Annoying creationists

[Yahzi]

I think Dr Schneider’s model is essentially correct.

Why do you think his model is correct?

For normal people, the answer is: because it models the observed facts.

You quite clearly feel it does not model the observed facts. What, then, is your basis for asserting the model is accurate?

 

[Beleth]

Uh oh. Multiple sig lines by the same person.

~~ Paul

Awww man! And during a thread where the one you took out (mine) is just so freaking appropriate!

 

[Yahzi]

Awww man! And during a thread where the one you took out (mine) is just so freaking appropriate!

I believe the accepted and appropriate comment for me to make at this juncture is:

Nyaah nyaah nyaah!

 

[joobz]

Let's not attack kleinman until he places his cards on the table.

Instead of the suspense building, kleinman, post the entire theory and supporting equations that make you feel that the model is wrong or that the model shows evolution is wrong. We can debate those claims once we've seen them.

So far, you've posted nothing earth shattering, nothing remotely difficult to disprove or find inaccurate. Right now, it is up to you to support your case.

We wait patiently.

I will state quite plainly, this is the first time i've ever dealt with any in science who didn't lead with their results. the cooler more dramatic the results, the more that person will shove it under your nose. Normally I know i'm gonna see something exciting when a person runs into my office sayinig, "Check this out!"

 

[Bronze Dog]

Is this just more magical thinking - the law of contagion in action (having gained a piece of evolutionary theory, it magically relates to the entire thing)? Or does he think he's doing the same thing we do when we use a Bible contradiction to overturn the inerrancy of the Bible?

I'd go with the voodoo hypothesis.

 

[kleinman]

Annoying Creationists

There are four user modifiable controls below which allow the user to specify convergence conditions including when a Perfect creature has evolved which occurs when the random point mutations and natural selection process has located all gamma number of binding sites on any genome without erroneously identifying a binding site where one should not exist.

Surely, the point is that there is no perfect creature. By prespecifying an end point you're trying to determine the time taken and probability of reaching that end point. In the real universe, assuming ID isn't correct, there isn't an end point and evolution goes to where it goes to.

The “perfect creature” terminology is not of my choice. What a “perfect creature” is according to the ev concept is a genome that has evolved to the point where every binding site that should be identified is identified and no binding sites are identified that should not be identified. What happens at that point is the evolutionary process plateaus. This is probably why Dr Schneider proposes this is an example of punctuated equilibrium.

I believe prove my case that ev shows that macroevolution by point mutation and natural selection is mathematically impossible...

Who says that macroevolution occurs by point mutation alone?

What mechanisms do you want to propose?

I think Dr Schneider’s model is essentially correct.

Why do you think his model is correct?

Tomorrow I will describe the theory behind the model and why I believe the model is essentially correct.

For normal people, the answer is: because it models the observed facts. You quite clearly feel it does not model the observed facts.

No computer model will ever accurately simulate reality when input parameters do not reflect realistic values. This is what Dr Schneider has done.

What, then, is your basis for asserting the model is accurate?

I’ll post this now since you have asked twice.

Ev and the basics of information theory.
The Shannon definition for information turns out to be mathematically equivalent to the negative of the quantum mechanical definition for entropy. If one considers that entropy is the measure of randomness this relationship becomes intuitively apparent. Increasing the information in a system reduces the randomness and thus reduces the entropy. So how does this relate to genetic evolution. One of the basic problems of Information theory is to take an initial ensemble with an initial probability distribution to a final ensemble with a final probability distribution by the input of information. In other words, you take a more random higher entropy ensemble to a less random lower entropy ensemble by the input of information. When information theorist talk about 1 bit of information, they are saying that based on a single yes or no question, the answer to that question allows them to decide which ensemble has a lower entropy. The answer to the binary question allows you to reduce the entropy and therefore the randomness by 1 bit.

In the ev model, the initial ensemble is the initial random genome. In the ev model, the questioning process are the random mutations. If a random mutation hits a locus in a binding site and improves the match between the weight matrix (binding protein) and that binding site it increases the information in that binding site. So the weight matrix gives the answer. Natural selection then chooses the ensemble with the higher information (or in this model, with the fewest number of mistakes in the identification of binding sites).

Unlike many IDers who have criticized Dr Schneider’s simulation, I find his basic concept plausible. What I don’t find plausible is Dr Schneider’s use of unrealistic parameters. Dr Schneider used a genome length of only 256 bases (2000 times smaller than the smallest genome known in any free living organism) and a mutation rate of 1 mutation per 256 bases per generation (4000 times higher than the average mutation rate in prokaryotes) in his publication on ev. When more realistic parameters are used in the model, the rate of acquisition of information become profoundly slow as will unfold as the data from ev is presented. This rate of information acquisition becomes so slow that it shows that it takes hundreds of millions of generations to evolve less than 100 loci on a genome of only 100,000 bases. Point mutation and natural selection is a profoundly slow process. Far too slow to evolve a gene de novo.

