Complex systems remember their past. Conversation with Prof Alicia Juarrero - part 1

 

Transcript

Marc:

Alicia, our listeners would probably want to know how does one become a complexity philosopher?

Alicia:

I'm not sure how that happened, but I'll tell you what my problem was. I had completed a dissertation in ethics and the dissertation in ethics, which I have never looked at again, because it was a pretty bad dissertation, but it was about the difference between explaining behavior and justifying behavior. So moral justification is you have to give moral reasons for that would show the behavior to be right or wrong, good or bad, but explaining behavior is how did the behavior come about? And the two go together, because as I then started playing with the phrase, what is the difference between a wink and a blink? If you just see an event, someone doing something you really, and especially back then, you can't tell if it's a wink or a blink because the actual physical movements of the eye muscles are very similar. But on the, on the one hand, if it's an intentional behavior, like a wink, you are morally responsible and legally liable for intentional behavior. But if it's just a blink, just a reflex reaction that somebody flashed the torch, a light, a flashlight on your face, then you have no moral responsibility. So you have to know what happened before you even can get to the, to the question of moral justification. And so the first chapter was on, how do you explain what happened? And I really was never at all pleased with that first chapter. And that's why I think it's a bad dissertation. And part of the problem is if behavior's intentional, then that means that supposedly that it was caused by an intention. Well, how on earth does an inten, what is an intention and two, how does it cause behavior? And the standard notion of science and philosophy has been since the scientific revolution in the 16th, 17th century, is that causes your billiard ball like one impact, transfers energy to a, a recipient, an object, like a tic transfers, energy to a ball. And that's what causes are. Well, obviously, even back when I wrote the dissertation in the middle ages, I, it wasn't a form. There is no one neuron that pushes another neuron anywhere. So clearly that was wrong. I already had the doctorate. I was teaching here in DC. And the nice thing is that, the national endowment, the national Institute of health would let you come listen to, to speeches. And so I'd go there and I'd listen to ideas about networks in the neural and the actual, so wetware in the brain, you know, and I'd heard about systems, systems, engineer, systems, science, some science. And so the combination it's got to be somehow an intention must be some kind of network in the brain, not one impactful energetic transfer, but a network dynamic phenomenon. And then the question was, well, how does that happen? Because clearly the network arose from the individual neurons, but then it seems to affect the individual neurons. So you've got this strange loopy, circular causality. That was a huge no-no. And still often is a huge no-no in so far as the theoretical understanding of causes. And so slowly by looking at systems thinking took me to [...], And then Prigogine came out with the notion of dissipated structures. And I happened to be reading Kant and Prigogine at the same time, because Kant has an idea that in order to explain this, he says in one of his Critiques of Practical Reason, you have to have a form of causality that Kant says–in translation–"is unknown to us". And when I read how individual particles of a Bénard cell or a convection cell align to form a system that then loops back down and affects the individual water molecules in the Bénard cell, I thought Prigogine got the answer to Kant's puzzle, and that is complexity theory. And so then from there on in, I never looked back. So that's a long way around as to how I got involved in this. And I think whenever you have phenomena that have properties that are the result of collective phenomena, but coherent collective phenomena, then you have complexity and, and there's no way to reduce it, I think. You can manage, you can understand, but you can't reduce it. And so that's how I got involved in this.

Kristof:

And, and you are also on the intersection between different parts of, or different areas where complexity applies. You're in, in biology with your business, you're in philosophy, you're, you're in well in technology also with your business. So was that from the beginning that you were already on that intersection?

Alicia:

No,I was interested in tech. I'm from Cuba originally, and I wanted to expose Cuban [ and Cuban philosophers to complexity theory because they did not really know about complexity theory. And so there were a number of people in Cuba who knew the work of Pri-go-gine they kept telling me it's not Prigo-gi-ne it's Pri-go-gine. Oh, okay. and so I think that became an interest and many of them were involved in public health in Cuba. And so that was the connection with public health and with technology.

Kristof:

How does looking at the world through complexity change things for you?

