Evolution Simulations and Countering the 'Information Theory' Argument

[Wowbagger]

There are those to think that various aspects of Information Theory debunk the Theory of Evolution. My standard counter-argument has been "If that were true, then computer simulations of evolution would not be possible".
But, the deniers keep insisting that every simulation of Evolution, for one reason or another, does not really simulate Evolution.

Sometimes their understand of the theories (Evolution and/or Information) are just flat wrong.

Some argue that the sims are too basic: They claim there is no demonstration of speciation, nor the emergence of separate genders, etc.

Sometimes they claim the sims have a "goal" programmed into them, and that real evolution does not. (They fail to realize the "goal" is a simple version of fitness measurement.)

Sometimes they will argue that the simulations were created by an intelligence, and therefore, we are only "proving" that life can only be the product of intelligence.

So, what further factoids can anyone offer, to counter such claims?

And, incidentally, if anyone knows of a good sim. that demonstrates some aspects of Evolution, really well, such as speciation, or emergence of separate genders, you can link to them, here, please.

Thanks!

 

[T-Diddy]

As I tried to point out on this thread, I don't know about any good numerical proofs (or simulations, if you will). We really know way too little about how evolution actually works at the genetic level to make a valid computer simulation.

Things that have been successfully modeled, such as solid mechanics, fluid mechanics, etc. are all very well understood; there are no properties left to chance. Note that none of them deal in statistics appreciably, and most are closed systems. When you branch out into simulations involving probability, things get a lot hazier. Probabilities must be known with incredible precision to be valid. You can get around it for some processes (nuclear fission, for one, is not perfectly statistical, but this is surmounted by base-lining to test data and using open-ended solutions to run checks), but for the most part you are a slave to assumptions. Think about the stock market, for instance.

Evolution predictions would be nothing if not incredibly statistical, and I know of no testing or theory that can supply useful data for simulation. It makes global climate models simplistic in comparison, and you know how many different predictions there are on that argument.

 

[Dr Adequate]

Some argue that the sims are too basic: They claim there is no demonstration of speciation, nor the emergence of separate genders, etc.

Sometimes they claim the sims have a "goal" programmed into them, and that real evolution does not. (They fail to realize the "goal" is a simple version of fitness measurement.)

Tierra seems to answer both those objections.

Sometimes they will argue that the simulations were created by an intelligence, and therefore, we are only "proving" that life can only be the product of intelligence.

To which the obvious answer is that the fact that we can simulate something happening naturally does not prove that it happens supernaturally. It proves that it can happen naturally, and that we are in no need of a supernatural hypothesis.

If we do a simulation which shows that natural processes are capable of producing lightning, that does not prove that lightning is produced by Thor's hammer. It proves that natural processes are capable of producing lightning, and that we have no need of the "Thor's hammer" hypothesis.

If we do a simulation which shows that natural processes are capable of producing snowflakes, that does not prove that lightning is produced by Jack Frost. It proves that natural processes are capable of producing snowflakes, and that we have no need of the Jack Frost hypothesis.

And if we do a simulation of evolution which shows that natural processes are capable of producing adaptation, that does not prove that the adaptation of species to their environment was done by God by a miracle. It proves that natural processes are capable of producing adaptation, and that we have no need of the hypothesis of fiat creation.

 

[drkitten]

Things that have been successfully modeled, such as solid mechanics, fluid mechanics, etc. are all very well understood; there are no properties left to chance.

Right. Boltzmann statistics are completely deterministic.

Basically, and once again,

 

[Wowbagger]

We really know way too little about how evolution actually works at the genetic level to make a valid computer simulation.

So, the algorithm of Natural Selection, though well understood on its own, somehow, magically, gets all screwed up, when applied to the particular medium of genes?

Now, we might not have been able to completely document every single selection factor of every gene, everywhere, in every life form, that ever existed, yet. But, I suspect they will all fall into the algorithm of Natural Selection, very nicely. (Unless, of course, you got evidence otherwise?)

