia, Twitter), SFI Sabbatical Visitor and professor of computer science at the Universidad Nacional Autónoma de México, where he leads the Self-organizing Systems Lab, among many other titles you can find in our show notes. For the next hour, we’ll discuss his decades of research and writing on a vast array of core complex systems concepts and their intersections with both Western and Eastern philosophical traditions — a first for this podcast.

If you value our research and communication efforts, please subscribe, rate and review us at Apple Podcasts or Spotify, and consider making a donation — or finding other ways to engage with us — at

For HD virtual backgrounds of the SFI campus to use on video calls and a chance to win a signed copy of one of our books from the SFI Press, please help us improve our scicomm by completing a survey linked in the show notes.

Or just a copy of the recently resurfaced SFI Press Archival Volume Complexity, Entropy, and The Physics of Information.

There’s still time to apply for the Complexity GAINS UK program for PhD students – apps close March 15th.

Or come work for us! We are on the lookout for a new Digital Media Specialist, an Applied Complexity Fellow in Sustainability, a Research Assistant in Emergent Political Economies, and a Payroll, Accounts Payable & Receivable Specialist.

You can also join our Facebook discussion group to meet like minds and talk about each episode.

Podcast theme music by Mitch Mignano.

Follow us on social media:
Twitter • YouTube • Facebook • Instagram • LinkedIn

Mentioned & Related Links:

Carlos publishes the Complexity Digest Newsletter.

His SFI Seminars to date:
A Brief History of Balance
Emergence, (Self)Organization, and Complexity
Criticality: A Balance Between Robustness and Adaptability
Festina lente (the slower-is-faster effect)
Antifragility: Dynamical Balance

W. Ross Ashby & The Law of Requisite Variety

by Timothy Morton

How can we think the complex?
by Carlos Gershenson and Francis Heylighen

The Implications of Interactions for Science and Philosophy
by Carlos Gershenson

Complexity and Philosophy
by Francis Heylighen, Paul Cilliers, Carlos Gershenson

Heterogeneity extends criticality
by Fernanda Sánchez-Puig, Octavio Zapata, Omar K, Pineda, Gerardo Iñiguez, and Carlos Gershenson

When Can we Call a System Self-organizing?
by Carlos Gershenson and Francis Heylighen

Temporal, Structural, and Functional Heterogeneities Extend Criticality and Antifragility in Random Boolean Networks
by Amahury Jafet López-Díaz, Fernanda Sánchez-Puig, and Carlos Gershenson

When slower is faster
by Carlos Gershenson, Dirk Helbing

Self-organization leads to supraoptimal performance in public transportation systems
by Carlos Gershenson

Dynamics of ranking
by Gerardo Iñiguez, Carlos Pineda, Carlos Gershenson, & Albert-László Barabási

Self-Organizing Traffic Lights
by Carlos Gershenson

Dynamic competition and resource partitioning during the early life of two widespread, abundant and ecologically similar fishes
by A. D. Nunn, L. H. Vickers, K. Mazik, J. D. Bolland, G. Peirson, S. N. Axford, A. Henshaw & I. G. Cowx

Towards a general theory of balance
by Carlos Gershenson

A Calculus for Self-Reference
by Francisco Varela

On Some Mental Effects of The Earthquake
by William James

Self-Organization Leads to Supraoptimal Performance in Public Transportation Systems
by Carlos Gershenson

Alison Gopnik on Child Development, Elderhood, Caregiving, and A.I.
Complexity Ep. 99

Simon DeDeo on Good Explanations & Diseases of Epistemology
Complexity Ep. 72

David Wolpert on The No Free Lunch Theorems and Why They Undermine The Scientific Method
Complexity Ep. 45

The Clock of the Long Now: Time and Responsibility
by Stewart Brand

Michael Lachmann

Stuart Kauffman

Andreas Wagner

Cosma Shalizi

Nassim Taleb

Does Free Will Violate The Laws of Physics?
Big Think interviews Sean Carroll

Episode Transcription

Machine-generated transcript below.  Human edit coming soon!

Carlos Gershenson (0s):

If we just look at chemical reactions from all possible chemical reactions, most of them will never take place in our universe. Just so many that there's no possibility that they will happen. And still we have such a rich chemistry that led to biology and the same all possible biological interactions will never take place all possible social interactions will never take place, all possible ideas will never materialize. But still that minimum infinitesimal region of the space of the possible that actually gets instantiated even just once in the history of a universe, needs to be viable in in some way, even for a short moment.


Carlos Gershenson (45s):

And of course, that constraints which ones will survive or which ones will endure, which ones will evolve.


Michael Garfield (1m 16s):

How do we get a handle on complex systems thinking? What is the implications of this science for philosophy and where does philosophical tradition foreshadow findings from the scientific frontier? Welcome to Complexity, the official podcast of the Santa Fe Institute. I'm your host, Michael Garfield, and every other week we'll bring you with us for far-ranging conversations with our worldwide network of rigorous researchers developing new frameworks to explain the deepest mysteries of the universe. In this episode, we speak with Carlos Gsen, SFI sabbatical visitor and professor of computer science at the Universidad, where he leads the self-organizing systems lab among many other titles you can find on his website.


Michael Garfield (1m 58s):

In our show notes for the next hour, we'll discuss his decades of research and writing on a vast array of core complex systems, concepts and their intersections with both Western and eastern philosophical traditions. A first for this podcast, if you value our research and communication efforts, please subscribe, rate and review us at Apple Podcasts or Spotify and consider making a donation or finding other ways to engage with us at Santa For HD virtual backgrounds of the SFI campus to use on video calls and a chance to win a signed copy of one of our books from the SFI press, please help us improve our sycom by completing a survey linked in the show notes or just buy a copy of the recently resurfaced SFI press, archival volume, complexity, entropy, and the physics of information.


Michael Garfield (2m 46s):

There's still time to apply for the Complexity Gains UK program for PhD students apps close March 15th or come work for us. We're on the lookout for a new digital media specialist, an applied complexity fellow in sustainability, a research assistant in emergent political economies and a payroll accounts payable and receivable specialist. You can also join our Facebook discussion group to meet like minds and talk about each episode. Thank you for listening, Carlos Grenson, it's a pleasure to have you on Complexity podcast.


Carlos Gershenson (3m 20s):

Pleasure to be here.


Michael Garfield (3m 20s):

Yeah, it's been great watching you give these thematic talks in our seminar series over the last few months and I'm excited to see the whole thing and we'll link to the first five and we're gonna, we're gonna dip into a little bit of that in the show, but you've also written a lot of interesting stuff at the intersection of complex system science and philosophy that I wanna get to. So anyway, let's just start by giving people a little bit of background about who you are as a person and how it is that you came to devote your life to the exploration and the inquiry around all of this stuff.


