Posted by Phil on August 01, 2003 at 11:25:22:
In Reply to: Ooh ooh ooh.... posted by David on August 01, 2003 at 11:02:43:
The population starts off randomly, OK, but I was thinking that if you gave each poem 4 or 5 random snip points when you generated them all and say there's 2000 poems, each poem is bound to have a couple of hundred others that share at least one of the snip points, and so could mate with it.
Don't fully understand what you mean about phermones but I'm sure you know what you're talking about. Can't wait to see what you come up with.
: You've just given me a great idea. I do like the idea of matching snip points to determine if two poems are of the same "species" however there's a "get there from here" problem as the population starts off randomly.
: But if I combine this idea with my old pheremone model, i.e. where there is both the species identifier as well as a "similarity preference mask" then we being to have a cool speciation mechanism. In other words, members of the initial population wouldn't really have a preference in terms of snip points, but that factor would evolve over time. Furthermore, the matches wouldn't have to be exact, but the more similar they were the more likely mating would happen.
: I haven't thought through this completely but this could be worth some good "feet on the desk, staring into space" time. Thanks. :-)
: --David
:
: : Further to my reply to 'Responses to postings below' below ;)
: : Snip points are treated as an extra word e.g. : : This could be a way of introducing speciation, i.e. poems are only allowed to breed if they have a snip point in common. This way each poem could only breed with a small subset of the population but there'd be a six-degrees style of thing where each poem can share genes with all the others indirectly. : : (alternatively if they don't have a matching snip point you pick a random one just for this mating).