Alicia:

I think that it makes you really take context seriously. Because that molecule molecule of water in a Bénard cell acts differently, behave differently often has different properties and other phenomena like, footfalls of soldiers, marching have all sorts of different properties if they are part of an entrained synchronized march, than if they are marching separately on their own. So you have to look at the context in which the entity is embedded, whereas in Newtonian mechanics and in sort of what I call science 1.0, space and time are like containers. They're like featureless containers that entities get plugged into: they don't change because of their embeddedness in a context. And I think that's why biology was the science that first refused to fall in line with mechanics, because biologists understand eco ecosystemic characteristics. And they refuse to accept that, that those systemic properties could be reduced away. And so I think that's context changes a lot.

Kristof:

But even now... I studied genetic engineering and we're still trying to reduce everything to genes and, and whatever genes are doing. I think it's more the frustration that it just doesn't work this way. So there's something missing.

Alicia:

But I think people are starting to realize that because with epigenetics, people are realizing that the activation of a gene or the suppression of a gene changes dramatically, depending on context. And when I say context, I don't just mean spatial context or even cultural context. I mean, temporal context as well. So what happened in the past, in that organism will affect whether that gene gets today or not, or whether that effect of that expression gets transmitted through several generations, which can happen. The Dutch winter example is as good an example as, as any of that problem. And so we are realizing how time is also a contextual feature because these–unlike Newtonian mechanics which are Markovian, they forget their past–complex systems are characteristicly path-dependent, so they don't forget their past. I was lucky enough to meet Prigogine several times. and he used to use a phrase and I've never, I don't think he ever published it in a written form, but he used to say 'Complex systems carry their history on their backs.' And the example he would give were snowflakes, the actual shape of the snowflakes reflect or embody or enact, or I don't know what the word is, the conditions under which that snowflake was created. Today that snowflake reflects or carries with it, the, the path in which it was created and through which it has traversed. So that I think that's very important. And certainly for human socio-technical complex systems that is really important. We tend to figure, 'Oh, we can start from scratch.' 'No, you know what? You can usually not start from scratch.'

Kristof:

Yes, exactly.

Marc:

Can you quickly tell me what the Dutch winter example is because I'm not familiar with it?

Alicia:

It was December '44, right. Or December or winter '44 to '45 or winter '45 to '46.

Kristof:

I know about the flooding and I thought it was the flooding that cause it, but I'm not sure.

Alicia:

I think it was, it was still remnants of Nazi occupation I think that really created–there might have been environmental, floods and droughts or whateve–that created a horrendous famine, for people in the Netherlands at the time. And it lasted for quite a while. There was a study done of women who were pregnant at the time and they followed the pregnancy to birth. They tracked the lives of those offspring, but then they tracked to the following generation and the genetic expression, for example, of obesity changed dramatically as if the offspring two generations down knew that it had to acquire and store as many calories as possible to, stave off the effects of starvation. And so that was the first clear evidence of what Lamarck would've called the inheritance of acquired characteristics, but it wasn't acquired characteristic. It was the fact that the genes didn't activate or they didn't turn off, or they didn't turn on as the case may be, but that was triggered by an environmental condition of an ancestor, for example.

Marc:

So that's experience of individuals reverberating down couple generations.

Alicia:

The contextual changes in the genes' methylation or whatever it was, persisted more than just one generation. I mean, you could say the first generation while the woman was pregnant with that child. So that's where the effect came from, but not the next generation.

Kristof:

And they've seen similar things also with I think with horses or, or it might mice also that depending on the food that's available, really big, big changes start happening, even before there's traditional genetic changes.

Alicia:

Right. I've always thought, for example, here in the United States, where there is so much gun violence in the inner city which we know affects the triggering effects of cortisol and so on and so forth, to what extent is that a form of Dutch winter? It's a violence-based form that must get transmitted more than one generation. I would think it's a... because for a person to live under horrible, strict, I mean, scary conditions for years and years, and years and years when you're growing up that has got to somehow affect things more than just now or tomorrow. It has to. So that that's a very scary contemporary version.