Carlos Gershenson (3m 54s):

Well, I guess I was curious since I was a child, my parents stimulated at curiosity. For example, when I was six, my dad took me to a program that was organized by the Mexican Academy of Sciences to teach kids how to program with logo. I had many teachers that also stimulated that in different ways back in Mexico City. And I studied their computer engineering and also a bit of philosophy later I went for a master's in evolutionary adaptive systems at Sussex University in the uk. So that's about 20 years ago. Then I did my PhD in Brussels at the free university and then New England Complex Systems Institute. Then I joined the faculty back in Mexico City at the National University.


Michael Garfield (4m 37s):

So you've, like I was telling you before we started this call, your list of research citations is extensive, especially someone who seems as young as you're, so I've talked about David Crackower about this. He says there's two kinds of people that seem to populate our research network and some of them are people that are interested in very specific narrow dominions, but they want to bring this whole pallet of complex systems tools to it. And then there are these people that are just interested in everything and you seem like one of those second category. So I'm curious what kinds of questions based on this enormous bouquet of stuff animated, you continue to animate you.


Michael Garfield (5m 22s):

Like how would you delimit or explain what it is? It seems like you came in through computer science, so like yeah, what are you really, what windmill are you really tilting after in your career here? Yeah,


Carlos Gershenson (5m 35s):

I chose computer engineering at the university. I studied because it was like the broadest program I could find. So it was not only computer science, it was also math and physics and even economy and philosophy and history because I was interested in all of that. So think that you can use the lens of complex systems to observe any phenomenon, it can benefit any discipline. So in a way, I've tried to do that at the theoretical level, at the engineering level, even the philosophical level and also at the arts level. So in my case, it has been, I think, combine all of this because for example, you will have better theory if you manage to test it with applications.


Carlos Gershenson (6m 18s):

And of course your applications will be better grounded in solid theory. And of course you can have better arguments to philosophize about it if you already have experience with trying to solve real-world problems. And I guess in art you can explore creativity that can be beneficial also for the other three.


Michael Garfield (6m 37s):

Well, what I'd like to do, I think is some, we don't typically do this on the show, but because you've written both your own quantitative research as well as lit reviews, as well as more like philosophical pieces about the way that complexity science changes classical thinking in philosophy, I'd like to run that circuit actually kind of in a circle. Yep. And start by inviting you to break down for us this paper. Francis Halligan. Yep. Who was that your PhD advisor? Yes. You've written a lot of papers with this guy, especially early on, and you've got this one in particular. How can we think the complex Yes.


Michael Garfield (7m 17s):

That was a book chapter. And I'd like for you to just break this down cuz you make about half a dozen key distinctions between complexity thinking and classical or Cartesian thinking. Yes. And even if somebody's been listening to this show for 101 episodes, I find that the distinctions that you've made here are really clear. So take us through please.


Carlos Gershenson (7m 40s):

Yep. So indeed during my PhD, the title of my thesis Design and Control of Self Organizing Systems, but then it was also something I felt like doing to understand better what was complexity, what was self-organization, what was emergence and so on. But the fact that I did during my PhD hasn't stopped. So I still write and think about these issues. We also collaborated with Paul Sears from South Africa who passed away a few years ago. I have the opportunity to interact with Moran in a couple of occasions. So trying to summarize everything that has been done because we could spend several episodes just on this topic, like traditional philosophy, which we could call reductionists, tries to simplify phenomena to predict that has been extremely successful.


Carlos Gershenson (8m 30s):

We have duplicated life expectancy because of these cars, air travel, internet, everything is thanks to reductionists. But there's this phrase of mora gelman reductionism is correct but incomplete. So what is missing from reductionism? When you simplify, you tend to ignore interactions and that's what characterizes complexity. So what, what happens when you put attention to interactions or you consider them as something real, not something that's just something extra. And then you perhaps most relevant aspect is that your prediction is limited.


Carlos Gershenson (9m 10s):

Reductionism tries to simplify or to predict which is desirable because then you can try to solve problems before it's too late. However, if you have complexity meaning interactions, these will generate novel information. And we are talking in high school that if you have, let's say initial and boundary conditions and the laws that rule the system, you'll be able to predict the future. And this has its roots in Lala system. But we know that because of these interactions, they generate new information. So it's not enough to have initial amount conditions and the laws that govern a system you need actually to run the system. And the technical term for this is computational irre reducibility. So then what to do when you can't predict what we have been exploring is to adapt.


Carlos Gershenson (9m 52s):

And self colonization is one way of adapting. And like that many pieces start to fall down that where implications of our reduction is scientific perspective that have been taken as true in philosophy. So we explore all of that and we keep on exploring all of that and what are the implications of considering interactions, what should we do with our limited predictability? And in general how to face complexity.


Michael Garfield (10m 22s):

So one of the things that you mentioned also in the work you did with Dirk Helbing on the slower is faster Yeah. Effect. You talk about, we just had Alison Gopnik on the show and Gopnik talks about what she calls the explorer exploit tension. Yeah. Which is, you mentioned this in a couple places that due to demands made on decision making that there you can act with too little information or you can act with too much. And classical thinking tends to, like you just said, tends to assume that you're gonna make better decisions if you have all of the inf This is like assuming that it's even possible and we know that it's not. But yeah, I'd love to hear this piece cuz this also came up when we were talking with Simon Dadeo about Simon sees science as successful in large part because there's a tension between the ways that different people care or like the aesthetics they have about what constitutes a satisfying explanation.


Michael Garfield (11m 21s):

Some people want a a an all embracing conent explanation and some people want something that's very limited and parsimonious and can fit on a t-shirt. And that's, that's the difference. And that's kind of related to the way that Alison Gopnik talks about the way that children kind of relate to problem solving and ex and the way that adults do. And so just one more thing I'll stack on this is when we're, when we talk about in in this and in other work, actually you talk about it kind of skipping ahead to all of these things, but you talk about it in the piece that you co-authored with a team led by Fernando Sanchez Puig on heterogeneity and how systems will tune themselves to have heterogeneity across time.


Michael Garfield (12m 6s):

And so you talk about some of the work of somebody like Andre Deus who looks at the way that ecology benefits from understanding organisms as having these diff distinct life stages where the larva might be feeding on something else from the adult. And so yeah, so there's, I don't know, there's a lot there. Yeah. But I'd love to hear you talk a little bit about the value of heterogeneity and diversity and how it also relates to these questions that you pursued with Dirk about when we get counterintuitive results in how systems adapt by splitting up and exploring multiple different strategies.


