The problem with Dawkins' Weasel Program

[spinoza99]

With Weasel you have to tell the computer what the target is. Natural Selection does not know what a target is. I do admit that NS preserves beneficial changes. Remember I believe in NS, I just believe that NS is not the only mechanism driving the change in species. I compare NS to the Free Market. In the Free Market NS gets rid of the bad products. No one will continue to buy a bad product for long and ultimately its manufacturer will go out of business, or, in the animal world it will die. The New Products are built by intelligent design, not by Natural Selection. NS simply gets rid of the bad products and enables the good products to persist. Natural Selection selects, it does not design.

The problem with Dawkins' Weasel program is that one small change will be preserved. In the real world you have to coordinate anywhere from 100 to a thousand, even 27,000 amino acids in the cases of Titin, to get a change that will be preserved. So all those changes that lead up to methinks it is like a weasel will not be preserved because they are useless and NS will discard them.

Let me also point out that I believe Humans and Monkeys descend from a common ancestor.

[Thinking Aloud]

Dawkins's program merely illustrates, in a very simple way, change that is selected for, and how change over time can produce far more complex patterns than the original. It matters not the mechanism that does the selecting, be it environmental, predatory, domestication, or a user on a computer - that's not what the program is about.

Intelligent design can be seen in operation around us. Dogs, pigeons, cattle, sheep, horses - any domesticated animal shows intelligent design: shaped by selective breeding by humans. That's why the changes have happened so quickly. In nature change happens extremely slowly, and that makes me ask why, if there is a "designer" out there, it's taking so long about it.

[FBM]

Natural Selection, not being an organism, cannot know anything. As a process, it can, retrospectively and metaphorically, appear to favor certain results, the simplest of which is mere survival for the individual. This retrospective perspective can easily be depicted as a sort of knowledge of the end, but it's a mindless "knowledge" that simply reflects statistics and probability in physical processes. This selection repeated over millions of years is sufficient to explain the apparent favor allotted to reproductive success. No grand designer behind the scenes is required to explain the data.

[spinoza99]

why, if there is a "designer" out there, it's taking so long about it.

Time is relative. Atoms think in terms of Planck times which is 10^-43 seconds. So that above post, if it took you 5 minutes to write, took 3*10^46 Planck Times to write. That's a big number. But I admit that designing humans has taken a long time which is why I believe that the designer is not omnipotent. On the other hand, designing a Trilobyte might have only taken 5 million years to change it from a sponge into its form. Trilobyte's are the creatures that existed in the Cambrian, sort of the first advanced animals.

[Feck]

Atoms don't think .

[my_wan]

With Weasel you have to tell the computer what the target is. Natural Selection does not know what a target is.

Let's suppose, for the sake of argument, your definition of "target" wrt the Weasel program is correct (I'll explain why it's not later). The Weasel program has a food source, and the key to eating this food is the "target", such as "METHINKS IT IS LIKE A WEASEL". Now the offspring best at catching this food source (target), gets to live and have kids which mutate. The ones that didn't catch the available food starve and die.

So you say NS does not have a "target". Are you saying that wolves don't have a food source to catch? Or that if other wolves better at catching that food will survive famines better than the wolves that are not so good at it? Food and reproduction is the "target" of NS. It's not a "target" because NS planned it that way, it's a "target" because those that didn't randomly acquire those targeting skills are dead, and have no children.

This means it's not a target, only a result of NS. Not unlike air leaking out of a tire in previous discussions. So what about this claim that we must provide the Weasel program with a "target"? Certainly we must provide a food source, for our evolving sentence to learn to catch and eat better than the others to survive. This food is what you are calling the "target".

So does the Weasel program know what its "target", or food source is, does the algorithm need to know this target? Absolutely Not! Here's why. The algorithm works exactly the same, no matter what sentence you choose for a food source. Now compare your argument in the thread "How human language refutes atheism". You claim that, because the number of way sentences can be structured, evolution can't explain organizing that much information. Yet no matter which of these sentences is used for food, even purely random food sources which only has "meaning" as Weasel food, the Weasel program learns to catch it. So the number of sentence combinations the Weasel program can learn to catch and eat, because it can also learn to catch and eat random letters, is trillions of times larger than the number of real sentences that is possible in you language refutation thesis.

So no, the Weasel program is not "given" a "target". It is only given a "food source" that it must catch faster than its siblings to live. The only "information" about the food source the Weasel program has comes from a function called "Random()". It works the same if it has one or a million different kinds of food sources. With a million different kinds of food sources, and a large population of Weasels, then different populations of Weasels will evolve eat different kinds of food. Some will even go after more than one kind of food, so long as that secondary food doesn't have too much competition from a population better at catching that food. The different populations are like different species.

