here are some references to start out with:
i have bookmark lists on complexity convergence.
but i can't help but rant:
some stuff not categorized yet:
something stops being interesting if you can learn to predict it.
something is alive if you can't predict it.
definition of life is relative: a super-being may be able to `play
croquet' with something we consider alive---even ourselves.
some people use the word `emergence' to talk about life. a system is alive if `new' behavior emerges from a system whose rules don't code that behavior `directly'. how can we make a science of such `emergence' if we can't measure it objectively?
we draw a qualitative semantic barrier between two levels of a system if the connection between them is too `complicated' for us to understand, even though the basic rules on either end are understood. this is the barrier against which classic scientific `reductionism' hits.
personally i try to parameterize a very large space, and risk producing some duds. when bomb is used interactively, duds would be considered `bad notes' or errors; hopefully the user recovers and moves on. in auto-mode, when the system is improvising on its own, you must suffer with them. working on auto-mode is an attempt to change coordinates on the parameterization so that `interesting' areas are easy to find. a first step in this direction is automatic detection of `low variance' images, and rejecting them. thus we have returned to the search for a objective (ie exists as a program i can run) definition of noise. for example, we might do well to reject images that don't have `pink' frequence-power distributions.
[digression could you build a living system (one with OEE) that had a small parameter space? what if you introduced a meta-level: imagine a simple CA supporting a complex meta-behavior such as wire-world or neuron-models. but the large parameter space isn't eliminated, it just moves up to the initial conditions for the CA.
so say the True laws of physics are just simple equations. we know they support life, but only just the right initial conditions are alive: empty space is dead.... or maybe you need a lucky sequence of quantum improbabilities. or maybe you just have to wait a very long time (best estimate: 20 billion years). ]