Michael Garfield (12m 47s):



Carlos Gershenson (12m 47s):

So I think all of these can be seen as particular cases of balance in sense that's, let's say explor exploration exploitation. We want to have balance between both. That will, let's say make search ottum. And of course we know from work of David Wil Northland serums that depending on the search pace, basically you need different strategy to search that search space. So this balance is moving, which is the same what happens in the slower faster effect. And also the case of Simon in some cases you, you might want to know all the details and to predict all the information of a phenomenon. But then that of course has a very high cost. You need lots of information.


Carlos Gershenson (13m 29s):

And on the other extreme, you might want to have very general theory, but then that won't be very practical because it'll say something like stuff is, and that's always true. And, but then there's not much you can do with that. So of, again, the balance lies depending on the context that you're trying to apply a theory and what are your interests or the purpose of that theory, what are you trying to achieve? And then the, let's say, how much detail is needed will depend on that. So again, a balance, but you cannot specify what's the optimum balance beforehand. So it's something that's, that is shifting, letting it to heterogeneity. It seems that when you don't know beforehand what's the optimum balance that will be best.


Carlos Gershenson (14m 12s):

If you have a variety of elements either spatial or temporal or functional, then you don't really need to find the precise parameters. You don't need to find two parameters because some of those elements probably will be close to the optimum. So then you can exploit those seems. So it is something general that nature does. And I don't know whether with a specific purpose or it's just easier to have heterogeneity because actually you would need to put some effort into having homogeneity in many cases.


Michael Garfield (14m 47s):

So this, we've never actually talked about this on the show before, but since you're not the only Buddhist on campus here, we're here, we're in Fred Cooper's office and I've had some great conversations with him about this particular intersection. You've written some interesting work on the relationship between Buddhist philosophy and complex system science. And something that came in again on this, how can we think the complex with Hagan is you talk about how in more classical thinking Aristotelian logic phenomenon begin belongs as you right to either to category A or not A, it cannot be both, neither in between or it depends. And yet Buddhist tetra logic also includes it can be both A and B or neither A nor B or these kinds of things.


Michael Garfield (15m 34s):

And then later on in the section on indeterminacy, you make a distinction from a paper you wrote in 2002 between absolute being and relative being, which is now even be, say you picked up Tibetan Buddhism in 2009, but this is something I've really only ever encountered in more esoteric philosophical tracts. Yeah. Where or the writing of Timothy Morton who talks a lot about in object oriented ontology, how you can never actually know the whole thing. So I'd love to hear you talk about this issue, which is fundamental to both the science and the philosophy of this whole thing about how observers are inherently finite. Yeah. And they can't gather complete information about something.


Michael Garfield (16m 16s):

And so there is, we talk a lot about the, just the, the fact that, like you said a moment ago that yeah, these things are only pluralistic and therefore there isn't like a right way. There are only more or less functional ways Yeah. Of understanding. And I'd love to hear you talk a little bit more about how this works in practice and how you've unpacked this in your own Yep.


Carlos Gershenson (16m 36s):

Research. So think that that figure of the circle sphere helps illustrate this point. So imagine you have a sphere and it have white half black, but actually you can just perceive it one perspective. So some people will see a black circle, some people will see a white circle, some people will see half and half. And then we can fight holy wars trying to decide or convince each other that the color is, that the circle is white or black. But of course that will not change the sphere. And we cannot make an exercise of democratic mathematics and just take the average because it could be that the majority is perceiving the sphere from a particular perspective.


Carlos Gershenson (17m 16s):

So we can say that these different perspectives are contexts that different people have, or the same person can have different context in different conditions. So you cannot really say that the circle is really white or really black because you cannot really observe it from all perspectives. And real phenomena can have, let's say an infinite number of dimensions because you can always describe them from novel perspectives. So instead of trying to say, okay, it is this way or it's not that way, it can be more productive to say, okay, from this perspective it's this color. From that perspective, it is that color. And of course that doesn't mean that anything goes.


Carlos Gershenson (17m 58s):

And then all the colors are valid because from a specific perspective you can really check, okay, it's from this perspective it's white and we can all agree on that.


Michael Garfield (18m 7s):

So, so how does this affect the way that you actually conduct research? Like how does this, when the rubber hits the road, and you're talking about model selection? Yeah. Like we, we have this kind of an inside joke here at sfi. There were a few years where everyone wanted to see everything as a network. Yeah. Or everyone wanted to see everything as the outcome of a scaling law. And the longer that this science matures, the longer a trail of these different sort of preferred framings stretches out behind us. So how do you, in, in practice, in your work, how do you actually decide the correct approach for the appropriate context? Yeah,


Carlos Gershenson (18m 43s):

So I'm, I tend to be pragmatic and of course there are these discipline jokes of mathematicians making fun of physicists because they're not rigorous enough. And then physicists making fun of engineers because we are not rigorous enough. And of course we can make fun of, I know doctors or psychologists because they're not rigorous enough according to our standards. So since I have permission as an engineer, then in many cases I just try to make things work and afterwards look for explanations. And it might not be the most elegant thing, but in a way I could kind of guard myself behind Wienstein and say, okay, that's, let's be pragmatic about it because of course we can spend all efforts trying to justify something.


Carlos Gershenson (19m 28s):

But if it does what, what it should do, then I guess it's, it shouldn't be too bad, at least on the, in the right direction. Well,


Michael Garfield (19m 36s):

So this just cuz it, it's kind easy to go in circles around this kind of a question. Yeah. One of the things that comes up again and again in your work and related work is the way that as you, you talk about adaptation in an organization and adaptation requires some kind of balance between memory and forgetting. You need to be able to forget in order to adapt. Yep. So I was actually talking about this with Michael Lockman the other day. When you ask a question such as, well we want to tune a research organization like SFI to produce novelty. Yep. In order to do that you have to know which holes you've already dug in this kind of a terrain.


Michael Garfield (20m 16s):

And so the question of how do you even orient, say scientific research or technological innovation in an organization also requires the, it's a complexity economics question where it's like you have to know what you're actually measuring for or like what the, yeah. Like what you're actually tuning this thing to do.


Carlos Gershenson (20m 36s):

Also, you don't want to lose what you already have in sense that's okay, we want novelty, but we still want to keep the good things that are working and actually evolution has to solve the same, same problem. So it, this has been studied by Kaufman, by and Wagner and many others because how evolution works the genetic level at the cultural cultural level, the economic level is by exploring but organisms, in order to be viable, they need to keep on functioning. So there's again this balance between robustness and adaptability. That robustness basically keeps your functions as they should. Then adaptability allows you to explore new things and actually heterogeneity seems to favor both as well.