The problem with Dawkins' Weasel program is that one small change will be preserved.

No, this is a common false claim. The one I wrote was just as likely to mutate a good letter to a bad one as it was to mutate a bad letter to a good one. Even more like for a good letter mutation to mutate to bad one, than the other way around, because there are a lot more bad letter for the "food source" available, and only one good letter available. Yet it still got it perfectly in under a minute.

I had all upper an lower case letter, four different punctuations, and a mutable blank space. That's a 1 out of 109 chance of any one "Random()" mutation getting a letter right. That's also a 108 to 1 chance that when a correct letter mutated it would mutate to a bad letter, thus getting worse rather than better at eating the food. Yet no matter what "food source" you provided, it still got that food (sentence) mastered in under a minute.

No good changes are "preserved" to make the program work.

In the real world you have to coordinate anywhere from 100 to a thousand, even 27,000 amino acids in the cases of Titin, to get a change that will be preserved.

Yes, and a 4 * 1000 = 4000 letter sentence would get mastered by the Weasel program in no time. I suspect it would be quiet a wait for 4 * 27,000, but hey, evolution took billions of years. I don't think waiting a few weeks is meaningful up against a few billion years

So all those changes that lead up to methinks it is like a weasel will not be preserved because they are useless and NS will discard them.

Download AutoIt, it's a free scripting language, not limited in any way, and easy to learn. Learn to write some basic genetic algorithms, so you can be sure that it actually works using the Random() function *without* preserving good mutations. You see, no amount of debating will convince you of anything, till you take the time to actually test these claims yourself. I wrote my Weasel in part to test the difference between preserving and not preserving good mutation. With a 109 character sets not preserving good mutations barely made any difference in speed at all.

It makes it fairly obvious why, after writing and testing the programs, but explaining it in words is even more complicated than the statistical mechanics of randomness from the previous thread. Learn to do it yourself, and test these claims you make, because they are so easily testable. Then see what kind of rebuttal you can come up with.

Let me also point out that I believe Humans and Monkeys descend from a common ancestor.

Good. We have solid evidence of that even without a theory of evolution, as a result of telomeres in the chromosomes. We also have equally good (different) evidence that mice and humans had a common ancestor, via endogenous retrovirus. The anti-evolution hypothesis need not find a rabbit in the Cambrian to make their case. It's actually much easier. Just find a gene sequence with a signature across species that matches a variance in a species that came later than the variance. If a rabbit in the Cambrian could be found then these variances, that are genetic versions of rabbits before rabbits should exist, should be all over the place in cross species genetic variances, but their not. Each and every cross species gene variance follows only from the exact same genetic tree, and variances from one limb will not be found in another limb unless it existed before the limbs branched into different species. There are literally thousands of different ways to statistically test the same exact thing, all with the same result pointing to the exact same tree of speciation. The same tree constructed completely seperately from the fossile record.

[FBM]

my_wan, from the nosebleed section:

[Gawdzilla Sama]

Why do I get the feeling that all those unanswered questions in the last thread will now be promptly, thoroughly, and completely ignored?

[spinoza99]

So you say NS does not have a "target". Are you saying that wolves don't have a food source to catch?

Wolves have to catch food but NS can not intentionally mutate amino acids so as to make the species better at catching food. Let's say that if your DNA spells the sentence: methinks it is a weasel, it will enable the species to catch food better. NS cannot know that that is a correct sentence ahead of time.

Certainly we must provide a food source, for our evolving sentence to learn to catch and eat better than the others to survive. This food is what you are calling the "target".

Clearly there are beneficial mutations and sequences, but we cannot tell NS what they are. For example, let's say we have an organism with 200 genes and each gene is 100 amino acids long. With 100 amino acids that's about 10^120 possible sequences (but I forget the exact number), but let's cite Douglas Axe's work who suggested that of the 10^120 sequences only 10^46 are useful which makes the odds of getting a possible sequence one in 10^74 (since you just subtract the number of correct sequences from the total number of sequences)(I haven't read Axe's actual paper because the scientific community ridiculously charges exorbitant rates for these papers). You then multiply 10^74 by 200 genes and get about 10^14000, even though it's more than that because you have left-handed and right handed amino acids (though I think there are some organisms that form using right handed amino acids and there are also peptide bonds that do not link amino acids). So, yes, hidden within those 10^14000 possible combinations of genes there are winning combinations but NS does not know what they are.

You claim that, because the number of way sentences can be structured, evolution can't explain organizing that much information. Yet no matter which of these sentences is used for food, even purely random food sources which only has "meaning" as Weasel food, the Weasel program learns to catch it. So the number of sentence combinations the Weasel program can learn to catch and eat, because it can also learn to catch and eat random letters, is trillions of times larger than the number of real sentences that is possible in you language refutation thesis.

