This post is realy Awesome for what im working on now, a game bassed off Evolution and survival of the fittest.
One thing to remember about survival of the fittest is not that it directly applies to a predator/prey relationship (but maybe indirectly), but it applies to members of the same (or similar) spiecies (its not weather or not you can out run a cheeter, but weather you can outrun the slowest person :D ).
Quote:
I do get your aplication of the symbols, my original idea came from Shapes that you combine to create new shapes and so on until you came up with the perfect shape. When i realised there was no perfect shape i went another direction but you get the idea.
FOr a good evolution game I would say that not having a "perfect" shape is good. This needs players to constantly adapt and re assess their strategies to "win" the game, and also thet there will nearly alwayse be a better solution to the one that they are useing (increasing the replayability).
For those of you who found interest in this topic I've been thinking long and hard about a way to mix behaviors in a way different than that of mixing them through genetic algorithms (the GA's seems too mechanical and unnatural to me). It's just a concept right now, no real math or extra thought put into it... yet.
Anyways, say I have two symbols, one that groups two objects that have a variable Cake that equals 1 and a symbol that sets all symbols' Cake variable within a 10 pixel radius to 1. Say that the user fuses those two symbols together. Now what if there was a "post action observer" module that looked for recurring patterns in merged symbols' behaviors (acting as two scripts running linearly), that created the new symbol's behavior after finding patterns in the parents' behaviors acting together. So basicly, the system would observe that any two symbols in a 10 pixel radius automaticly were grouped and would attribute the newly created symbol with a script that tells it to group any two symbols in a 10 pixel radius and drop the set variable command (it would not look at variables, just 'physical' observations).
I could make it assume that the purpose of variables is to classify symbols for some type of action to work on them, in a way treating the variables as strictly metadata.
Edit: upon reading this over, it looks like I could actually use this with a GA algorithm where script commands having to do with setting variables are looked at as "recessive genes" and actions are looked at as "dominant genes". Damnit, why the hell do GA's keep invading my thought processes here... maybe I should pick up a book on them :p
Bloody hell, I don't think I can delve into this deeper, or even try to simplify it. Just going to go insane. Maybe when I do a shitload more research on the topic, I'll be confused enough to try again, but for now I prefer to keep my mind intact.
