[-empyre-] self-modification, emergence

I'm going to have a go at self-modifying this tangent back in the direction of a-life. Self-modification might also be thought of as adaptation, learning, innovation... and depending on your view of biology and what constitutes a "self" then you might also find it in evolution. Lamarckian evolution theorises adaptive self-modification which can be passed on to the next generation. While it's been overtaken by Darwinian theories, it's not a done deal - there is some controversial science that suggests that parts of the immune system can pass on acquired traits (Ted Steele, Lamarck's Signature - see for eg http://www.abc.net.au/rn/science/ockham/stories/s14075.htm). If you stretch the idea of "self" out to a species (or in to a gene) then Darwinian evolution is self-modification too.

A-life, and a-life art, involve the pursuit of the emergent moment of excess and surprise. Emergence is a knot unto itself, but the ongoing emergence that characterises biological life, requires adaptive self-modification. Thinking about self-modifying code (or other technological systems) quickly runs us into the problem of brittle grammars: basically, how likely is it that some random "mutations" applied to a bit of c++ code, are going to result in code that is even functional, let alone interesting / adaptive? Imagine a robot trying to self-modify by picking parts from the shelves of an electronics store. The predefined grammars of our technological forms are not an ideal substrate for self-modification. Most artificial evolution gets around this by creating a grammar of its own. Some grammars are very rich (eg Karl Sims' or Steven Rooke's image breeders) and some are more limited (eg Latham's virtual sculptures), but all are constructed and involve their own constraints.

My favourite example of self-modification and emergence is the work on evolved circuit designs by Adrian Thompson of the COGS lab at Sussex (http://www.cogs.susx.ac.uk/users/adrianth/ade.html). Basically he used a programmable chip (a FGPA, field gate programmable array) to "breed" and automatically test thousands of electronic circuits, on a task such as distinguishing between two different frequencies at their inputs. Many many generations later, circuits were evolved that indeed fulfilled the task. But when the evolved circuits were analysed, they were found to operate according to no known principles of circuit design... for one thing they were tiny and incredibly efficient (using only a small portion of the array) but also they used no internal clock, and instead seemed to use the dynamics of interlinked feedback loops to analyse the input. Best of all, the circuits didn't work so well when the same design was transferred to another chip: the evolved design made use of the specific physical characteristics of its substrate.



This archive was generated by a fusion of Pipermail 0.09 (Mailman edition) and MHonArc 2.6.8.