I don't know exactly what you're saying. If you could restate that would help but it seems like what you're saying is that: the strong survive, the mediocre survive and, yes, the blatantly handicapped survive as well. That's not true. Some amino acid combinations result in the immediate destruction of the organism and if not that then certainly its inability to reproduce.

The only "information" about the food source the Weasel program has comes from a function called "Random()".

Not true. You have also told the Weasel what a correct sentence is.

It works the same if it has one or a million different kinds of food sources. With a million different kinds of food sources, and a large population of Weasels, then different populations of Weasels will evolve eat different kinds of food. Some will even go after more than one kind of food, so long as that secondary food doesn't have too much competition from a population better at catching that food. The different populations are like different species.

10^6 is nothing compared to 10^150. I'm sure you have excel. Excel will do the math for you, just put in excel =20^100, that will give you how many combinations of sequences there are in a 100 amino acid protein, to say nothing of the 27,000 long amino acid, titin.

The problem with Dawkins' Weasel program is that one small change will be preserved.

No, this is a common false claim. The one I wrote was just as likely to mutate a good letter to a bad one as it was to mutate a bad letter to a good one. Even more like for a good letter mutation to mutate to bad one, than the other way around, because there are a lot more bad letter for the "food source" available, and only one good letter available. Yet it still got it perfectly in under a minute.
I had all upper an lower case letter, four different punctuations, and a mutable blank space. That's a 1 out of 109 chance of any one "Random()" mutation getting a letter right. That's also a 108 to 1 chance that when a correct letter mutated it would mutate to a bad letter, thus getting worse rather than better at eating the food. Yet no matter what "food source" you provided, it still got that food (sentence) mastered in under a minute.

If the odds of getting a beneficial mutation were 1 in 108 perhaps NS might be plausible even though that would not explain the fact that the properties of objects are coordinated. But the odds of a beneficial mutation is not one in 108, it's well beyond one in 10^10,000. Moreover, there are only 10^80 atoms in the universe and 10^60 Planck Times in our universe's history, so there is not enough time to hit the lottery.

In the real world you have to coordinate anywhere from 100 to a thousand, even 27,000 amino acids in the cases of Titin, to get a change that will be preserved.

Yes, and a 4 * 1000 = 4000 letter sentence would get mastered by the Weasel program in no time. I suspect it would be quiet a wait for 4 * 27,000, but hey, evolution took billions of years. I don't think waiting a few weeks is meaningful up against a few billion years.

You're doing the math wrong. There are probably a minimum of 200 genes needed for a one celled organism, even though Sean Carroll (a passionate Darwinist) says that the minimum number found so far is about 1560 (I thought it was 400). In order to calculate the odds of a beneficial mutation you have to compare homologous genes across species. They range in similarity from 60% all the way up to 98% (see attachment). So you take a 150 amino acid protein and you figure that maybe, if you're lucky only 60% of those sequences are important, though clearly there are some sequences in the other 40% that result in the being's immediate death. You then take 20 and raise it the 90th power and that's roughly the odds of forming ONE successful protein.

By the way, how did you come up with the idea that the odds of getting a beneficial mutation are one in 108? I noticed Calli thinks that too, clearly you two are reading the same source.

You see, no amount of debating will convince you of anything, till you take the time to actually test these claims yourself.

Such statements cannot be proven and hence are useless in this debate. By the way, I could make the same unprovable statement about you, but I know better than to make statements that can not be supported with reason.

The anti-evolution hypothesis need not find a rabbit in the Cambrian to make their case. It's actually much easier. Just find a gene sequence with a signature across species that matches a variance in a species that came later than the variance.

Finding homologous genes in different species does not prove that those genes were built by NS and random mutation. Darwinists never cease to make this mistake.

This is the ultimate Darwinian fallacy:

Common ancestry is true therefore Natural Selection through random mutation is the only mechanism by which species change.

[Feck]

And what about the canard 'It can't have happened by random ...the odds of a human genome happening by random chance are too high ' ?
Please don't talk about atoms thinking and 911 number theory and then say things like ' but I know better than to make statements that can not be supported with reason. '

[charlou]

The New Products are built by intelligent design, not by Natural Selection.

Quite a leap and quite a claim ... and you're not the first to make it. Yet to see evidence of the requisite designer, let alone actual design, though.

Methinks you have your understanding arse-backwards, approaching it as you are from a default position of belief, rather than falsifiable facts.