Let's not attack kleinman until he places his cards on the table.

Wait until I put my cards on the table and then attack.

Instead of the suspense building, kleinman, post the entire theory and supporting equations that make you feel that the model is wrong or that the model shows evolution is wrong. We can debate those claims once we've seen them.

I posted the basic theory above so now you can attack.

So far, you've posted nothing earth shattering, nothing remotely difficult to disprove or find inaccurate. Right now, it is up to you to support your case.

I thought it only fair to warn you that your earth was about to be shattered.

I will state quite plainly, this is the first time i've ever dealt with any in science who didn't lead with their results. the cooler more dramatic the results, the more that person will shove it under your nose. Normally I know i'm gonna see something exciting when a person runs into my office sayinig, "Check this out!"

Hey joozb, check this out! By the way, I like your new sign on your office door.

Is this just more magical thinking - the law of contagion in action (having gained a piece of evolutionary theory, it magically relates to the entire thing)? Or does he think he's doing the same thing we do when we use a Bible contradiction to overturn the inerrancy of the Bible?

I'd go with the voodoo hypothesis.

Yahzi, no rabbits out of my hat, Bronze Dog, no voodoo dolls, this one is mathematics with a computer model of random point mutations and natural selection written by the head of computational molecular biology at the National Cancer Institute and published in the peer reviewed journal Nucleic Acids Research. This model has been around for years and scrutinized by numerous scientists and mathematicians. The one thing nobody has done until now is a systematic parametric study. These results put the cornerstone of evolutionary theory in a severe mathematical vise. I’m not a bad applied mathematician and computer programmer and I see no way of rescuing this aspect of the theory evolution from this mathematical conundrum.

 

[T'ai Chi]

The experiments are "intelligently designed" as well. The reason to use a computer simulation is that there are some things that are much easier to do with a computer.

I'd think real-life experiments are always worth more than a computer simulation of real life. Why? Because it shows directly it could happen more realistically in real life.

If Paul's program shows something could happen in X number of generations, why not just go through X number of generations of flies and see if it actually occurs in real life?

 

[fls]

Originally Posted by Kleinman
I believe prove my case that ev shows that macroevolution by point mutation and natural selection is mathematically impossible...
Originally Posted by fls
Who says that macroevolution occurs by point mutation alone?

What mechanisms do you want to propose

I'm guessing that in addition to showing that point mutations alone will not suffice as a mechanism for macroevolution, you want to dismiss other mechanisms that are a part of current evolutionary theory such as crossing-over and inversion.

You seem to be spending an inordinate amount of time arguing something that most people may already agree with - point mutation alone does not drive evolution. Maybe you'd make more headway if you could explain why you think it is reasonable to exclude other mechanisms from your model (other than simply dismissing them).

Linda

 

[Paul C. Anagnostopoulos]

Awww man! And during a thread where the one you took out (mine) is just so freaking appropriate!

I know, I'm sorry. Three sigs just don't work. For anyone who has not seen this beautiful sig:

ID has no answers. It can only make itself look palatable by making evolution look less palatable. It lives in a cardboard refrigerator box and throws rocks through the windows of evolution's unfinished mansion. ---Beleth

 

[Paul C. Anagnostopoulos]

Surely, the point is that there is no perfect creature. By prespecifying an end point you're trying to determine the time taken and probability of reaching that end point. In the real universe, assuming ID isn't correct, there isn't an end point and evolution goes to where it goes to.

But it's an endpoint that we know has been reached in the real world. Don't be mislead by the word perfect, it's just used to mean that the creature has no binding site errors.

~~ Paul

 

[kleinman]

Annoying Creationists

What mechanisms do you want to propose

I'm guessing that in addition to showing that point mutations alone will not suffice as a mechanism for macroevolution, you want to dismiss other mechanisms that are a part of current evolutionary theory such as crossing-over and inversion.

How do you get the original gene to cross-over or invert? You must still have some mechanism to create the original gene.

You seem to be spending an inordinate amount of time arguing something that most people may already agree with - point mutation alone does not drive evolution. Maybe you'd make more headway if you could explain why you think it is reasonable to exclude other mechanisms from your model (other than simply dismissing them).

I don’t know if that is what most evolutionists agree on. Recombination may drive diversity but recombination can not drive macroevolution because recombination without errors can not add information (create new genes) to the gene pool. Recombination with natural selection can cause the loss of information (alleles) from the gene pool. Neither of the mechanisms (inversions and cross-over) that you propose can create a new gene these mechanisms require and existing gene that must be created by random point mutations and natural selection.