Evolution predictions would be nothing if not incredibly statistical, and I know of no testing or theory that can supply useful data for simulation.

I agree that trying to predict the future of life, on Earth, is difficult, to say the least, and highly dependant on statistics and chaos theory, and all that.

But, the simulations I would like to see are only required to demonstrate evolution in a basic way. Even if the entities are not entirely realistic.

I rather like this video: http://www.youtube.com/watch?v=mcAq9bmCeR0
What if clocks could evolve?

Tierra seems to answer both those objections.

Ah, very much worth a good look at. Thanks!

To which the obvious answer is that the fact that we can simulate something happening naturally does not prove that it happens supernaturally. It proves that it can happen naturally, and that we are in no need of a supernatural hypothesis.
(snip)

Yes, I know this. But, the creationists would argue that computers are not nature.

What would be really cool, is for someone to physically demonstrate that a novel life form can emerge in the actual world. Like taking the Miller-Urey experiment many, many, many steps further. To an actual life form.
I am sure someone will probably do that, some day. Though, I suspect it may take a long, long, long while.
So, for now, we are stuck with computers. *sigh*

 

[Dr Adequate]

Yes, I know this. But, the creationists would argue that computers are not nature.

At which point I should ask them in what way that statement is relevant to the question of whether lightning is produced by Thor's hammer or snowflakes by Jack Frost.

What would be really cool, is for someone to physically demonstrate that a novel life form can emerge in the actual world.

Do RNA species count?

 

[Shalamar]

S

I rather like this video: http://www.youtube.com/watch?v=mcAq9bmCeR0
What if clocks could evolve?

Ah, very much worth a good look at. Thanks!

I LIKED that Video! In theory, one could build one of those 4 handed clocks that 'evolved' in the simulation!

 

[Wowbagger]

Do RNA species count?

Well, from our own open-ended definitions of 'life form' it could. But, I doubt it will for the general public. (They hardly qualify viruses as life forms.) They will need to see a novel "amoeba"-like-thing, at the very least, I imagine, for it to count.
How close do you think we are to something like that?

 

[rocketdodger]

Tierra seems to answer both those objections.

Do you know of any simulations that start at a lower level? I am trying to find something that simulates abiogenesis rather than evolution, specifically how something like the DNA coding system could have arised.

 

[Paul C. Anagnostopoulos]

Do you know of any simulations that start at a lower level? I am trying to find something that simulates abiogenesis rather than evolution, specifically how something like the DNA coding system could have arised.

I think such a thing would require a detailed simulation of chemistry or an analog to it.

Apropos:

http://www.us.net/life/


~~ Paul

 

[rocketdodger]

I think such a thing would require a detailed simulation of chemistry or an analog to it.

Apropos:

http://www.us.net/life/


~~ Paul

Holy crap... I hadn't heard about that!

I will get right on it...*starts writing code*...

 

[Agent : Orange]

Hey all.

Though it's a little further removed from what some of you might consider strict simulations of evolution, I work with genetic algorithms (GAs) quite a bit. Though these routines do not attempt to mimic biological evolution in strict detail, they are analogous to the process of evolution. I don't know about any simulations done with sufficient detail to illustrate the differences between male and female members of a population so I don't have much to contribute to this area. However I do have experience using evolution as a strategy for solving difficult optimization problems.

Speciation effects are common in the VEGA multi-objective GA as detailed by Schaffer (1984), who discovered speciation occurring through members of a population that satisfy a single objective well at the expense of other optimization objectives. It was later shown by Richardson (1985) that Schaffer was essentially doing a single-objective problem than a true multi-objective one, with each species lying at one extreme of the trade-off-surface (Pareto-front) between the objectives. The process of niching (sharing) in modern genetic algorithms allows the population to occupy unconnected distinct areas of the parameter space simultaneously. This niching process essentially splits up the population to inhabit unique biological niches, similar to speciation. However unlike speciation as exhibited by the VEGA routine, niching is a tool used to explore the search space and delay convergence of the population due to the effects of genetic drift.