Kristof:

But it it's often in families, that's violence carries over from generation to generation.

Alicia:

We're calling it nature versus nurture, but it really isn't quite nurture. It's sort of a nurture that goes through a genetic activation. It's not that the gene is there. It's not that gene mutates it's that it turns on and off that. See, and that's another thing that is context, timing, the timing of these genes and how they are expressed or suppressed makes a huge difference in the outcome. And so if you thinker with the timing and if the switch stays on or the switch stays off, that can have incredible consequences. That can happen in a business, in any kind of a cultural situation where, where you, where the stories of reprisals by bosses can outlive the presence of those bosses and, and force people to self-censor or self, you know, it affects the behavior of the individual in that team, in that culture.

Marc:

I often also thought about, so in company speak, we say 'red tape', being the remnants of past experience of the organization and the difficulty often, that we're experiencing with red tape is that we do not understand the context in which it was issued as in it's like a scar, but we can't remember what caused the injury. And so governance people, for example, they might be very sympathetic when we come to them and say, 'Hey, this is really hindering us,' but because they can't remember what the pain was that this is supposed to mitigate, they're very reluctant to remove it.

Alicia:

But it might be that the wound started, was closed by butterfly-clips or whatever they're called. And that was designed to keep the wound and the two sides of the wound straight. So it need not have been noxious to begin with. It might have been a conservation, a stabilization process that then scarred over if you will. And what would a scar be? A scar would be that it didn't continue to adapt to the context. It became sclerotic. Right? And so then now we call it red tape. When it originally was probably a very useful attempt to keep the system stable and preserve it.

Marc:

Yeah. That's always my assumption because we can see when new rules or new restrictions are created, we know they can explain to us why and that makes sense. assuming that the creator explained it sufficiently, But it's, it's old red tape. So that's often, especially in larger organizations, also older organizations where you come across a lot of red tape that nobody can explain anymore. They do not know, 'So why did we put this in place?' But because it's not known, you would say, 'Okay, nobody could tell me why I have this. Why don't we remove it?' But there is a reluctance because also the experience of removal of red tape without knowing the cause often then, 'Ah, now it triggers again, the problem, that it actually was mitigating. We just weren't aware of the problem until it reoccurred.'

Kristof:

Under the influence of your work have started calling those 'debts constraints,' constraints that are like sitting in the system and there are no longer performing the work they were supposed to perform. I'm not sure if that's, semantically, but that's how I started thinking about those.

Alicia:

I think "restrictive constraints" are the, I think standard understanding and connotation of the term constraint. Because of what I had mentioned earlier about the mess with billiard ball type causality, I was looking for a term that would give me a cause-effect relationship, but that I wouldn't call it cause because as soon as it didn't have the energetic transfer, people would say, 'It's not a cause at all.' And so Howard Pattee, who is a very well known, biologist in the seventies, he is still alive. He one of the first to use the term "enabling constraints". That is conditions that facilitate the emergence or the appearance of those complex coherences that I was mentioning earlier, like synchrony or like entrainment, or like coordination dynamics, such as a bunch of pendulums sitting on a common shelf will end up in training and become synchronized in time, but nobody pushed them in that sense. The shelf serves as an enabling constraint, like a common ground in terms of which the dynamics of the individuals organize and become aligned. To the best of my knowledge there's something similar going on with lasers, lasers are nothing but fotons, but they are more than individual photon streams once they align. It is what I would call an emergent coherent structure that has emergent properties. Like a laser can cauterize a wound, whereas individual foton streams can't. The entrained footfalls of soldiers, marching over a bridge will collapse the bridge, which, and it isn't the number of footfalls because if they break step, the same number can go across the bridge, but it will not collapse the bridge or the stadium. you know, when, when the fans begin stomping in synchrony. It has qualitative effects that separately the individuals do not.

Kristof:

You're mentioning a couple of words that I've very busy with lately and what we've had some discussions already with Marc about previously. I'm curious to hear what you think, so "alignment" is a term that's used a lot in business, right? Creating alignments, in an organization that's organized through power structurres, probably that is one of the key ways to create coherence. But do you see–well, it's kind of a leading question, but–what is according to you, the difference between coherence and alignment, is there other types of coherence than alignment?