Since all you evoblogophiles now understand the basics of the basics of ev, I know give you your first homework assignment if you are willing to enter this realm of danger. Your first assignment is to take Dr Schneider’s baseline case (the case which he published in Nucleic Acids Research) which are the default values in evjava. Click both the Pause on Perfect Creature check box and Pause on Rs>=Rf check box. Also change the maximum number of generations for the program to run to 1,000,000. You only need to run any case the number of generations required to meet the convergence criteria. Then click the Run button. When either convergence criterion is met, record that number of generations required to meet that convergence criterion, unclick the corresponding check box, then click the Run button again and let the program run until the other convergence criterion is met. Record the generations required to meet this condition as well. Now click the New button. Change only the number of Potential sites (the genome length) from 256 to 512 and click either the Pause on Perfect Creature check box or Pause on Rs>=Rf check box so that both pause conditions are checked. Then click the ok button to return to the program execution screen. Click the Run button and repeat the process you just performed for the 256 base case for this 512 base case. Once both convergence criterion are satisfied for this case and you have recorded the generations for convergence, click the New button and double the genome length from 512 to 1024 and make sure both the Pause on Perfect Creature check box and Pause on Rs>=Rf are clicked on and repeat the process for potential sites (genome lengths) 2048, 4096, 8,192, … Report these result in the following format:
Genome length/Generations for Perfect Creature/Generations for Rs>=Rf
256/662/675
512/2412/2925
1024/***/***
2048/***/***
4096/***/***
8192/***/***
16384/***/***
I’ve given the first two cases in this little table so you can compare if you obtain the same results. Once someone posts their filled table, we can discuss the meaning of these results.

 

[Paul C. Anagnostopoulos]

I posted the results on the Evolutionisdead forum on the following page:
http://www.evolutionisdead.com/forum/viewtopic.php?t=348&postdays=0&postorder=asc&start =285
It is on my Oct 09, 5:39PM post.
That case took about 300Mb of RAM and 4 days of cpu time.

Merging your data with mine and ignoring populations below 2048, I obtain the following curve fit for a genome size of 1024 and 1 mutation / genome / generation:

[latex]$12{,}138p^{-.23}$[/latex] with r = -.92

So increased population size does help.

~~ Paul

 

[Roboramma]

If Paul's program shows something could happen in X number of generations, why not just go through X number of generations of flies and see if it actually occurs in real life?

Because to do so would be far more expensive? Because to do so would take much more time? Because you can run the program many many times with different conditions for less money and in less time than you could run your fruit fly experiment once.

Possibly (I don't know if this is the case or not) by building a model that focuses on one aspect of the process, we might get more insight into that specific aspect of the process than doing an experiment on complex living organisms where many more things can muddle our interpretation.

There might be other reasons as well. Certainly we have to be careful as to how to interpret the results of a computer model, but that doesn't mean they are useless.

 

[delphi_ote]

I'd think real-life experiments are always worth more than a computer simulation of real life. Why? Because it shows directly it could happen more realistically in real life.

If Paul's program shows something could happen in X number of generations, why not just go through X number of generations of flies and see if it actually occurs in real life?

Gee. I wonder why nobody ever thought of that.

Oh.

 

[Paul C. Anagnostopoulos]

I'd think real-life experiments are always worth more than a computer simulation of real life. Why? Because it shows directly it could happen more realistically in real life.

The purpose of Ev was to show that information could evolve, and that one can predict how much information will evolve in a particular context. The program succeeds in demonstrating this. If you think it doesn't pertain to real life, it would be helpful if you could explain what part of mutation and selection real life doesn't exhibit. Heck, if you could convince us that DNA doesn't contain information, that would pretty much do it right there.

Ev was never meant to replace fruit fly labs.

~~ Paul

 

[kleinman]

Annoying Creationists

Awww man! And during a thread where the one you took out (mine) is just so freaking appropriate!

I know, I'm sorry. Three sigs just don't work. For anyone who has not seen this beautiful sig:ID has no answers. It can only make itself look palatable by making evolution look less palatable. It lives in a cardboard refrigerator box and throws rocks through the windows of evolution's unfinished mansion. ---Beleth

Beleth, it’s not an unfinished mansion, it’s a house of cards, and I’m not throwing rocks, I throwing mathematical data. And don’t make fun of my house, it happens to be a double door refrigerator box that came with built in ice-maker. I’m planning on adding a chest freezer box room addition as soon as cardboard prices drop a little.

So increased population size does help.

