Monday, July 24, 2006

Attributes of Complex Adaptive Systems (Part II)


The Simply Complex Law
By J.B. Ruhl
Post 4: Attributes of Complex Adaptive Systems (Part II)
July 24, 2006

The previous post introduced terms describing internal attributes of complex adaptive systems (CAS). Now we turn to terms describing co-evolution between CASs.

Co-evolution: No CAS is a closed system, thus any CAS has a set of other CASs with which it co-evolves. CASs can co-evolve horizontally on the same macroscopic scale, or vertically in nested hierarchies. Except in computer-based CAS models, defining the boundaries of a CAS thus is an artificial conception, as it may be that a seemingly small link between two CASs provides the information flow that sets of large effects in one or both, though it may be difficult to trace that causal property by studying either CAS.

Radical Openness: At some point a CAS may become so open to effects from another CAS that they essentially merge into a new CAS.

Fitness Landscape: Taking all of the attributes discussed in this and the previous post into account, each CAS has a fitness landscape, each point of which represents the fitness level associated with a particular configuration of its agents, rules of interactions, and so on. The point of adapting is to maintain high fitness. If the fitness landscape is rugged, attributes such as feedback, emergence, and path dependence will be pronounced, and it could matter significantly what the agents do on each move. Also, the CAS might find when it reaches a particular fitness “peak” that there are higher peaks nearby, but getting to them through incremental evolution is a poor strategy as it involves walking downhill before walking uphill. The CAS may need to cross a critical threshold in order to “jump” to the fitness slopes of higher peaks. All of this is complicated by the “Red Queen” effect—because a CAS co-evolves with other CASs on their own fitness quests, the former’s fitness landscape is constantly shifting below its feet. A CAS may need to run just to stay in place.

Next: The basic framework for thinking of law’s domain—its subject matter and the legal system itself—through the lens of complexity theory.

11 Comments:

Anonymous jecado said...

Still new to thinking about this, and I am appreciating your posts, and the blog generally.

This sentence in part 4 threw me off: "The point of adapting is to maintain high fitness." This seems a little too end-directed. Isn't it just that a "highly fit" CAS as calculated by whatever you use to define "fitness" is considered "well-adapted?" Maybe we need definitions for "fitness" and "adaptation," too.

It's interesting that in contrast to the terms in Part 3, which seem primarily internal to a given CAS, the terms fitness and adaptation might be seen as having meaning at least in part to references outside of the CAS itself.

7/24/2006 3:30 PM  
Anonymous Frank Pasquale said...

Could we say that nations are CAS's? And that they compete over time for dominance?

I'm interested in this as an aspect of evolutionary economics, based on a presentation by Robert Frank at the HBES
(see
http://www.psych.upenn.edu/~kurzban/HBES2006/

He suggested that certain traits of individuals might be evolutionarily advantageous to their bearers, but set in motion an evolution disastrous to the group. For example, imagine peacocks with larger and larger tail feathers...birds with such features would do better reproductively, but the species as a whole suffers as they become easier for predators to catch.

Frank has a sense that "high spenders" in America, who fail to save, may be our "peacocks:" successful within our society but setting a pattern of norms that will ultimately lead us to lose global influence.

7/24/2006 3:51 PM  
Anonymous Anonymous said...

Of course, the point of adapting is NOT to maintain high fitness. As any evolutionary biologist or memeticist knows, high fitness is a secondary value, in support of reproduction. Reproduction of the genes or memes, if you believe Richard Dawkins. Or of the organism itself, if you go with the classic Darwinian explanation. Complexity and emergence translates this basic feature of the individual agent--whether gene, meme or organism--into fitness landscapes at the level of the collective.

7/24/2006 4:35 PM  
Blogger J.B. Ruhl said...

These are good comments. Thanks for them.

Perhaps I am conflating too much into the sentence in question and getting ahead of myself. True enough, a CAS has no normative objective or end state in mind. If adaptation leads to higher fitness, that's a "good" thing only in the sense that the system is more likely to exhibit resilience to perturbations, etc.

We also need to keep track of where in a nested hierarchy of CASs we are measuring fitness. Dawkins’ gene-centric view of evolution, for example, contends (as I understand it) that natural selection works at the genetic level to promote inclusive fitness of individuals in the species, toward the end of improving the success of the species in passing on genes, and that selection at the level of individuals or populations thus will not override selection on genes. Natural selection therefore favors genes that code for behaviors that lend themselves to increased fitness. Parents, for example, will exhibit altruistic caregiver behavior toward their offspring even though it does not always maximize the parents’ well being.