Carlos Gershenson (21m 17s):

That's how the idea of heterogeneity arose because we're studying rank dynamics with colleagues from the physics institute in Mexico. And it turns out that the most important elements of a wide variety of complex systems change slower than not so important elements. So let's say since I knew this literature from evolution, it made sense to me that you don't want your relevant elements to change. So that gives you a business, but then the least relevant elements have the freedom to explore without breaking everything down. So that heterogeneity also gives you the opportunity to explore and kind of possibly innovate in a sustainable way.


Carlos Gershenson (22m 4s):

Because if everything is changed, it's too bad and if nothing changes, that's also bad.


Michael Garfield (22m 9s):

So yeah. And that, but and then it's like we're, and then we're back at the question of, well in order to know which elements are important to conserve, yeah you have to be adaptable at a different time scale then. So that's to the sort of question about whether the economy as we have it now, is that paperclip machine that's optimized for the wrong thing, it's turning the whole world into paper clips. So you talk about this again, to go back to the paper with Halligan, you mentioned, and I think this is gonna be familiar to anyone listening, you say an entrenched culture in an organization can be very difficult to change as new measures are actively or passively resisted, ignored or deflected such as system destroys distinctions as distinct causes a lead to the same outcomes.


Michael Garfield (22m 49s):

That's a robust system, but it's a system that's not necessarily tuned to achieve the best results. And so how do you, based on this work, how do you imagine that organizations can better implement this understanding into accommodating adaptability without destabilizing themselves?


Carlos Gershenson (23m 11s):

Yeah, yeah. So Ashby, who was a super British Sian from middle of the 20th century, one of his contributions was the law of requisite variety. So this basically says that in terms of controls theory, that a controller should have at least the same variety as that which is trying to control. And by variety we can just think number of states. So imagine you have robot in a factory and you want the robot to manipulate six objects. So it should distinguish at least those six different objects in order to function. But then when you take that love requisite variety to organizations, to governments, we see that they don't have requisite variety to deal with economy, with societies and with many other things.


Carlos Gershenson (23m 56s):

So that kind of explains why they fail so often and it's not trivial to either increase the variety of disorganization or decrease the variety of that which they're trying to control. Which for our purposes we could use variety as a synonym of complexity. But then it's, it is like a thermometer that you can apply to organizations and say, okay, how many things we have to deal with And then we have the capacity to deal with all those things, yes or no, how many unexpected things we have to deal with every month, how well are we dealing with those? Does it break everything else that we had planned or not?


Carlos Gershenson (24m 37s):

So this sort of questions can help adjust the organization to an appropriate level of adaptability because basically the things that are constantly occurring and you can plan for them, then you don't need to change for them. Then there are other things that every now and then it's something new and then you need to dedicate time and effort to address those things. And of course you want to be able to do everything in parallel.


Michael Garfield (25m 5s):

Yeah. There are a lot of ways that we could move from there, but I want to bring it back to wind. Slower is faster just cuz this is Yep. This is a review full of work, some of which is your own, like the work on self-organizing traffic light control that I think gives people a lot of really tangible, real world examples of how systems either do or could accommodate this kind of flexibility in their design. So yeah, you gave a lot of these examples in the talk that you gave here that will link in the show notes. But yeah, let's talk about stuff like logistics and supply chains and transportation infrastructure and how this stuff is actually working in practice. Yeah,


Carlos Gershenson (25m 44s):

So, so traditionally most of these were solved with in algebra or some other formal approach. And there again, you basically try to predict or optimize and then you idolize the problem and in an abstract way you solve it and supposedly you're it, you are, you're done. The thing with this problems is that they're changing constantly. The technical term is non-stationary and it's basically that the problem itself is changing. So if you thought you had a solution, the moment you implement it's already obsolete and they are non-stationary precisely because of complexity. If you have lots of elements interacting and they generate new information, that that information change the problem, then your solution needs to adapt at the same time skills at which the problems change.


Carlos Gershenson (26m 31s):

So it seems that in many cases engineers or just that deal with these problems prefer to ignore these facts and then just try to say more or less cope with it it and leave some margins. Then when you take a different approach and instead of trying to predict something that you know will change and that you cannot really predict, you shift your approach to adaptation, then you can achieve much better performance in some cases optimal in some cases even beyond optimum. So that's this paper for public transportation systems about super optimality, which perhaps I didn't send it to you


Michael Garfield (27m 10s):

But No, but I did see it Yeah. On the Google scholar page. And then again, so it's funny cuz actually this was a conversation we were just having in the comms office about crisis control and when things cited Daniel Kahneman and when things require an an immediate response versus when they don't, and when I had the first time I had Rajiv Seti on the show and we were talking about the confusion of those categories in a world that moves extremely fast. Yeah. And so he was talking about the problem of police violence and the way that stereotypes sneak into these interactions between authorities and citizens when you don't have a lot of time to get to know somebody and you end up making a snap judgment and someone's life is on the line and it's, it seems those kinds of experiences or those kinds of situations where we're, we're thinking on the wrong time scale to address the issue, are proliferating in a society where the technological infrastructure is going faster and faster all the time.


Michael Garfield (28m 14s):

And so when you talk about, again, in, in this piece you're talking about how, and this really, this is one of those kind of paradoxes that's inherent to this that I love when you mention in talking about requisite variety in cybernetics, that the greater the variety of perturbations system may be subjected to the larger the variety of actions it needs to remain in control. Right. So there's kind of a fork here and I'm, you can take it either way you want or both or neither, right? But there's that famous preprint from, I can't remember who read it, the artificial intelligence piece on how optimal policies tend to seek power, right? And like the definition of intelligence is about navigating that kind of uncertainty and being able to make decisions across timescales and tuning for the appropriate level of variety.


Michael Garfield (29m 3s):

And then to me this sounds a lot like to just be a kind of an armchair guy about this. You're a man on the street. This sounds a lot like what you hear Buddhist teachers saying about the importance of mind training practices in the intensity and the pace of modern life that in a weird way there's a kind of identity between more control and more unattachment or like the, the ability to relax and allow things to percolate before making a decision. So this is me trying to edge into the other piece that you've done on Buddhist philosophy, but I'm, I'd love to hear you riff on this. Yes.


Carlos Gershenson (29m 39s):

So in general in a network, when you have more and more interactions, that creates more and more change that propagates through the network. So you end up with chaotic dynamics very quickly. And of course this can have its negative sides. Again there's this balance between order and chaos we could dive into. But of course we tend to notice the negative effects of increasing connectivity, of accelerating interactions. And in the stock market with electronic trading, this was very clear with flash crashes a decade ago and you can start seeing it everywhere. And of course in if you are in a Catholic regime, there's very little you can control because changing is constant.