Paul, I didn’t say that ev doesn’t show that population helps, what I said was that huge populations don’t markedly reduce the number of generations for convergence sufficiently to allow macroevolution to occur. In addition large populations contradict Gould’s concept of punctuated equilibrium which he said occurs in small sub-populations.

We can discuss the population effect but why don’t we give the bloggers a chance to familiarize themselves with ev so that we might have more than a two way discussion on these issues.

 

[Paul C. Anagnostopoulos]

Paul, I didn’t say that ev doesn’t show that population helps, what I said was that huge populations don’t markedly reduce the number of generations for convergence sufficiently to allow macroevolution to occur. In addition large populations contradict Gould’s concept of punctuated equilibrium which he said occurs in small sub-populations.

We don't know this, because we haven't modeled more than a measly million creatures. You won't let me extrapolate using that fitted curve, but if I did extrapolate to a lousy billion creatures, it would require 103 generations; to a trillion creatures, 21 generations. Of course, there is some asymptote it's approaching, although I haven't the slightest idea what that is.

I think that is a sufficient reduction for macroevolution to take place. Wait, what was macroevolution, again? And how small is a small population of bacteria? Hang on, does punctuated equilibrium pertain to the evolution of an ancient mechanism such as genetic binding?

~~ Paul

 

[joobz]

I really thought you had more to offer. I guess we know now.

Ev and the basics of information theory.
The Shannon definition for information turns out to be mathematically equivalent to the negative of the quantum mechanical definition for entropy. If one considers that entropy is the measure of randomness this relationship becomes intuitively apparent. Increasing the information in a system reduces the randomness and thus reduces the entropy. So how does this relate to genetic evolution. One of the basic problems of Information theory is to take an initial ensemble with an initial probability distribution to a final ensemble with a final probability distribution by the input of information. In other words, you take a more random higher entropy ensemble to a less random lower entropy ensemble by the input of information. When information theorist talk about 1 bit of information, they are saying that based on a single yes or no question, the answer to that question allows them to decide which ensemble has a lower entropy. The answer to the binary question allows you to reduce the entropy and therefore the randomness by 1 bit.

So you comment on the relationship between information theory and statistical mechanics. Your simply stating increased order(information) equals decreased entropy. So? Why does this matter to your kinetic arguement?

In the ev model, the initial ensemble is the initial random genome. In the ev model, the questioning process are the random mutations. If a random mutation hits a locus in a binding site and improves the match between the weight matrix (binding protein) and that binding site it increases the information in that binding site. So the weight matrix gives the answer. Natural selection then chooses the ensemble with the higher information (or in this model, with the fewest number of mistakes in the identification of binding sites).

Unlike many IDers who have criticized Dr Schneider’s simulation, I find his basic concept plausible. What I don’t find plausible is Dr Schneider’s use of unrealistic parameters. Dr Schneider used a genome length of only 256 bases (2000 times smaller than the smallest genome known in any free living organism) and a mutation rate of 1 mutation per 256 bases per generation (4000 times higher than the average mutation rate in prokaryotes) in his publication on ev. When more realistic parameters are used in the model, the rate of acquisition of information become profoundly slow as will unfold as the data from ev is presented. This rate of information acquisition becomes so slow that it shows that it takes hundreds of millions of generations to evolve less than 100 loci on a genome of only 100,000 bases. Point mutation and natural selection is a profoundly slow process. Far too slow to evolve a gene de novo.

That's it? Um, again, Not enough time? THere are never situations where mutation rate changes? I'm sorry, the model is interesting in it's ability to show increased information with improvements in the binding site sequence.
But you ignore, not all point mutations are created equal. Some loci are not as improtant as others (consider consensus sequences).
You ignore insertion/deletion errors
You ignore genetic exchange
You ignore the ADMITTED limitations to the code presented

I'm sorry, you claim an understanding of engineering, but haven't exercised any of the critical analysis required in the disipline.
As an engineer, you MUST know what a math model is and what it can and can not do. I don't want to believe that you are being delibertly missleading, but you are leaving me little room to think otherwise.

 

[T'ai Chi]

Because to do so would be far more expensive? Because to do so would take much more time?

Yes, it probably would be more prohibitive for various reasons. It would also be more convincing IMO, because it would deal directly with real life.

 

[T'ai Chi]

The purpose of Ev was to show that information could evolve, and that one can predict how much information will evolve in a particular context. The program succeeds in demonstrating this. If you think it doesn't pertain to real life, it would be helpful if you could explain what part of mutation and selection real life doesn't exhibit.

That's all good and stuff, but why not do the same with actual biology instead?

But what I'm really wondering is why, if you admit Ev shows information could evolve, and you admit that Ev is an intelligently designed program, that you are not admitting that it really shows that it takes intelligence to get information?