what do you mean by "control is cheap"? stability of patterns once emerged?
Conversation
Replying to
To make a computer, you need a strongly constrained dynamically stable system. Cost of control is related to entropy
1
Replying to
if you mean entropy/energy gradient I probably agree
but that's the "boring" argument :)
1
although my idea is fragile: hinges on intelligence being a predictor of complexity and that complexity being irreducible
3
Replying to
IMHO all observable complexity is reducible to computation + perfect white noise; but practical resource constraints limit us
1
Replying to
exactly: what if that computation is just hard enough to approximate in a smaller system that no emergent intelligence makes the jump
2
Replying to
yup, but then again no intelligence has ever emerged from a random-seeded golly run to our knowledge
1
also, fsvo robust: a single misplaced cell can level your entire universe (if you're playing at the low level)
1
Replying to
which is precisely the case because single cells are totally causally reliable
1
Replying to
but does low level GoL 'naturally' compute anything interesting (e.g. with agency) if we don't design a second substrate on top of it