Talk:Complex socio-ecological systems/Genetic algorithms

This week’s readings on genetic algorithms went far beyond my smallish realm of knowledge, but I was guerdoned by reading some of this material over again and realizing how incredible the potential for some of this research is. And while most of the texts are not very new, a casual perusal of the some recent titles in Google Scholar reveals that researchers in the past few years are using these concepts in work on developing subdivisions, project scheduling, and even job training. Perhaps not quite the sexiest venues for this research, or at least ones we would would associate with thinkers like Axelrod, Gell-Mann and John Holland. Still, it illustrates how some higher level computing can be applied to some of the most mundane tasks. Apart from Axelrod’s extensive description of the Prisoner’s Dilemma, and Gell-Mann’s exhortation that we not confuse complex adaptive systems with everything else, what I took away most from this week’s set of readings was the promise of this research evident in Holland’s SA piece. The mere idea that one can transfer nonlinear genetic complexity into binary numbers is, I suppose, not particularly new, especially for the technorati out there, but it is incredible to imagine that strings of bit information of varying type have the ability to encode everything from a gamete to multiple generations of a population. And even more complex still is the simulation of evolution and natural selection with something like 8,000 rules (as of 1992, at least—one can imagine that the processing power and know-how has increased since that time). From the standpoint of complexity and systems thinking, what I find both alluring and potentially dangerous about this line of research is that one can go from computing a single genome, to a series of 1’s and 0’s, to the development and troubleshooting of a jet engine, to modeling of an ecosystem, and then back again to the 1’s and 0’s. It seems that the analogies in these schematics are encoded in numeric criteria with endless variability, using incomplete parts (chromosomes or other “coarse” elements, as Gell-Mann calls them) to create larger wholes. (Or as Holland says, “The remarkable ability of genetic algorithms to focus their attention on the most promising parts of a solution space is a direct outcome of their ability to combine strings containing partial solutions.) Whither go these wholes? If there is no design evident, is there not greater risk? This, to me, speaks about what’s compelling about complex adaptive systems, connections to AI, and ultimately to sentience (http://illigal.org/2005/01/29/computer-sentience-is-possible-says-holland/): we want to design intelligent, complex systems—but do we want to live with the outcomes? Tegeticula

Similarly to Sam, I was allured by this week's readings elucidation of how much processes of evolution can be applied to technology, computer science and even decision-making processes by the rule of choice, combination, outcomes and learning or adaptation. I always like to think how much duality is infused in our lives and the enormous possibilities of binary systems. Thinking about genes, where the possible re-combinations of genetic codes are infinite, is both exciting and overwhelming. I like to think about applying microcosmos functioning to macrocosmos reality, such as when Holland refers to mutation, for example. Mutation would be the unpredictable playing its role, does not matter if it's good or bad. So, in Resilience science, the "surprise" plays an important role on unnanticipated events that can put the system under risk or take it to another stability state or regime shift. I like Hollands discussion on exploration and exploitation, thinking about how these ideas could apply to Resilience Science. Another example of microcosmos reflecting in macro phenomena is the idea of emergence or emergent properties, that are also shown through computer simulation. Axelrod states that agent-based modelling can help to make us understand emergent properties of bigger systems - there is something in common in the minds and choices of humans. Social norms emerge from social/cultural systems and sustain cooperation. People choose sides based on affinity. What might be the role of social norms in cultural resilience? I was comparing the idea of biological evolution with the idea of cultural evolution, or, to be more realistic, biocultural evolution (Gell-Man) - with the genetic algorythm combination and the learning processes that lead to adaptation. But is it adaptation always good? The graph in Axelrod shows that people will learn and will tend to reach the fitness landscape. How can we think about humanity evolution and learning, when the same processes that lead to adaptation also lead to destruction of our natural resource base?Simoneathayde 13:52, 28 January 2011 (UTC)Simone

SOME QUESTIONS: 1. What genetic algorythms can teach/offer to improve our understanding and management of socio-ecological systems? 2. What is the difference and consequences of applying inductive X deductive methods in scientific research? 3. What is parallelism in genetic algorythm functioning? What are some advantages of parallelism? 4. What is the rational-choice paradigm and why scientists have shifted to focus on adaptive behavior rather than rational choice? 5. What would be some applications of agent-based modelling in the study of resilience of socio-ecological systems? 6. Is adaptation always good? What are some examples of maladaptation resulting from anthropic manipulation of the environment?SimoneathaydeSimone