Uncommon Descent

22 September 2007

Real Simulations, Cartoon Simulations, and Evolutionary Informatics


Computer programs that purport to validate the grand claims of Darwinian (i.e., chance and necessity) biological evolution are a hoot.

In early August my aerospace R&D company sent me off to Livermore, CA for a four-day course in using a finite element analysis (FEA) simulation program called LS-DYNA, originally developed at Lawrence Livermore National Laboratory. It models the laws of physics and material properties with astounding fidelity. It is so powerful that it is used heavily in the automotive industry to simulate entire vehicles and how they behave during impacts.

On the first day of the course the instructor warned us that it is very easy to create “cartoons” with LS-DYNA (it not only generates all kinds of data, it produces beautiful animations). By this he meant that if your initial assumptions are not correct, or if the FEA tools are not used correctly, you can get results that look really cool but don’t comport with reality. Much of the course focused on avoiding cartoon-generating pitfalls.

I’m in the process of completing my first suite of real-world LS-DYNA simulations for a project at our firm, and I can attest to the fact that the instructor was right on the money.

If a simulation program that can model the laws of physics and material properties with utter faithfulness is subject to all these pitfalls, what chance do you think you have of modeling Darwinian biological evolution and having any confidence that the results correspond with reality?

Interestingly, investigating the validity of evolutionary simulations is one of the things Marks and Dembski have been doing with their evolutionary informatics research. In the case of an evolutionary program called ev, they demonstrated that all but 8.8 bits of information out of a total of 131 were smuggled into the program, and then it was squandered with an evolutionary algorithm. It turns out that random queries outperform the evolutionary algorithm by over 10,000%. I believe Dembski and Marks have plans to conduct a similar analysis of Avida.

Avida and ev are cartoon simulations.

Of course, this is what got Marks in trouble at Baylor. There is no need to investigate the validity of evolutionary simulations, because we know in advance that Darwinian mechanisms explain everything in biology.

These icons link to social bookmarking sites where readers can share and discover new web pages.
  • Digg
  • del.icio.us
  • Netscape
  • Reddit
  • Fark
  • Furl
  • NewsVine
  • StumbleUpon
Email This Post Email This Post Print This Post Print This Post
5 Responses




3:45 pm


Very Interesting Insights Gil,
Do you think it is possible to faithfully and robustly mimic the foundational parameters in a computer program to see which of the two mo^dels (RM/NS or ID) is correct? And do you think it is possible to correctly modify the correct theory from such a hypothetically robust computer program?




4:22 pm


Gil, I’m thinking exactly what bornagain77 is thinking. We hold our breath!

More seriously, when you say physical laws, and then give the example of a car crash, do you mean the simulation involves mostly mechanical equations coming from classical mechanics (non- quantum mechanics) and hence excluding electrical/chemical interactions?

IOW, it might be great for solid bodies but not very applicable to quantum systems—which is where most of the magic of biology likely resides.




4:56 pm



“quantum systems are where most of the magic of biology likely resides”

Why do you say that?




6:42 pm


I think we just need to discuss the limits of FEA in more detail but Gil’s conclusion is correct.

It would be better just to see FEA as a metaphor for Avida and ev. The similarities are well founded and described, but like all good metaphors it has to break down at some point.

The point here would be the fact that FEA has very little relation to the generation Specified Complexity and Avida and ev is all about generating Specified Complexity.

Hope this help.




8:09 pm


More seriously, when you say physical laws, and then give the example of a car crash, do you mean the simulation involves mostly mechanical equations coming from classical mechanics…

Correct. Although LS-DYNA is so sophisticated that it can model explosives (this was its original design purpose, to produce more sophisticated, smaller, and more powerful nuclear weapons). It now includes FSI (fluid structure interaction) through ALE (Arbitrary Lagrangian Eulerian) modeling.

In answer to bornagain77’s question: In my opinion, computer simulations have decisively demonstrated the vacuity of Darwinian suppositions. Meaningful results are only obtained through the introduction of intelligence and information into the algorithm and the design of the computer code. In addition, in bio-evo simulations, the search space is always cleverly reduced to the point where the program can reach a pre-specified goal with the available computational resources. Almost every aspect of the design of bio-evo computer simulations represents the antithesis of Darwinism.

So, as a bottom line: Attempts to demonstrate the validity of Darwinian mechanisms through the use of computer simulations have demonstrated the exact opposite. Design, information, and goals are the essential ingredients for success.

Post a Response