[ 5/January/22 ]
Hi Dirk & Mark,
To me, evolution has been a fascination for 50+ years.
What sort of systemic characteristics are actually required for it to work?
What is actually going on?
How deeply, recursively, complex is it, really?
What are the major systemic drivers?
It took me about a decade to start to get a handle on the major systemic drivers; and a further 30 or so years to really build a useful model of depths of strategy and system present in complex organisms like ourselves; and the sorts of evolutionary histories required to deliver such things (and, more importantly from my perspective, the sorts of systemic environments that give us a reasonable probability of long term survival with reasonable degrees of freedom).
Part of understanding what seems to be going on is understanding “search”. When one looks at search for a fully loaded processor then it is clear that the most efficient search possible is the fully random search.
When one looks at the evolution of neural networks, and the biases necessarily present to allow them to make any sort of sense of reality in any usefully short space of time, then that does present issues for how such a network might approximate random search (hypothesis – enter the autism gene – writing as an autistic).
Going back deeply to the start of replicating systems, the idea of irreducible complexity applied to life alone doesn’t help – it is an over simplification of something. And to be able to get a clear idea of why that may be so one needs to be able to deal with all of the concepts present in Wolfram’s NKS (New Kind of Science) – in terms of the irreducible complexity of computation more generally as a fundamental characteristic of the reality in which we seem to exist. When you do that, it takes away the restraints on the evolution of self replicating systems; but it does impose constraints of complexity only being able to emerge in contexts where cooperation can actually survive long enough for effective sets of cheat detection and mitigation systems to emerge (and those necessarily become evolving ecosystems in themselves at every level of complexity). Examining the space of all possible contexts for general classes of such contexts is not for the faint hearted (there be dragons, monsters and demons – metaphorically speaking).
It is seriously complex, writing as someone who has been doing complex math inside my head for 50+ years, and who knows it would take me hundreds of years with current tools to write out something that others could follow in a simple step by step process – the use of “random search” does deliver that unsettling characteristic – there really are random jumps, some of them across vast systemic landscapes, and the experience of “flying” over such “landscapes” is very different from that of walking through them (writing as both pilot and tramper {both metaphorical and literal}).
Ted Howard NZ's Blog 