Alicia:

Yes. I think that, let me start by saying what I think it is not. And that is, I think that coherence is not a solid monolithic structure. I think coherence allows for individual variations under a set of more generic governing constraints, strengths. And I think that sometimes you need all the individual particles to be absolutely aligned in the same direction. I think that's probably true with the laser, but, Marc, the person I've mentioned to you before Peter M. Allen, who was at Cranfield, he has a beautiful example of, schools of fish, and he had a simulation where he called one of them Cartesians and other ones stochasts . And the Cartesian schools fish all followed exactly in the same direction. Well, what happens oftentimes is that they deplete those fisheries. Whereas if you have stochastic schools of fishes, where there is room to maneuver. So I don't know if the word alignment in business means absolutely marching in sync, but if it does, that's not what complexity is about because especially natural complex adaptive systems require enough flexibility on the part of the realizers, the individuals that are part of that overall alignment, to explore the state space and perhaps find new resources. Too much alignment will deplete the resources probably. It might work very nicely for the short run, but you don't want that for the long run, because then one thing we know again about biology is that if a species is absolutely fit for that niche, it will deplete that niche very quickly. And it will have no, if it has a perfect one-to-one match between its characteristics and the resources of the niche, if that resource starts going away, then there's nothing for this to do. But if this one has more of a flexibility, that's why my friend Dan Brooks and his colleague Sal Augusta talk about "sloppy fitness". So talk about sloppy alignment, it's aligned enough for purposes of the organization to achieve its goal or its function, but it is sloppy enough to perhaps discover a new path that might make, that might achieve that function better. So I don't like the rigidity of a monolithic block and I don't think nature does that at all, ever. So the word fitness, which would go with alignment, apparently Darwin used it in the sense of when you talk about going to a tailor to be fit for a new suit. So it's, you know, they're gonna take it in here and they're gonna adjust it there. And so that's what happens between the individual organism and the niche. So in business, that would be ideal as to having that kind of fitness function, which is not, I think we think of fitness nowadays in more of a sports metaphor that you are perfect for this game, and that's not the way Darwin thought of it at all.

Kristof:

I have a hypothesis that, the companies that are gonna be most successful tomorrow in this faster changing world, they'll have the sloppy coherence, instead of a very strict alignment and that this hyperefficiency is not going to keep working.

Alicia:

Efficiency I've learned is very different from effectiveness. And so what you're looking more for is more effective, but of course, if you have to worry about the stock price next quarter, then unfortunately efficiency often trumps effectiveness. And so it's very difficult to explain to somebody that, 'Look, this is not gonna work tomorrow,' when they're worrying about today. That's part of the problem.

Marc:

In business, we do actually now find a range. There are even concepts that we've came up with how we could create–so let's use the complexity word–coherence through alignment. Then even these concepts get actually debated in how do they need to be influenced? So, objective and key results is a current popular theme on that. In short, called OKRs, but there's a school of thought where objectives and key results have to be sort of chained across the organization. So you have your V-exec. And so she articulates objectives for the whole organization or your executive board does that. And then so their key results become the objectives of their next in line managerial level. And then their key results become the objectives one level down, et cetera. So you have this whole chain. And so I believe I might be wrong about this, but I think this is the Intel model or Google model. But there's somebody call Christina Vodtke, she comes from the other side. She originally started at HP. So she also has large corporation experience of this, but advocates for the opposite. And that has especially proven–in her experience–successful with startups and product focused organizations where the objectives might still be somewhat connected, but more through a meaning sense, and each level needs to figure out and learn how to articulate their own. So what would be the results for us to achieve these objectives? So there's no chaining, which also ties in for some... Drucker found that when he tried to do management by objectives, that he said, the most difficult thing is actually articulating good objectives, which is why he abandoned it because he says the average manager, it takes him two to three years to actually get there, by which time they get a new role so they have to relearn what good objectives are. So sloppy adherence would actually, so from a complex, could we then say from a complexity informed point of view, this chaining is what makes you quickly align, but then too rigid to react to any movement of your environment, but also make you miss opportunities that might already exist because there's no face space to explore.

