Thursday, June 05, 2008

Could computer models of biological systems disconfirm Darwinian biology?

This is by pro-design biologist Douglas Axe.


Anonymous said...


Just the opposite; it confirms biological evolution by natural selection:

Anonymous said...

How does it confirm biological evolution since computers are programmed by superior rational minds. Not the non-rational forces of nature like evolution would call for.

Anonymous said...

Just the opposite; it confirms biological evolution by natural selection

Anon, would you mind explaining how the article from sciencenews supports the above claim?
I've read Doug Axe's article and the 2nd one you posted and I just don't get the connection.

Mike Darus said...

Victor outs sloppy naturalistic reasoning related to the brain and reason by his "Mr. Brain" reference. Too often naturalism injects language inconsistent with naturalistic claims. It seems we have a new character named "Mr. Life" who guides the evolution of increasingly complex machines.

Ilíon said...
This comment has been removed by the author.
Ilíon said...

"Materialism is the philosophy of the subject who forgets to take account of himself." -- widely attributed to Schopenhauer.

"The whole is greater than the sum of the parts." -- a frequently silly, and probably always fallacious, claim and belief.

"Computers are Magick." -- a generally unspoken, yet widespread belief.

Understanding the above statements will go far towards enabling one to understand the basis or genesis of the faulty reasoning and misplaced will-to-believe behind *all* such (false) claims as those advanced for the Avida computer program, or often for 'genetic algorithms' in general, and the expansive (and frankly silly) claims about the soon-to-be-realized wonders of "Artificial Intelligence."

There is nothing at all magic about computers or computer programs.

A computer is *just* a glorified and mechanized abacus. Computers compute -- which is to say, that based on the program being executed, they contingently count and mechanically compare different counts -- that's all they do, that's all they can do, that's all they can ever do.

A computer program -- which is the relevant thing when talking about what computers do -- is just a larger or smaller collection of algorithms. An algorithm is the set of instructions for doing some specific task: the classic example of an algorithm is the instructions on a bottle of shampoo; and this is also the classic example of a faulty algorithm. When we're talking about a computer program, an algorithm defines how/what to count, what results to compare, and which instruction (or further algoritm) to next execute based on the result of the comparison.

There is nothing a computer program can ever do that you cannot do with a piece of paper and a pencil. OK, perhaps these days you'd need a ream of paper and several pencils. The computer merely does it faster and, since it's a machine, with essentially guaranteed accurately (this "guarantee" must ignore a physical problem in the hardware which could cause the program to malfunction) ... and, of course, you, being an agent, can do far more things, and far more interesting things, than any computer program can *ever* do.

When people make claims that some "computer simulation" "proves" this or that, what they are *actually* claiming -- and they may or may not be correct on this point -- is that someone successfully and correctly represented the various parts of some logical argument (which is not to say that the argument is successful logically) via various computer algorithms and then ran the program with some inputs and got some output.

My point at the moment is that regardless of whether some program does, in fact, accurately represent the argument someone wishes to advance, it's an absolutely pointless exercise: if the argument is sound and valid, then it's sound and valid without attempting to represent it via a program; and if the argument is *not* sound and valid, then *nothing* one can do will make it so.

If one does not believe or accept some argument when another agent presents it, how is it even rational to believe or accept it based on the output of some computer program, regardless of whether the program does actually accurately represent the argument?

Also -- frequently (and this is the case with Avida) -- when attempting to represent some argument via a computer program, the creators of the program flubb it: they incorporate into the program aspects/assumptions which are contrary to the argument the program is meant to "prove."

Avida is meant to "prove" the primary/fundamental "Darwinist" assertion which may be expressed thusly: "Information can/does come from nowhere!" But the information "generated" (i.e. the outputs of a specific execution) by any execution of the Avida program does no come from "nowhere" -- the information is *already* encoded in the program and its inputs from the very beginning.

The outputs of a computer program do not come from nowhere; the outputs are fully, totally, absolutely, *determined* by the logic of program and its inputs! The persons who make these silly expansive claims for the Avida program (and for 'genetic algorithms' in general) are overlooking -- sometimes intentionally -- this fact: it is meaningless -- or dishonest -- to speak of the outputs of a program without reference to its inputs, *all* its inputs.