Multi-population GAs and the effects of speciation and genetic drift on these optimizers is well-studied. If anyone is interested in Genetic Algorithms in general, you should make it a point to check out Goldberg's monograph from 1989. It's a great read and explains many of the sticky points of using evolutionary strategies in optimization problems.

 

[Radrook]

The above might not be as all-pervasive as it seems:

Why Mathematicians, Computer Scientists, and Engineers Tend to be More Skeptical of Darwinian Claims
http://www.uncommondescent.com/darw...end-to-be-more-skeptical-of-darwinian-claims/

 

[Paul C. Anagnostopoulos]

I guess we need a list of mathematicians, computer scientists, and engineers who are significantly skeptical of Darwinian claims, and a list of those who are not.

I'll start the list of those who are not.

Paul C. Anagnostopoulos

 

[Wowbagger]

The above might not be as all-pervasive as it seems:

Why Mathematicians, Computer Scientists, and Engineers Tend to be More Skeptical of Darwinian Claims

The guy in that article talks about himself, as if what he thinks matches what so many others tend to. This is not a demonstration that Information Theory disproves Evolution. It is just a couple of mis-informed questions, from one guy, with a seemingly exaggerated headline.

Do you have any evidence-based information you can provide, that would indicate my effort should be futile?

I guess we need a list of mathematicians, computer scientists, and engineers who are significantly skeptical of Darwinian claims, and a list of those who are not.

Science is not a matter of Argumentum ad Populum.

If the best argument Radrook can come up with is that "lots of people disagree", I say let him keep that one. I prefer to pick my fights with demonstrations and evidence.

 

[rocketdodger]

The above might not be as all-pervasive as it seems:

And, although we start out (apparently) more skeptical than normal, we end up more convinced than normal because we can just write a simulation to answer our questions on our own. Kind of like what I did. And Paul did. And Adequate did. And Delphi did.

When code I wrote myself shows me that the mechanisms of evolution work, it is pretty hard to ignore.

 

[Paul C. Anagnostopoulos]

If the best argument Radrook can come up with is that "lots of people disagree", I say let him keep that one. I prefer to pick my fights with demonstrations and evidence.

Aw come on, the lists are fun, too. They are a kind of tally of where people stand on the demos and evidence. No one thinks they are a logical argument one way or the other. Oh wait, I guess some people do.

If you want to read the definitive commentary on Dembski's messed up ideas of information theory and Intelligent Design, here's the paper by Elsberry and Shallit:

http://www.antievolution.org/people/wre/papers/eandsdembski.pdf

~~ Paul

 

[Paul C. Anagnostopoulos]

When code I wrote myself shows me that the mechanisms of evolution work, it is pretty hard to ignore.

Indeed.

~~ Paul

 

[kjkent1]

Is the arrangement of grands of sand on a typical beach designed or random? How would one use Dembski's math to differentiate a designed beach from a randomly-arranged beach?

Seems like the beach arrangement is highly improbable, so it must be designed.

But, if there is a universal designer, then isn't EVERYTHING designed, by definition, no matter how improbable it may seem?

Or, is everything designed except for those things which are truly random?

And, how does one prove absolute randomness?

I think that all of the above represents the reason why philosophy was invented: to fill in the gaps while we research the answers.

 

[drkitten]

Is the arrangement of grands of sand on a typical beach designed or random? How would one use Dembski's math to differentiate a designed beach from a randomly-arranged beach?

Well, since his math demonstrably doesn't work, it would be difficult to use Dembski's math to do much of anything.

But in theory, if his math did work, you would look at the properties of the particular beach that could be legitimately prespecified in advance. For example, a sand arrangement that is both symmetric and gravitationally unstable in the presence of water is a sign of a sand castle, not a naturally occurring beach formation.