I would call the blood type thing trivial as well, but not the concept of using genetics in-game. You call it trivial, but one of the things I am most impressed with that Toady does is how the worlds become mathematical models of populations, civilizations, ect. As such, Genetic Algorithms themselves could indeed add an interesting aspect to the game. Genetic algorithms are nearly by definition a generator that generates some factors starting from a seed (so for our purposes, random), plus selective factors on those traits.
It relies on having randomized ranges. So if a shark has initial size 10+-2, it could be from 8-12. And some parameter is needed for selection. Lets say that we set it, for sharks, to have increased feeding need based on size, and has whatever tag that was where it attacks anything smaller then it. So there's two ways it could play out. 1. those with increased need to feed would need to attack more. On the one hand that is putting it in danger more, but on the other hand the increased size makes it better able to deal with it. So if it is up against size 9 things that usually kill sharks without effort, then being small might be good so that it does not attack those things, while being size 12 might increase success. So now, lets say we run this for a while.
Most size 10s die, size 9's and 8's don't attack those things but go a little hungry, size 11's and 12's do and manage to pick off a few and feed. So the size parameter of living sharks is now used to generate the next generation. Size 12's still have the +-2, so they are now size 10-14, and the 8's have 6-10.
Now, those of size 14 whoop arse of their normally mortal enemy, feed a good bit, and reproduce. Size 6's attack only smaller animals by the water and survive to reproduce. Even if in-game the smalls 'mate' with the larges, this system allows the population to develope into two subpopulations with various traits. We've selected for small and large variants based on a single natural food/enemy...
This could be a way, also, of making the PLAYER or the DWARVES THEY ARE PLAYING be the selective factor. I once played a game where randomized 3-D triangles attacked you. They had random speed, shooting ranges, basically random 'behaviors'. Each level used the ones that did the most damage to you as the seeds for the next level's enemies. So it adapted to your style of play. If I am the sort to stand around and shoot far away, then thoser that shoot from afar will do more damage then those thatdart forward to attack and get killed. If I am the type to go in close and shoot wildly, those that shoot from afar are less advantageous because they won't often hit me, so those that rush forward are best. After the third level, I had all triangles that rushed forward, and at that point I was done for. Likewise, there was one time I decided to play the other way, and I selected for the far-rangers. I ended up with so many shooting from afar that I couldn't dodge them all anymore, and again died repeatedly. So I selected for two separate tendencies both of which were successful against me. In this way, the player can be the selective factor.
So if you tend to hole up and shoot through fortifications, goblins with A. the trait of shooting more from afar up at your dwarves behind the fortifications and B. higher marksgoblin skill, will do more damage. If that is the factor used to generate the goblins in the next wave, there will be a steadily increasing selection toward long-ranged high-skill marksgoblins. If, however, you are the sort to, for example, set up walls and catch goblins by surprise with hordes of well-armored, well-armed dwarves, then you will be selecting for well-armored, well-armed goblins.
Depending on how wide the range of a randomized trait, and how much that trait matters for doing damage, the game could quickly adapt to a particular mode of play, and possibly add a level of difficulty to the game.
So do not discount the idea of genetic algorithms as a wonderful mathematical model, to possibly someday make their appearance in DF.
Author
Topic: Evolution and Genetics (Read 3004 times)