Alicia:

What have you added zooming in and out to chaining to the chaining. So that, to me, if you zoom out, you see a wider context and I would assume the executive board's job would be to see that wider context, correct? And then zooming in to the more concrete realization of those conditions would be the next level down. I don't think the job of the board would be to, to even keep track of the change so much as to see what the overall picture looks like. And either expand that adaptive space, a la David Woods. Kristof, I hope Marc has talked about David Woods's paper called "Graceful extensibility", right? Where he talks about adaptive space and how you must either move away from a threshold of instability if you think you're reaching the limits of your adaptability, or you must expand that space in some way.

Alicia:

I use in the next book an example that I really liked. In the Mid-West United States are huge prairies with grasses and the grasses compete for nutrients, with flowering plants and horses and predators like horses. Now, if you allow me the anthropomorphism, the grasses figured out that a lot more energetically expensive for them to compete against the flowering plants. It's easier for them to compete against the horses, but what have they done? These grasses put out what are called Mary stems, which are sort of tendrils that go out in a sense, those tendrils are easier for horses to eat, 'cause they're very, tender tendrils. I wonder where that word came from. And so in a sense, by doing that, the grasses have invited the horses to come in and thereby co-op, preempted the competition with the flowering plants. In a sense, what that is doing is expanding the ecosystem. It is no longer the grass ecosystem period, competing against foreigners like the flowering plants and the horses. But rather now it's expanded into a grass-horse, more encompassing ecosystem. And I really like that example because in a sense the original ecosystem extruded into what was initially foreign territory to create a more... Not to destroy but to create a more encompassing ecosystem or to put it differently, it allowed the outside to become part of a new, more expanded inside. I really like that example. And there's a similar one with chaparral. Chaparral is an oily plant that encourages fires. And you would think that in the very dry area of the South-West in the United States, you'd try to get rid of the chaparral. But on the contrary, bringing in chaparral encourages small little brush fires, but what that does is it keeps the underbrush clean. And so you won't have a horrible conflagration, like these horrible forest fires we've been seeing lately. And so in a sense, it's brought in the chapparal into a more expanded ecosystem. I like those examples because it seems to me, they are metaphorical for what could be a lot of human behavior. Instead of seeing it as a competition, it sees it as a co-opting potential, if you will, and enlarging the ecosystem...

Kristof:

It brings the horses in coherence with the grasses.

Alicia:

Exactly. Exactly. And in order to do that, you have to have enabling constraints if I can sneak it in, that include catalysts, feedback loops, iteration, recursion. All of these are non forceful causes that enable the formation of a new emergent coherent whole. I think so. I think that's the answer.

Kristof:

And that's a lot more fun than competing, at least for me.

Alicia:

Well, it's probably a lot more effective. Yeah. It's probably in the long run, it's probably a lot more effective.

Marc:

Sounds like it changes the nature of the competition. Correct. But to use an example I had from you from earlier is, is also interesting in this regard, because we talk about in this podcast, so we want to talk or reference to APIs, which is sort of exactly a question of, so how outsides and insides. You pointed out, at one point, so if you have a pray-predator relationship, so you have gazelles and lions, over time they make each other faster. So the fact that they've sort of become co-dependent and that their evolution becomes intertwined actually makes them a system in themselves that then within the wider environment, gave them some advantages.

Alicia:

Correct.

Marc:

So what did you said about the grass and the horses is a similar thing. So in a sense, yes, they were competing or they might still be competing for some resource, but they're also, the competition becomes more of a collaborative nature. Right?