Carlos Gershenson (30m 24s):

And of course we could try to apply masterful in activity but of course that will assume that things tends to their optimal state. And then we don't really need to intervene because it does happen that sometimes let's say there's situation and you respond to this, to the situation makes it worse. So it would be better just not to do anything and let things run its course. But in most cases that's not the case. And but of course what to do is an open question, especially in situations that are unraveled that we've never encountered before. How do you make decisions when there's no previous example and so many things that could go wrong or not necessarily wrong, but let's say in unexpected ways.


Carlos Gershenson (31m 7s):

And that has kind of a density of increasing and of course there's arguments for trying to slow down the change in order to make it more manageable.


Michael Garfield (31m 17s):

So this question of the balance yes. Between these two things. And I wanna go back just briefly to the frontiers piece. Heterogeneity extends criticality cuz you know, something that comes up a lot at SFI is the way that given the enormous breadth of the possible, what we find in complex systems is that they actually inhabit very narrow channels. Hmm. Of possibility. You talk about like the way that Chris Kemp and Jeff West have shown that all like biophysical scaling puts all of the possible forms of a tree within this sort of very narrow scope. And so yeah, I'd like, but just given that we could anchor this in a bit more of a sort of a rigorous formalization, I'd like to hear you talk a little bit more about criticality in light of all of the different ways that we see this manifesting in the world.


Michael Garfield (32m 9s):

And you give some really interesting examples here talking about arbitrary complexity. Maybe that's something that we can spend a little more time on.


Carlos Gershenson (32m 17s):

Yeah, well I would like to expand a bit on on criticality, which brought way of saying what criticalities, simply regime that's between order phase and co faced or between two phases. But in general we could describe the mass chaotic or order, but it could be also turbulent and laar flow in fluid dynamics or in traffic dynamics. It could be free flow phase where all cars are driving at the same speed or at a desired speed and jammed phase where you have to stop because of high density. So precisely this scaling phenomenon, power loss and many things have been found close to criticality and you can see criticality also as a balance and it's a desirable balance because then you get the benefits of both phases.


Carlos Gershenson (33m 9s):

Of course throughout evolution you we observe that many systems are poised near criticality and precisely because they benefits from finding this balance. And as you mentioned, this is like a very narrow region from all possible parameter spaces and they're would like make a a small bitter and then to turn by to return. Yeah, yeah. If we just look at chemical reactions from all possible chemical reactions, most of them will never take place in our universe. Just so many that there's no possibility that they will happen. And still we have such a rich chemistry that led to biology and the same all possible biological interactions will never take place.


Carlos Gershenson (33m 54s):

All possible social interactions will never take place, all possible ideas will never materialize. But still that minimum infinitesimal region of the space of the possible that actually gets instantiated even just once in the history of a universe needs to be viable in in some way even for a short moment. And of course that constraints which ones will survive or which ones will endure, which ones will evolve. So returning into criticality traditional models of complex systems were homogenous because simply it's convenient to treat all the elements of a model in in the same way like in acellular automaton.


Carlos Gershenson (34m 42s):

So you consider all with the same rule, all with the same time, all with the same initial states perhaps or just with random states and then you observe what happens. But then what we began seeing, of course you look at the real phenomena and you diversity different types of heterogeneity, but then when you include that heterogeneity in your molds, we realize that the properties that we usually found in criticality that was difficult to get because you need to fine tune the parameter to reach this phase transition. This same properties you would find for a broader region of the parameter space, the more criticality you add. And this applies for structural criticality, which basically the network topology that has been studied thoroughly in network science, but then also the temporal heterogeneity, different elements of data, different times.


Carlos Gershenson (35m 34s):

And it turns out that this was known since the sixties in physics they're called griffith's faces. But it's, it seems it's a well kept secret, at least nobody I know about them. And what we're finding is that also if you introduce hetero in the functions, that also has a similar effect. And not only that, but if you combine them then they have an additive effect. So basically when you include these heterogeneities, you make it easy for evolution or whatever process to have these properties that are desirable cuz of criticality. So then search becomes even easier.


Michael Garfield (36m 13s):

So the, there's a question that comes up on the show a lot and I love interrogating this from as many angles as possible. And we're here and we've already kind of teased it, which is, and I'm gonna talk about this with Jeff West and mounted laber, I have them on the show cuz Manfred's been writing about the enth anthropo scene. Does era of geological history on this planet defined by the predominance and geological record of human activity? Yes. This is an unprecedented time. Maybe you can look at the formation of a modern atmosphere 2 billion years ago as a kind of a precursor. Yeah. But you have this time where it's the network topology as you're talking about it is one where suddenly these ephemeral fleeting creatures on the surface of the planet are capable of adjusting the outcomes of so many different things over such a large timescale.


Michael Garfield (37m 6s):

And yet we're not actually, we're not actually adapting at Yeah those timescales, we're not actually, so you've got these, this weird situation where you can, you have a, like a multibillionaire who decides that they wanna launch a new industry and then you know they're gonna be then pulling up all our rare earth minerals and like laying out of tons of concrete and changing the atmosphere with rocket fuels and like all the, and so this is a very, this is related to the problem that I talked about with Dne farmer where DNEs work in market ecology showed that you mentioned flash crashes a moment ago and din showed that systems like Robinhood, which allow for fee free trading by retail investors lead to enormous market instability.


Michael Garfield (37m 49s):

And so on the one hand it's an opportunity for people and on the other hand it's like arguable that more people are getting wrecked by this system then are benefiting from it. And so yeah, I feel like I just keep going back to this, but like when it comes to when slower is faster, one of the big questions seems to come up a lot with respect to complexity economics is how do we slow down, right? Like how do we lower the temperature of this system enough that we can actually aggregate information at the time scale where we're capable of making decisions and yeah. And how do market incentives actually achieve this? How can we bring this closer in practice?


Michael Garfield (38m 31s):

And I'm curious what your thoughts are on all


Carlos Gershenson (38m 32s):

That. Yes, there two ways are the problem of let's say when slower is faster is that if you go beyond the face transition, then your performance decreases. So you want to maximize your performance and there are two ways of achieving that. One, of course it's slowing down, but another is changing the system so it can go as fast as you want to go, but then of course that increases the incentive of going even faster and then you have the same problem again. And also many of the issues you mentioned could be argued that it's unbalancing our, our planets, our economy, our way of life. But on the other hand it is true that it, these changes kind of drive systems out of the natural states or prefer states, but at the end, unstable configurations are not sustainable.


Carlos Gershenson (39m 18s):

So sooner or later they will create a new balance state, just like the great oxidation events transform the atmosphere actually created a niche for more complex and richer life. The question of course within our lifetimes is, okay, how much suffering can we prevent? How much damage can we avoid? And in many cases, even when we have enough information or enough tools to be able to say, okay, this will be the consequences of this trajectory, it's like we won't do anything until it's too late and even after it's too late probably we will just say, oh, too bad, it's too late then we cannot do anything.