The comment from "anonymous" says "high fitness is a secondary value (of adaptation), in support of reproduction." Well, what is sexual reproduction doing for the species? It's facilitating adaptation: it increases the storehouse of genetic variation; it allows for sources of variation besides mutation; and thus it allows for adaptation to changing environments. Nice example of a CAS. But not all CASs must use sexual reproduction to facilitate adaptation. Sexual reproduction is just an example of a CAS.

Where I'm headed with this, of course, is that in social CAS settings, such as law, human actors have normative goals and we get to design our social systems. And one goal of how we design the legal system may be, and ought to be I would think, to improve its fitness. So I think it is fair to say in that setting--i.e., when we have the luxury of trying to design a CAS (something genes, birds, and hurricanes can't do)--maintaining high fitness ought to be one objective of building adaptive mechanisms into the system. None of this would be even slightly interesting if it weren't for the fact that we get to design the legal system.

Bob Frank's observations are quite important, as they illustrate both the conflicting constraints problem and the Red Queen problem.

And yes, I think nations are CASs. There is also a tremendous amount of literature on complexity theory and economics, and on evolutionary economics in general. I'll offer some links that lead there in the next post.

JBR

7/24/2006 5:44 PM  
Anonymous Anonymous said...

JB Ruhl's comment seems to demonstrate, again, a confusion about whether adaptation or fitness are ends in and of themselves, or secondary values supportive of reproduction. He is right that sexual reproduction contributes to increased opportunity for adaptation, relative to other forms of reproduction like asexual reproduction. But one should not take from that the notion that reproduction is designed to serve adaptation. Again, adaptation serves the end-state of reproduction, not the other way round.

The distinction is important because, once we introduce the notion of "designing" systems toward some end state, some might say the Lamarckian model of evolution, then we have to specify what we are designing towards. Ruhl says fitness. Memeticists like Dawkins would say that "we" become the vehicles through which ideas reproduce themselves, even though we think we are selecting for fitness. Besides, fitness must be measured in some way, and traditionally, in evolutionary theory, fitness is measured by ability to reproduce (check out Axelrod's work on cultural transmission, for example).

7/24/2006 8:30 PM  
Blogger J.B. Ruhl said...

It appears Anonymous and I each believes the other is confused.

Anonymous seems to be arguing that the point of adaptation is to support reproduction, and I argue the reverse. Sexual reproduction is an adpatation strategy.

I've had this discussion with evolutionary biologists and "memeticists" before, and can't get from them a clear answer about why they think reproduction is "supported by" adaptation and not the reverse. What is the point of reproduction?

There would be no point to expending the energy to engage in sexual reproduction if the environment were absolutely static. That's exactly when mitosis (not much of a CAS stratgey as they go) is an advantage. Why go through the trouble of meiosis? It's to adapt to a changing environment.

Moreover, not all CASs engage in reproduction in either sense. An ecosystem does not "reproduce." It "cycles" energy and nutrients.

Sexual reproduction is a variety of CAS, not the other way around. Complexity theory is a bigger tent than evolutionary biology. So one can't try to fit CAS behavior into the theory evolutionary biology or memetics; rather, it's the other way around. Kauffman's At Home in the Universe is a nice read to get this sorted out. JBR

7/25/2006 12:25 AM  
Anonymous jecado said...

I will need to sit down and look at Kauffman's books (among other resources), but I'm still a little stuck at this early stage on what counts as "fitness" in any particular context -- whether in biological terms or legal ones. I can envision a lot of ways to determine "fitness;" the most straightforward (and least end-directed one) would seem to start with the proposition that if a a CAS exists at a given moment, it is "fit" at that moment. A further level would seem to be a CAS's resilience (maintenance of the same / similar agents and rules over time and in the presence of perturbations); a resilient system would have "fitness" over time -- and amount to a more "fit" CAS at any given moment.

As long as you're going to avoid the notion of an end-driven process, is it that the main task is one of definition: Can you define the CAS's agents and rules in a way such that they are constant over time and in the face of perturbations (i.e., disasters a la Katrina). If so, the CAS is well-adapted (even if the "real world" results of a CAS are very different at time T+100 than at time T).

I'm trying to decide how the question of "growth" fits into this approach. If you have a CAS that is fit to a particular niche environment (whether in law or the real world), and that environment is destroyed (or made irrelevant) in the face of a perturbation, the CAS is not particularly fit regardless of how stable it may have been in that initial environment. If the CAS expands into other environments, though, it's not as susceptible and thus more fit (because it exists with the same rules and agents in the long run) at a particular moment.

I'm certainly missing something -- or perhaps simply seeing the obvious -- and look forward to further posts...

7/25/2006 4:00 PM  
Blogger J.B. Ruhl said...

jecado's last comment raises a very important question--what is "fitness" in a CAS in general, and in the legal CAS in particular--which I think I should address in a general post dring the current set of posts.