Alicia:

Correct, correct. And I think that is in the nature of the problem with boundaries and complex systems in. In Newtonian systems, you have nice clear boundaries, right? They are kind of walls. And that fits in very nicely with Aristotelian logical categories, right? They're very clearly bounded. Especially in complex adaptive system because of these feedback relationships, they are more like membranes. The boundaries are porous. They're more like membranes whose behavior is governed by a set of rules or algorithms. Instead of having walls, you've got passports, right? Or visas. I mean, that, that's a big change, correct? Instead of putting a moat in front of a medieval pass and a wall, you start using passports. But what that means is you've got rules that regulate the functioning of those membranes or those boundaries. And so I think you can think of what happened with the grasses and the horses as a widening of the aperture of those pores, widening, broadening, widening the lenses, enlarging the pores, so that more stuff comes in and gets incorporated into the dynamics inside.

Alicia:

Look at the problem of the brain drain there. The paper today was full of the number of Ukrainian IT folks that are going to have probably left, or I'm sorry, Russian IT folks that are probably gonna have left for good. You know, obviously that was the purpose of the Berlin wall, but they are still using walls. Clearly now with apps and so on, a physical wall won't do. And so the question is, how do you organize the principles, regulations, rules, whereby insides and outsides are negotiated, because that's what they are, they're negotiated, right? It's like an eardrum. And I got that from Paul Cillier, who's a very good complexity philosopher from South Africa. And he said, 'You know, complex boundaries are more like an eardrum, a membrane that is an active site that on the basis of the overall context allows more sound, broader sound waves. And that's what homeostasis is all about. The human biological capacity to keep everything in balance. You know, you eat an incredible meal today, the overall governing constraints of the system adjust the insulin level and the metabolic rate and so on to handle that extra load, or if you are starving, it handles it in that sense as well. So those are flexible regulatory constraints, which are governing constraints that adjust what's coming in, and what's going out in order to maintain the dynamic equilibrium that your engineering teacher was worried about.

Kristof:

I've been thinking about this, like resonance chambers, like if you have a guitar body, you have the snares where you're playing the music, but the sound waves are amplified inside of the residence chamber, but there's the hole where the snares and the resonance chamber are communicating with each other. And it's the fact that there is this chamber, this boundary, where a very different regime that can be formed, that actually creates the music that makes it amplifies it. And something similar that also happens with APIs and with ecosystems and platforms and all of this new stuff that's evolving in the it world. So this thing that you described about co-opting others into your game and like expanding your game.

Alicia:

Nasty...

Kristof:

Yeah. Well, that's kind of like what's going on right now, correct?

Alicia:

Correct. Marc taught me years ago at a conference at a workshop in Wales about object oriented programming. And I was fascinated by that because to me that is to think in terms of a class of variables that... am I correct in saying that's a form of sloppy programming? 'Cause what you're doing is you are looking in terms of classes of variables and functions rather than individuals. I don't know how to explain that any better.

Kristof:

I think it's similar. In the API world, in the software world, there's... I got this explained to me by the head of APIs at TUI. He was doing a presentation and he said there's a really clear distinction between how we used to do integrations in enterprise software, which was about integration. It was like everything was connected together and you were building exactly what you needed specifically for bringing all these pieces together. And the way that we're doing APIs today, which is, we have a constraint interface that hides what's behind it. And because you're hiding, because you're constraining what is going through this aperture, between these themes and the rest of the organization, you are making it harder for people to fail. So a developer, who's using an API, now they don't need to know this whole context of whatever you have on your end. They can just look at the interface and say, 'Well, you know, I can do something with this.' And they can do innoative things, things that you did not even imagine that are possible.

Alicia:

Yeah. And so we realize that boundaries are, or active sites at the boundaries are important enabling or not enabling constraints. They can facilitate this emergence of a more complex dynamic or they can boil it if you will.

Kristof:

Well, it's biology all over being repeated in software? Same with mycorrhizal network that are possible because of active sites in tree roots that know, 'If you present me this, then I'll accept the connection with you.' Or the way that sugars are being exchanged.

Alicia:

For so many centuries physics was the model to aspire to. And I think now biology is starting to give physics a big run for its money.

 

End of part 1 of the episode we recorded with Professor Alicia Juarrero. We invite you to tune in to part 2.