Carlos Gershenson (40m 1s):

So with the pandemic, I was thinking, well maybe it'll have the added benefit that we'll be able to be better prepared to kind of face global challenges. But I think it showed that we are


Michael Garfield (40m 13s):

Incapable of


Carlos Gershenson (40m 14s):

Dealing with global challenges and if we were unable to coordinate internationally to deal with the pandemic, I don't think we have much chance with climate change. So we'll just take the hits as they come and yeah, try to solve issues as they appear and think we have the tools for global social coordination and decision making to make better decisions or well, yeah, we'll have better actions at this stage.


Michael Garfield (40m 40s):

So kind of a related question, given that you just gave a talk on anti-fragility Yeah. Is that this is the third regime that we haven't really spent a lot of time on in this conversation. Yeah. The notion that there are ways in which certain systems can actually benefit from perturb patients and and so maybe like I've seen a lot of conservation ecologists come around to this, and you mentioned this in the talk, you know that there are ways that rather than trying to preserve a kind of retro romantic Eden type landscape, that we can accept the fact that things are changing and then we can try and tune both our systems and the kind of perturbations that we hit them with. Yes. So as to create structures that actually become stronger Yes.


Michael Garfield (41m 23s):

Through through our meddling. And so I'd love to hear you bring in unpack antifragility for people and then Yes. Bring in, bring that piece into this conversation and


Carlos Gershenson (41m 32s):

Yes, so antifragility is a concept defined by more than 10 years ago in his book of the same name. And he asks, okay, what's the opposite of fragility? We kind of know what's fragility and people will tell, well, robustness. And he was like, well no, robustness is the lack of fragility. I'm not interested on things that let's say don't care whether there is perturbation or noise or not. How would you call things that benefit from noise that thrive with perturbations? And since he didn't find a better term, he coined antifragility. And there specific phenomena that we can call antifragile like stacastic resonances or, and many other examples that you can say, okay sis, let's say systems where noise improves the performance of the system.


Carlos Gershenson (42m 24s):

And however, to be able to design antifragile systems, you need to be able to predict or to know what will be the magnitude of the perturbations the systems will have to deal with. And yeah, that, that's tricky precisely because of computational reducibility. Because in, in many cases since complex systems are limited in their predictability, then we cannot answer a priority what would happen if we have this or that intervention. So we still have an ideology where we want to be sure of the decisions we make, what would be the consequence even if we were wrong, at least we have delusion that we wanted something to do.


Carlos Gershenson (43m 7s):

So it would be very difficult to convince anybody to say, Hey, let's try in this ecosystem to intervene with these substances or with these species and then we'll see whether it's good or bad. Most people say that doesn't sound like a good idea, but what's the record sole bio terraforming or biosphere? Yes. But not only, we don't have better way of doing this, but it seems that unless we get very sophisticated computer simulations and that will take a few decades, so far we don't have better alternatives. So in a way it is difficult to say, okay, we want the planet to be with all these properties and that's how we will get there.


Carlos Gershenson (43m 52s):

So I guess that we will have to satisfy ourselves with a more modest approach and just see, okay, we know recent history, we know where we are now we know possible trajectories and we have a very limited interventions we can have. And from those we we have to choose which ones viable. And from there we'll keep on learning how to better take our place in in the universe. But I think that all this already turning back into philosophy until recently or perhaps now science had a vision, okay, the more more we know we will be able to control nature and then we will be able to do whatever we want for our own purposes.


Carlos Gershenson (44m 39s):

And since we are slowly accepting that there are many things we cannot know a priority and that our control is limited, it's more productive to see ourself not as controllers or of nature, but as part of nature. And then it's, the question is not how can we transform nature for our purposes, but how can we better take our place in nature?


Michael Garfield (45m 2s):

So that brings us right around to where I wanted to take this, which is back into your philosophical writings. And so you mentioned again and how can we think the complex, you talk about how self-organization deals below to the dualism of classical thinking as it blurs the distinction between matter and mind. And you explain how from the cybernetic perspective, there's no strict boundary between material and mental components. You talk about the extended mind and kinda cyborg theory and when don't we add Caleb sharp on the show talking about the data own, these kinds of questions are really on a lot of people's minds now in a big way. Yes. With the fact you can't get away from conversations around artificial intelligence and what it means to entrust machine learning with tasks that we have conventionally understood to be the the exclusive province of human creativity.


Michael Garfield (45m 50s):

So from here, I want to go finally into this is piece that you wrote on the scale of selves information life and Buddhist philosophy. Because again there's, there is a kind of a sub current or an undercurrent of Buddhist thinking in the history of the complex systems domain. You go back to Varela and Marans piece on a calculus for self-reference back in the seventies, these kinds of things. And yet something that doesn't come up a lot in stuff. Anyway, I would love to hear you just drop this piece on us because there are a couple of really juicy interesting things you say in here that that I wanna address, but I wanna give you the chance to unpack it for us first. Yes


Carlos Gershenson (46m 29s):

Sir. Can say that western philosophy has been dominated by the success of physics in describing certain aspects of our world. But then that has led people including some from this institute to believe that let's say what reality is, what physics is basically modern energy and then let's say modern energy organize themselves and then you have biology and society and ideas and everything. But all of these are epiphenomenal. However you end up unable to explain, let's say what life is, what the mind is, what well, whatever that is in terms of physics. Because not only the properties of life, but simply we can say that living systems use information with meaning.


Carlos Gershenson (47m 13s):

And then you cannot define what the meaning of information or symbol will be from physics because it's arbitrary. However, we can agree that the meaning of information can have causal influence on maternal energy. So I would argue, and many people wouldn't agree with me, that there's this causal influence that cannot be reduced to modern energy. So one alternative, instead of describing the world in terms of modern energy is to describe it in terms of information. And you can describe modern energy as particular case of information. And there you can avoid this dualist trap where you cannot explain one from the other. And in, in a way, in Buddhist philosophy, you have a similar approach in sense that it is said that the distinction between the observer, the observed and the action of observing is an illusion.


Carlos Gershenson (48m 7s):

And what this means is that you cannot really speak of an observer without some physical world, but of course you cannot speak about a physical world without an observer to describe it. And of course the action of this observing process. So seeing this as part of the same thing can not divisible kind of solve, solve these problems of whether of objectivity or subjectivity are the best approach. It's just like you need both. And that's just one, one aspect of things that that could be said on the matter.