I am not sure I follow the second comment, regarding "growth." Assume a perturbation massively alters the environmental conditions within which a CAS had succesfully evolved--i.e., it was most successful among all CASs at "filling" that "niche." Then what? Well, we'd have to see. I think your point is that if it can't weather this storm, it wasn't so fit after all. Well, that's the conflicting constraints problem. A CAS that is highly dedicated to a particular set of environmental conditions may prove very hardy so long as those conditions are present, but pretty vulnerable as soon as they don't. Take the cichlids of Lake Victoria. They evolved into incredible niche specialists (evolving from 5 species to 400 in about 14,000 years), but throw in some Nile perch (an expert generalist) and that's that. Today the cichlids combined have fallen from 80 percent of the lake's biomass to one percent.

So, fitness is a complex metric. As I said in previous post, a CAS may seem fine, but really be stuck on a "fitness slope" that leads to a fairly small peak compared to others. That may not matter for a while, but when the "perch" comes along, it may be fatal.

Thanks for the comment. JBR

7/25/2006 10:49 PM  
Anonymous Anonymous said...

The discussion about fitness illustrates a couple of important points. First, absent some agreed upon definition of fitness, and an understanding of the relationship between fitness and persistence over time (i.e. reproduction), the insights offered by complexity theory become just a loose metaphor into which anyone can insert their favorite definitions of fitness (much like the debate over efficiency in law and econoland).

Second, most complexity theorists (including Kauffman) are quite okay about adopting the definitions of fitness offered by the evolutionary biologists, largely because those definitions avoid the confusion evident in the discussion between jecado and jb ruhl. Fitness is a relative term--it is measured only in relation to an environment. If the environment changes, then you have to recalculate fitness. So an organism that survives quite well in salt water might die quickly with an influx of fresh or brackish water. A legal rule that reduces agency costs in one environment might prove wholly useless in the face of new legislation.

As an aside, let me express a bit of skepticism about how productive an embrace of complexity theory by legal scholars can be, if folks don't take the time to really plow through the background foundational concepts.

7/26/2006 7:11 PM  
Blogger J.B. Ruhl said...

I am not sure whether Anonymous is addressing the comment chain alone or has also read today's post and is addressing that as well. I think today's post addresses some of these points, though I doubt to Anonymous's satisfaction.

A central theme of complexity theory is that the environment changes, the fitness landscape changes, and thus fitness changes. I'm not sure why (or whether) Anonymous thinks that makes fitness not a useful concept or complexity theory not a useful model. If CAS theory helps us understand how change affects fitness, and how different system designs sustain fitness as the environment changes, that strikes me as useful.

Today's post addresses the use of biological fitness in CAS literature, the more general throries of fitness in CAS literature, and what fitness could mean in the legal context. If the fact that we must agree on what fitness means in the legal context makes complexity theory not useful to law, then it is not useful to ecology, sociology, psychology, etc. Maybe it's not, but I'm trying to show otherwise.

As to whether complexity theory is just a loose metaphor, see the next post later this week.

If nobody was skeptical about this, it wouldn't be very interesting, so keep it coming Anonymous. JBR

7/27/2006 12:00 AM  
Anonymous Anonymous said...

I liked the post on fitness, though I only read it today, after I posted the earlier comment.

The post makes reference to survival, and of course for a legal system as for an organism, survival depends on reproduction over time. Is why biologists use reproduction and not survival of the individual organism as a measure of fitness.

Accordingly, I still think reproduction can be a useful measure of fitness with regard to particular laws and legal systems, so long as one doesn't get too general about what it is that is reproducing.

Reading the social norms literature across a wide variety of disciplines (not just the law and economics version), it is possible to think of the process of legal codification as a mechanism to transform a social norm into a legal rule, and thus ensure its reproduction over a longer period of time. A legal rule has a lot more staying power than a more informal norm or custom or rule. A community delegates responsibility for monitoring and enforcement of the rule (necessary to its reproduction) to the state.

The legal rule gets modified constantly--in operation, in geographic dispersal, via amendment, via interpretation, etc. (so reproduction with some recombination or mutation). Sometimes, the legal rule is thrown over by a competing legal rule, or extinguished by a change in the environment (so what would our abortion rules look like if the fetus technologically could survive from conception onwards outside the womb?).

Kauffman makes the point that you can measure fitness at different levels (for example, the ability to catalyze an enzyme), but the ultimate measure of fitness is still reproduction.

I think referring to the fitness of the common law system is still possible, so long as one recognizes that you must have a pretty good idea of what features characterize a common law system. I think it's more useful to focus on the fitness of those individual features (e.g., judicial interpretation of law that creates binding precedent).

7/27/2006 2:16 PM  

Post a Comment

Links to this post:

Create a Link

<< Home

Google
 
Web Jurisdynamics