Title: Quantifying Self-Organization

Abstract: All good complex systems people know self-organization when 
we see it.  Unfortunately, we sometimes disagree about whether we're 
seeing it, and anyway "I know it when I see it" is for art or 
obscenity, not science.  For instance, we can't seem to agree about 
whether only open, irreversible systems can self-organize. I will 
review the ways people have proposed to quantify the idea of 
self-organization, and focus on one which seems particularly promising. 
  This will involve a brief detour through the theory of optimal 
prediction.  (Brief for you; it took me seven years!)  I will then 
demonstrate how we can quantify the degree of self-organization in 
cellular automata.  (There will be movies, if the hardware is 
cooperative.)  I'll close by sketching some possible applications and 
listing some cracked-open questions, which I have an idea of how to 
answer, and some wide-open questions, where I haven't a clue how to 
proceed.