Michael Garfield (48m 40s):

So there's just a couple little pieces I want to quote out of here and then you can just riff on yourself here because I thought this was, it was interesting how you say, for instance, when we're getting back to where we started this conversation and talking about an inherent pluralism to the approach of modeling. Yes, the world you mentioned in scales, and we've talked about this on the show, the it's on a field like psychophysics works because if you back away far enough from people, then you can model people as though they are just molecules. But if you get people reject this, if you get in close enough, then you get in these systems where you don't want to make deterministic claims about the dignity of a free will and a human actor.


Michael Garfield (49m 24s):

And Sean Carroll is somebody who's written a lot on this as well. And the this question of finding one's level, what is the appropriate scale which to explore phenomenon you in connecting this to Buddhism, you make this the following claim that at the highest scale everything is included, therefore form is emptiness. While at the lowest scale everything is possible, emptiness is form. And then later as this pertains to the issue of sulfur at any scale except at the highest we will be leaving something out of a description. The highest scale is not always practical. We need to make distinctions to take distinctions. We can conclude at the best scales of descriptions ourselves will be those at which decisions are made, different decisions, different scales.


Michael Garfield (50m 6s):

If the decision is about our biosphere, we should forget about our individuality. But if we are hungry, we must focus on sustaining our decaying bodies. It's funny because this is so clearly instantiated in the biology of, I had the opportunity to talk to Penn State emeritus Gary Weber once, who meditated for 35 years consistently. And at some point he is no longer modeling a self in his experience of the world. He said the only time that a self appeared to him was when his blood sugar was dangerously low. And so there's, yeah. So it does seem as though, like you were saying earlier, that evolution has had enough time to find a way to create requisite temporary spatial heterogeneity within even a single person where sometimes you want a self and sometimes you don't.


Michael Garfield (50m 54s):

And it really just depends on the scale at which you have to be making your decisions. So that's the last little ball of yarn I wanted to put together for you and then just let you carry it home into, yeah, it's been a couple years since you've written this and Yep. Yeah, I'd love to hear your thoughts.


Carlos Gershenson (51m 7s):

So it reminds me of, from cosmic, he wrote that a model should be able to predict, well I'm kind of misquoting, but something like a model should predict the most with the list information or something like that. And again, predict the most about what it depends at which scale. So again, if you want lots of details, then your model will be inevitably more complex. If you want to look at things at a very high level, then you can get away with a very abstract model. And also in some cases it might seem that if your model is not predictive, it doesn't work. But we forget that modeling also can be useful just to understand phenomena.


Carlos Gershenson (51m 52s):

So the, there are many models that are not realistic at all, but you can understand the nature of phenomena much better throwing much of the detail. So the, this reminds me when we were exploring with some colleagues this synchroning traffic lights in a very abstract city grid model with solar automata where you had duality between Carson spaces, which and infinite acceleration. And the physics were like not completely unrealistic, but this simplification allowed those to very clearly identify 10 phase transitions that you have in this system. And if you add the realism, these phase transitions blend away. So you cannot identify them clearly.


Carlos Gershenson (52m 32s):

So again, like arguing for let's say a more inclusive perspective for studying phenomena in the sense that you will not find the best model, just like you won't find the best search algorithm because different search spaces have different shapes. So then there will be different algorithms that will be best for that. And the same for studying phenomena. You will be better off if you have multiple perspectives. And that's one of the benefits of multidisciplinary. And places like SFI were this is promoted of course in some cases it's not easier because difficult to communicate and different people with different languages. But I guess that's if we find better ways of facilitating this multidisciplinary interaction and we are more open to what other people think and, and not try to assert ourselves as the ultimate truth and everyone else is wrong, but okay, I might have some something to contribute and then maybe other people have something else to contribute.


Carlos Gershenson (53m 33s):

And if we agree that in a group, most probably will be reaching better decisions than fighting for dominating, let's say for finding a single best answer. Yeah, I think it'll be the, not that we will solve our all our problems, but at least we will be facing them with better alternatives. Yeah.


Michael Garfield (53m 57s):

So just kind of in closing, and then I wanna give you just an opportunity to show this book you're working on. This is, this comes up in term that you cited in that first paper meta representation, and I've also heard it called meta theory or meta methodological research. This is connected to a conversation that I heard David Krakauer talking about the work that he's done on national constitutions and how if we think about constitutions as a kind of regulatory network for the state, there are times when you want a more extensive prohibitive framework and there are times when you want to open it up and let it just as we were talking about with the child adult explore exploit tension.


Michael Garfield (54m 41s):

But we don't seem to have, generally as a species, we don't seem to have a very good grasp on how to do this. Yeah. Or like when you know, what kind of provisions we can put in place to know when it's important to close national borders versus when it's important to open them. And I'm just curious if you have links that you can point us to as far as research that's been done in this area by yourself or by others or real world instances where this is being done very well that that people listening might be able to model in the way that they think through the regulation of their own organizations or governance, yeah. Issues, et cetera.


Michael Garfield (55m 21s):



Carlos Gershenson (55m 22s):

Nothing we, we've begun exploring the good way of dealing with organizations or countries as you say, since the demands are so dynamic, we saw it with the pandemic that like almost all decisions were wrong because, Because okay, if we have two strict lockdowns, then people will get crazy, then we'll have revolves. But then if we let everything do whatever they want, we'll have so many million deaths. So you need to consider so many factors and at the end you will be wrong anyway. And people will hate you because of negative consequences not considering all the how much worse it could have been.


Carlos Gershenson (56m 10s):

And I guess that's, let's say our world becomes more complex precisely because of more interactions and more speed. We will face more and more this conditions where perhaps you won't even find like a Pareto front where let's say it's like the best combination of two variables, but then okay, now it's 20 variables and any possible solution will give you like very bad outcomes for most of them. So how to take good decisions in those contexts, it might be even overwhelming because you'll say, well anyway, most of the variables will be very bad, so what should we try to save or preserve?


Carlos Gershenson (56m 55s):

And you see it in many countries, no, in United States there's big high value given to democracy, but in many countries that they have violence, famine, even insecurity, they will say, well I don't care that much about democracy, let's just have stability. And then they will be willing to vote for dictator and they will be better off with let's say without certain freedoms, without free press, without many opportunities. But let's say if crime rates are not as high as they used to be, then they will gladly take that option.


Carlos Gershenson (57m 34s):

And I don't think we can say, oh, they're wrong. It's just like giving the conditions. Their options are so bad that the least worse possibility that of course can be criticizable, it's a bad choice, but all the other choices are even worse. So I don't know whether we are kind of heading towards more and more of those situations whether the speed of progress will start slowing down because of course tendencies tend to change and we have seen accelerating and accelerating change, but cars already invented, flight is already invented, the internet is already there. So of course there will be new technologies, but how transformative will they be?


