Comments on: GA This!

By: Patrick

Patrick — Tue, 06 Nov 2007 14:45:13 +0000

I would say that is possible to design a computer program so that every major function is very modular. Now most software engineers do not do this but in game development the scripting systems are sometimes designed this way. This allows game designers to drag and drop various components to create complex modifications quickly.

By: magnan

magnan — Sat, 03 Nov 2007 20:57:51 +0000

WinstonEwert: “You have picked a specific target, evolution is not bound to a specific target and as such any “good” simulation ought to have a multitude of possible results.”

This is the usual ploy, that carefully avoids the factor of complex specified information. Any coherent song, not just this particular one, is going to be a very tiny target in a huge landscape of random targets. Also, only this particular song or ones with very similar meaning (out of all possible coherent songs) are going to have good fitness functions for selection.

By: geoffrobinson

geoffrobinson — Fri, 02 Nov 2007 16:32:14 +0000

Granted and good point. We could create a virtual machine that models something akin to DNA. Like byte code for the Java virtual machine. I would think, in such a model, there has to be a pre-existant capability to accept and incorporate new code. In other words, I expect such a capability to be irreducibly complex.

By: bFast

bFast — Fri, 02 Nov 2007 15:12:39 +0000

Geoffrobinson, the problem with your hypothesis is that you are equating computers with DNA. However, DNA is programmed in a much more modular way than computers are. As a result the science of genetic engineering has developed which does exactly what darwinists claim happens naturally — they are taking genes from one organism, putting them into another, and seeing the semi-predictable result.

By: Patrick

Patrick — Fri, 02 Nov 2007 15:10:34 +0000

Well, HGT could be added by making it so that the randomize function can pull chunks of information from other songs in the database.

By: geoffrobinson

geoffrobinson — Fri, 02 Nov 2007 14:44:51 +0000

A thought occurred to me while lying in bed this morning:

Some Darwinists want to go with organisms transferring genes between each other as a source of new genetic information.

Try modeling that in a computer. Get two programs that try to swap source code/binary instructions with each other and see if they crash or keep going with new functionality. Should we give them the pre-existing capability of excepting new code or should the programs try to randomly evolve that?

By: apollo230

apollo230 — Fri, 02 Nov 2007 14:11:15 +0000

There is little reason to sweat over these evolutionary algorithms because none of them (to my knowledge) fully simulate a genetic system in its full glory – COMPLETE with virtual DNA, RNA, ribosomes and the whole works. Darwinists who think they have anything of worth are flat wrong because they have not simulated the actual problem. Rather, they have taken a simpler problem, “solved it” and declared “victory”. (Bill Dembski quoted some mathematician in No Free Lunch who said “if you cannot solve a problem, find a simpler problem, and solve that”…)

By: Patrick

Patrick — Fri, 02 Nov 2007 13:20:13 +0000

Fortunately I’m not claiming this is anywhere near to approximating the process of biological evolution. It merely illustrates a single problem. Though I suppose along with an English dictionary there could be a very large database of songs to allow a multitude of possible results.

By: WinstonEwert

WinstonEwert — Fri, 02 Nov 2007 05:02:08 +0000

You have picked a specific target, evolution is not bound to a specific target and as such any “good” simulation ought to have a multitude of possible results.

By: Patrick

Patrick — Fri, 02 Nov 2007 00:19:01 +0000

I could incorporate some more active information about the English language and the fitness function could check the “intended meaning” of a sentence block. If this intended meaning is very close to serving the same function as the target sentence then even though it’s using different words it will survive.

For example, “major facts” could substitute for “overwhelming evidence”. Or “you claim primers lie” for “you say our textbooks lie”. This would help the search since those phrase serve the same function yet are simpler to produce.