Here's an amusing little story about the Avida program. Or, to be precise, about the mindset of its advocates. A couple of years ago on the ARN discussion board (back when I could still easily connect to ARN), I spent *weeks* trying to get some Avida advocates to admit that the program does not live up to its billing. Well, that's not quite the case: I knew the advocates would never admit this, so I was trying to let the advocates help me demonstrate to the undecided that this is the case. Anyway, I finally played the ace, so to speak, and I was accused of "breaking the program." I kid you not! -- I was accused of breaking the program because I told the general reader how he could empirically demonstrate on his own computer the truth of one of the points I had been arguing.

Here is what that "ace" is: the specific results of any specific execution of the Avida program are generated by means of hidden inputs -- so-called "random" numbers which are internally generated by the program. Please *understand* this: Avida uses a set of publicly known and modifiable inputs (the various values in the program's configuration files) ... and a completely unknown and hidden set of inputs, but inputs nonetheless.

Please *understand* this: the expansive claims asserted about the capabilities of the Avida program (and frequently for 'genetic algorithms' in general) are advanced by means of deception -- and the deception is intentional, it is not a mistake or misunderstanding on the parts of those who initially advance it.

So, the outputs of the Avida program are actually determined by a set of "random" (and unknown/hidden) inputs.

Now, at the same time, there is an value in one of the configuration files by which one can "seed" (as we programmers call it) the "random" number generator which is part of the program. The default value in this configuration file is "0" (zero), which instructs the program to generate the "seed" based on various computations, primarily on a value derived from the current time ... That is, the "seed" used for that execution is unknown to the user of the program and the set of "random" numbers subsequently internally generated by the program *appears* (if one were to actually know the content of the set) to be unique -- of course, the set of internally generated "random" inputs cannot actually be guaranteed to be unique, but there are so many possible sets that it's a vanishingly small possibility that any user of the program will ever get the same set twice.

Now, here is how anyone who has the program (last I knew, it was freely available for download) can demonstrate on his own computer the truth of what I am saying. This value in the configuration file, which controls the "seed" of the "random" number generator can be set to some number other than "0" before one executes the program. [The configuration file in question is the file named: "genesis" The configuration value to change is, by default: "RANDOM_SEED 0" Change the "0" to some other number.] If one changes the "seed" to "1," for instance, than one will get the *exact same results* no matter how many times one runs the program using this value. If one changes the "seed" to "12," for instance, than one will get different results from when it was "1," but one will get the *exact same results* no matter how many times one runs the program using this value.

The thing is, *any* competent computer programmer knows about this sort of thing. One does not have to be a genious programmer to know -- without even looking at the program's code -- that its results are acheived by these means. There is no magic involved with computer programs. Ever.

Moreover, the authors of the program know what they did, and they intentionally did it that way -- and the rationale for allowing the user of the program it choose the "seed" is so that the Avida "researchers" *can* get the exact same results if they need to do so in doing their "research." It just so happens that this programming choice gives non-programmers an easy way to see with their own eyes a key truth about the program which all competent computer programmers already knew.

The claims asserted about the capabilities of the Avida program (and frequently about 'genetic algorithms' in general) -- the *interesting* claims, the claims which are the entire reason the program was written -- are false. The persons who initiated these false claims *know* the claims are false; i.e. they lied. Many, though not all, of the persons who secondarily assert these false claims know they are false; i.e. they lie. But at the same time, few of these secondary advocates who innocently (that is, they were deceived by someone else, and did not initially know that the claims are false) advance these false claims about Avida will *ever* acknowledge that the claims are false; i.e. few of them are intellectually honest.

No computer program will *ever* be able to do what the "Darwinists" want to believe a computer program can do (that is, generate information from nowhere). No computer program will *ever* be able to do what the 'materialists' want to believe a computer program can do (that is, be an agent).

I have not merely asserted these two statements, I have *explained* to you, the general reader, why the statements are, and logically must be, true.

All the wonderful claims about what a computer program can do are based primarily on three things:
1) Intentionally ignoring that a mind or minds created *every* aspect of the program;
2) Intentionally ignoring that the output(s) of any computer program are are fully, totally, absolutely, *determined* by the logic of program and its inputs;
3) And, frequently, the use by the program of hidden inputs, generally "random" numbers.

There is no such number as "random number;" the term is used as a place-holder to mean: "some specific number, the value of which is to be determined in some way or another at the time the code is executed."