Home » Intelligent Design » GA This!

GA This!

The concept of IC is that an IC system has no *functioning* precursors with the same function. Zachriel’s program (and other similar programs randomly generating phrases) don’t challenge IC because the target phrases do not have functioning precursors. As in, they do have precursors but they do not provide the same meaning. Thus, if IC was taken into account normally these words would not promote survivability. If Darwinists would deal with programs instead of phrases they would understand quickly what a functioning precursor is.

Here is a modification of the Phrasenation program that I’d find interesting and possibly even relevant to discussions of ID. I’d like suggestions on how to more accurately reflect the problem Darwinists face.

1. Start with one phrase line:

Overwhelming Evidence

2. Here is the target goal:

Listen, people
Evolution’s true
We can promise you,
Cause we’ve got evidence, people.

You say our textbooks lie,
But only in your eyes,
Cause you’re creationists,
While we are scientists.
At the end of the day,
The truth is what we say ­

Cause we’ve got
Overwhelming evidence,
Overwhelming evidence,
Overwhelming evidence ­
To prove that our theory’s true.

Oh, the peppered moths
Aren’t where we thought
It’s a big mistake
And the photos are fake.
We know it’s unreal,
But it’s no big deal ­

3. There are 20 additional lines or “components” on top of the starting point. Each line is treated as a whole. So the song in its entirety is considered IC as well as being CSI.

4. No front-loading of a correct line or “component”; the fitness function only checks for for one block (see point 6) and the entire song. Allow for single words. Partial words don’t count.

5. The randomize function not only randomly generates lines but the order they’re in. Fitness function eliminates if generated lines are out of order. Original phrase line “Overwhelming Evidence” always survives.

6. We’ll be “nice” and say that the 3rd of the 4 “blocks” in the song is like the TTSS and can serve an alternative standalone function made up of 5 components/lines. The fitness function WILL check to see if a series of lines match this block. BUT no survival benefit AT ALL (no free bonus points nudging it toward the front-loaded goal) is conferred unless the block is generated in its entirety. Only if all 4 blocks are generated at once will the song have any survival benefit. Anything outside of the original line, the 3rd block, and the entire song is eliminated since it doesn’t confer any survival benefit in the partial form.

7. Although I pasted the song as blocks for readability that isn’t considered by the program. Only the individual lines.

In short, the program must make the leap of first generating the 3rd block AND then all 4 blocks to form the song with its 21 components. I imagine that the program might be able to hit

Overwhelming evidence,
Overwhelming evidence,
Overwhelming evidence ­

every so often via duplication but I’d be slightly surprised to see it hit the entire block.

Of course, Darwinists will whine that we need to break the entire song into smaller pools of stepwise survivable blocks to allow for a reasonable indirect Darwinian pathway. Well, biological reality is not that nice so all I have to say is tough noogies.

What is the take home lesson from all this? RM+NS works with certain types of problems when it is DESIGNED and implemented properly. Do we have any indication that nature is designed as such? I would not say nature is designed improperly. Instead, I would say that such natural mechanisms were designed to complement other methods of design to allow the wonderful diversity of life we see.

  • Delicious
  • Facebook
  • Reddit
  • StumbleUpon
  • Twitter
  • RSS Feed

10 Responses to GA This!

  1. 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.

  2. 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.

  3. 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.

  4. 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”…)

  5. 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?

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

  7. 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.

  8. 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.

  9. 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.

  10. 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.

Leave a Reply