Carlos Gershenson (58m 17s):

So it might be that change starts slow slowing down a bit and maybe we'll be able to, let's say, take advantage of what we have. I don't know. Because it's also sometimes we have a romantic view of the past in sense that we worry about all the problems that we have now in justice and so on. But it'll be difficult to pinpoint any point in our history where we were better on or we didn't have other problems that we are glad that we don't have anymore. And okay, now we complain about the problems we have now.


Michael Garfield (58m 48s):

Yeah, now we're kind of into that Moore's Law critique. Yeah. Now, but it does seem as the the late model smartphones compared to the phones two years ago. Yeah. Not all that different. And maybe that's like former SFI trustees Stuart brand said when progress happens fast enough, people call it change and they want it to stop. Yeah. So yeah, I mean it does seem like we're on the cusp of some sort of mass call for regulation on technological innovation, the rights around that so that people have a bit more of a grasp of what it's gonna look like in five years.


Carlos Gershenson (59m 21s):

Yeah. There are incentives for companies and for people to change as fast as possible to get an edge in the market and so on. But if at some point that instability brings more drawbacks than benefits, then very probably regulations should kick in and see how can we still have innovation. But as you say, we don't need to have new phones every year because anyway, they're not that different. And if some companies lose a few billion dollars, but you save the sanity of millions of people, I guess it's a with exchange


Michael Garfield (59m 56s):

Or put it back in sort of a Buddhist formalization. I have an old friend who decided to experiment by refusing to speak in the first person for weeks, just to see what kind of effect it would have on his consciousness. And he said that within a few days he stopped experiencing an egoic bound himself. But that after a few weeks, his wife got furious with him and begged him to start referring to himself in the first person again because it was interfering with their ability to communicate as partners. So there isn't always that question of like, how far out on a limb can you get before it's time to crawl back to the tree? Anyway, so yeah. Let's talk about your book that you're here on sabbatical and you're working on this thing. Yes.


Michael Garfield (1h 0m 36s):

And I guess you have to hand in a manuscript later this year maybe. So, yeah. So let's lead out with just a little bit of a teaser for the book that you're writing. Yes,


Carlos Gershenson (1h 0m 45s):

Yes. So the excuse for my sabbatical here at sfi, which enjoys greatly already halfway through, it's to write a book about balance, which is a narrative that kind of helps bring together many concepts related to complex systems for a general audience. So also the strategy that I follow is to give a talk on one chapter every five weeks. And that kind of keeps the template to advance at a certain pace on the book. It's not at, by the end of the sabbatical, I already have the full manuscript, but maybe at the end of 2023 I'll have a first draft. So yeah, it's been a great experience because it's much easier to pitch ideas to all the community here before they're written and then get feedback and then I notice what I'm missing or maybe what's an excess.


Carlos Gershenson (1h 1m 39s):

And then by the time I write, then I already advance some of that. And of course the writing also gets polished, but then it's kind of very helpful to, to concatenate ideas in this way.


Michael Garfield (1h 1m 52s):

Any parting thoughts or burning questions for you or places you wanna direct listeners before we sign off? Yeah,


Carlos Gershenson (1h 2m 2s):

Been very interesting conversation, like in some aspects, like we need to focus on our individuality, but also there's this cliche that when there's an emergency, like in the Titanic, all the civility goes overboard. And of course there are are many scenarios that we might think about. And then we'll say, okay, all the sociality and cooperation that we've achieved, it might be more fragile than we think. But then at the end, if the conditions are proper, I guess that it's not an exclusive choice in the sense that we have to decide whether it will be more individual or more cooperative and kind of become part of the machine, or in the sense that we can at the same time enjoy our personal lives, but basically avoid conflicts at the social level and international level, at the global level.


Carlos Gershenson (1h 2m 56s):

And of course, this has been studied extensively with games theory. All these dilemmas are precisely when the individual goals are not aligned with the group goals or the social goals. But in a way, this a problem I say of a badly designed game because if you, if you design the games properly, then everyone should strive for the best collective situation because then everyone benefits from that as well. So if we manage to design our social systems or our incentives in such a way that if the decisions I make at the same time maximize my benefit and maximize the benefit of society, then there's no dilemma. And then we're all happy. So some of our work has been in that direction, and of course, you cannot change people in sense that, okay, let's, I don't know, get rid of people from this state, and then we'll bring people from other states because they think differently.


Carlos Gershenson (1h 3m 48s):

And then that will solve all the problems you cannot change. So when intervention we did in the metro of Mexico City, it was like that you cannot change the passengers of Mexico City, but you want to change their behavior. What can you do? Well, you can change their interactions and like that maybe the system will perform better. And we manage to do that. And it's like an illustration of how could you change an economy where you won't change business people, but maybe you can change their interactions in order to have a better distribution of wealth? You will not change politicians, but maybe if you change their interactions, corruption will be reduced and maybe governance will be improved, will not change teachers, but maybe arranging interactions will improve education systems.


Carlos Gershenson (1h 4m 35s):

So I think that by understanding better complex systems, of course we will be able to deal better with complexity in our daily lives, in our organizations and international relations. And also focusing not only on, let's say, objective side of things, but on how they're related to each other. And precisely because in many cases, the possible interventions are at the interaction level, not at the object level.


Michael Garfield (1h 5m 2s):

That's funny. Just reflecting on this, I'm gonna have John Cog on the show at some point this spring to talk about the writing he's done on William James. And William James has that famous essay on some mental effects of the earthquake from when he was teaching at Stanford in 1906, during the big one in San Francisco. And how he said, to his surprise, after the earthquake, everyone came out of their buildings and was just spontaneously helping one another. Yes. So back to the question that you asked earlier, which is kind of how much suffering is necessary to improve things. I was like, maybe the structure of the environment was just too partitioned. Yeah. It was just, it was too easy to forget that you're part of a society with other people until you have to come out of the building safety.


Carlos Gershenson (1h 5m 44s):

Well, I don't know. Maybe for our civilization it'll be similar like with double A that you need to hit rock bottom before you actually take things seriously. But then of course the question is how deep is that rock bottom? Because it seems we're just going deeper and nothing changes.


Michael Garfield (1h 5m 59s):

Yeah. Well, Carlos, it's always a pleasure to talk to you and thank you for the work that you've done. Not only as a researcher, but as a a synthesis and reviewer and communicator of this stuff. I always appreciate hearing you riff on these things. Yep. Thank you for listening. Complexity is produced by the Santa Fe Institute, a nonprofit hub for complex system science, located in the high desert of New Mexico. For more information, including transcripts, research links, and educational resources, or to support our science and communication efforts, visit Santa