The Minkowski Challenge

[shernren]

If shenren is correct, then this computer simulation is hopelessly flawed.

JR

I think you should refer back to relspace's excellent comment, he basically said that the flaw was a flaw of the programming language instead of any inherent methodological problem. The basic idea (take a program, subject it to random mutation + natural selection, see how fitness changes intergenerationally and see if new features evolve) is actually quite sound, genetic algorithms are roughly built off this idea (minus new features ) and perform optimizations quite well.

The problem is that C++ reacts differently to random mutation than DNA. Randomly mutate C++ and you get junk. Randomly mutate DNA and often (though not always or even most of the time) you get a subtly different but still roughly functional protein. This btw is a good indication of design, C++ is a language built exclusively for human programming with zero tolerance for random modification since none is expected. (Makes you wonder why DNA shouldn't be the same if it was designed. But that's a totally different issue altogether.)

If we had a programming language that could produce a subtly different, but still roughly functional program, and a programming environment where these subtle differences could have noticeable effects, then the simulation would actually be quite a good idea. But C++ and conventional assembly/OSes simply don't make the cut.

 

[laptoppop]

It would be interesting to see if we could create a model closer to the minimum "life" -- but such a computer model would have to be amazingly complex, and we keep learning more and more just how complex it is. Even one celled bacterias are still AMAZING creations. I would guess that true complete modeling is outside what we can currently do.

 

[cubanito]

I understood the objection on the basis of tolerance for informational "static" between C++ and proteins. An alternate explanation that this indicates deliberate design on the part of C++ but not proteins is that the hypothetical designer of biological systems was much more intelligent than us. Thus this hypothetical Creator designed a system capable of adapting and evolving within certain enviormental parameters. Perhaps such a Creator from the outset created a variety of genes that would each be selected for in various enviorments. But all this is peripheral to what I view as a far more crucial error in the experimental design.

The main problem is the lack of a defined end point that is comparable to answer the central question. The central question is the evolution not of various subspecies, or "species" as defined by political consensusl but rather of true full blown speciation. That is, not a wolf into a dog, and back again, but a wolf into some completely new animal incapable of back breeding at all.

There is a constant confusion between "microevolution" and "macroevolution." Everybody acknowledges microevolution. The question is wether the acrettion of microevolution can lead to macroevolution. Can a cow really turn into a whale, or a bear-like animal?

The computer simulation would be irrelevant to the center question if it merely showed that a virus program can modify. It would need to show it can become a completely different type of program.

JR