Home » Darwinism, Intelligent Design » Memo to physicist David Thomas: Make Darwinism work. Get an intelligent agent involved.

Memo to physicist David Thomas: Make Darwinism work. Get an intelligent agent involved.

 Dave Thomas's photo

Dave Thomas

Heck, Dave will do. Who are we to be fussy?

Physicist/mathematician David Thomas has boasted that evolution creates information and that he can show this by solving the Steiner tree problem using its powers.

Steiner tree problem: To connect 5 (or whatever) cities with roads using the shortest combined road length.

Thomas has challenged,

If you contend that this algorithm works only by sneaking in the answer (the Steiner shape) into the fitness test, please identify the precise code snippet where this frontloading is being performed.

The guys at Evolutionary Informatics Lab (the one the Baylor dean tried to can years ago, remember?) do exactly that:

The precise code snippet where this frontloading is being performed” from Thomas’s Fortran version of the program is shown below. It ensures that there are at least two interchanges (Thomas calls them variable points) during the initialization of
the population [21]:


Robert J Marks II comments:  ”In fact, we show the problem attacked by Thomas is pretty lame in comparison with other Steiner tree solutions in the literature.”

Here’s the goods:

“Climbing the Steiner Tree–Sources of Active Information in a Genetic Algorithm for Solving the Euclidean Steiner Tree Problem” (BIO-Complexity, Vol 2012 )

Winston Ewert, William A Dembski, Robert J Marks II


Genetic algorithms are widely cited as demonstrating the power of natural selection to produce biological complexity. In particular, the success of such search algorithms is said to show that intelligent design has no scientific value. Despite their merits, genetic algorithms establish nothing of the sort. Such algorithms succeed not through any intrinsic prop- erty of the search algorithm, but rather through incorporating sources of information derived from the programmer’s prior knowledge. A genetic algorithm used to defend the efficacy of natural selection is Thomas’s Steiner tree algorithm. This paper tracks the various sources of information incorporated into Thomas’s algorithm. Rather than creating informa- tion from scratch, the algorithm incorporates resident information by restricting the set of solutions considered, introducing selection skew to increase the power of selection, and adopting a structure that facilitates fortuitous crossover. Thomas’s algorithm, far from exhibiting the power of natural selection, merely demonstrates that an intelligent agent, in this case a human programmer, possesses the ability to incorporate into such algorithms the information necessary for successful search.


Follow UD News at Twitter!

(Visited 362 times, 362 visits today)
  • Delicious
  • Facebook
  • Reddit
  • StumbleUpon
  • Twitter
  • RSS Feed

6 Responses to Memo to physicist David Thomas: Make Darwinism work. Get an intelligent agent involved.

  1. Dave fails to appreciate that information can be represented algorithmically. The front loading doesn’t have to be the actual geometry of the solution but an algorithmic representation of the solution or part of the solution.

    And I have a surprise for him. He owes me a favor since I responded to his request for participating in a steiner solution.

    Now it’s my turn to pose a puzzle for him. He’ll get a free lunch (figuratively speaking in the form of a donation to the NCSE or some comparable Darwinist charity) if he is able to write an evolutionary algorithim to uncover the specified complexity of an artifact.

    I’ll post this later. Stay tuned Dave Thomas, you owe me a reciprocal response. :-)

  2. Related question:

    What relation, exactly, does a Steiner tree problem have to biological systems?

    I can see that a Steiner tree solution might be viewed in some sense as specified information. Depending on the particular problem, however, we might be justified in questioning whether it is complex information. Further, it seems that (at least in some cases) it may be the type of problem that might be amenable to brute force calculations. Finally, if a particular Steiner tree problem converges on a single “correct” solution, that might be somewhat different from creating a functional physical system from aperiodic digital information.

  3. 1. It will take some time to look into the paper in greater detail, but I’ll surely enjoy it!

    2. In fact, we show the problem attacked by Thomas is pretty lame in comparison with other Steiner tree solutions in the literature. So what? Isn’t that true for most evolutionary algorithms? Efficiency isn’t the great strength of GAs, but they are easy to implement and applicable in many situations….
    3. Bob Marks, have a look here for a very simple critic of a problem with your paper ”A Search for a Search”. I’d like to get you input!

  4. Biological evolution might be modeled as a GA, but it is inappropriate to assume that the biological evolution attempts to solve the problem of creating rube-goldberg machines and IC machines. The problem with biological (real-life) GA’s is they converge on solutions other than increasingly more integrated complexity.

    Empirical observation, clearly shows real-life biological GA’s converge on solutions of less complexity, not more, and often no solution is found (extinction).

    Thomas’s demonstration is misleading if he is trying to apply it to biology.

    Now if he is saying merely that he was able to find a free-lunch (algorithmically speaking), that has been refuted numerous times.

    The measure of information is the measure of the reduction of uncertainty. Just the fact that he knows in advance that a GA can solve the steiner problem implies he has front-loaded knowledge of the problem.

    To illustrate the point that not all realms of search are amentable to this, I will pose my own challenge to Dave. He better respond since I responded to his request to participate in finding solutions to the steiner problem.

  5. Gammarus minus and blind cave fish are good examples of how GA’s really work in nature. Complex solutions are selected against, not for. That’s what we would expect theoretically and it is confirmed observationally. There is no free lunch.

    The violation of no free lunch principles in biology is in the imagination of Darwinists.

  6. I’m still waiting for a demonstration of the efficacy of undesigned algorithms.

    Also, algorithms, by their very nature, are teleological.

    So have we allowed teleology back in to scientific discourse now?

Leave a Reply