Talk:Genetic algorithm

This definition is incredibly off. Genetic algorithms do not create "programs" - a more appropriate term would use data, or solutions to problems. The most common genetic algorithm programs generate solutions in the form of either numbers or strings of ascii characters, and typically treat these strings as chromosomes, modeling real-world genetics. The definition listed here is closer to the one for Genetic Programming, a technique which creates programs using an evolutionary program and then evaluates them according to a particular fitness function. The third confusing term, "Digital Organism" is used when the programs themselves reproduce, with or without mutation (i.e. evolution). In general, the three classes of algorithmic techniques are designed for entirely different purposes: -Genetic Algorithms are designed to "solve" a problem that lacks a straight-forward answer (such as NP-complete problems) -Genetic Programming is used to develop NEW algorithms to perform a certain operation (sometimes, the operation is specified, sometimes it is not) -Digital Organisms are designed to model evolution primarily, and are very similar to Genetic Programming techniques. The most theoretical use would be to include the Digital Organisms on an Astrochicken. --Wildhalcyon 13:44, 24 Sep 2005 (GMT)