[my_wan]

Wolves have to catch food but NS can not intentionally mutate amino acids so as to make the species better at catching food. Let's say that if your DNA spells the sentence: methinks it is a weasel, it will enable the species to catch food better. NS cannot know that that is a correct sentence ahead of time.

NS doesn't "intentionally" mutate amino acids to make the species better at catching food. The Weasel program doesn't "intentionally" mutate letters to make it closer to the food sentence. Mutations happen randomly. The Weasel uses the Random() function that says this letter has a % chance to mutate this letter to one of 109 letters, characters, or blank spaces, even if the letter is already the correct one and it makes the letter worse. DNA mutates the same way, without "intentionality" just like the Random() has no "intention", but some of those DNA mutations will increase and hurt the offspring ability to catch food, just like Random() did for Weasel.

Clearly there are beneficial mutations and sequences, but we cannot tell NS what they are.

How does NS 'decide' which mutations was the right ones? Because the wrong mutations died. How did Weasel know which ones were the right ones? Because the wrong mutations died. We did not tell NS what was food because we didn't have to put the food their for it. We had to put the food there for Weasel, and your pretending that's how we "told" Weasel what to do. But the food is already there in NS, and does the same thing.

For example, let's say we have an organism with 200 genes and each gene is 100 amino acids long. With 100 amino acids that's about 10^120 possible sequences (but I forget the exact number), but let's cite Douglas Axe's work who suggested that of the 10^120 sequences only 10^46 are useful which makes the odds of getting a possible sequence one in 10^74 (since you just subtract the number of correct sequences from the total number of sequences)(I haven't read Axe's actual paper because the scientific community ridiculously charges exorbitant rates for these papers). You then multiply 10^74 by 200 genes and get about 10^14000, even though it's more than that because you have left-handed and right handed amino acids (though I think there are some organisms that form using right handed amino acids and there are also peptide bonds that do not link amino acids). So, yes, hidden within those 10^14000 possible combinations of genes there are winning combinations but NS does not know what they are.

Did you not understand when I explained that the Weasel program can sort through even more food combinations than that without changing the program? I did not even put the food sentence in the program at all. I put it in a separate file for the Weasel program to try to eat. So I could change food sources as often as I wanted without changing a single byte of the Weasel program. The Weasel program is "Universal" for *all possible* alphabetic food sources.

I don't know exactly what you're saying. If you could restate that would help but it seems like what you're saying is that: the strong survive, the mediocre survive and, yes, the blatantly handicapped survive as well. That's not true. Some amino acid combinations result in the immediate destruction of the organism and if not that then certainly its inability to reproduce.

Yes. I just did when I explained above the the Weasel program is "Universal" for *all possible* alphabetic food sources, without any changes in the Weasel code at all. Yes, less than 30% of human pregnancies ever result in a birth. Most don't even last long enough for woman to realize they were pregnant.

Not true. You have also told the Weasel what a correct sentence is.

I'll try to explain this again. The correct sentence is *not* written into the Weasel program. The correct sentence is entirely separate in a different text file, and is nothing more than a food source. You do have to feed your pets to, if you expect them to stay alive. You can change this food source to any alphabetic food source, a variety far larger than your 10^14000, and it'll still evolve to catch that food source. I even used random sentences 500 letters long, and it still worked.

10^6 is nothing compared to 10^150. I'm sure you have excel. Excel will do the math for you, just put in excel =20^100, that will give you how many combinations of sequences there are in a 100 amino acid protein, to say nothing of the 27,000 long amino acid, titin.

So I'll say it this way: It works the same whether it's one or 10^1,000,000. In fact even that is small compared to the number of different food sources (sentences) a Weasel program can evolve to use, without any modification to the Weasel program.

If the odds of getting a beneficial mutation were 1 in 108 perhaps NS might be plausible even though that would not explain the fact that the properties of objects are coordinated. But the odds of a beneficial mutation is not one in 108, it's well beyond one in 10^10,000. Moreover, there are only 10^80 atoms in the universe and 10^60 Planck Times in our universe's history, so there is not enough time to hit the lottery.

I can change the odds of a beneficial mutation were 1 in 1000 and it still works. In fact I did, as it was only a variable in the code, only the one I released on RD.net defaulted to a 109 character set.

And here's where you really need to pay attention and learn. The properties are *NOT* "coordinated". They are defined by Random(), in both Weasel and NS. It only looks coordinated because the failures died.

Suppose every person born in the last 5000 years lived, and all their kids lived. The Earth would be so full of people you couldn't find a place to put your foot on the ground that didn't already have somebodies foot there. Imagine if England expanded its per capita population as much as the US did after we colonized here. England would have to be a large multistory apartment complex to house them. This means most of us will not successfully have descendent's in the future, just as most from back then don't. The few who do are the few with the successful mutations that gave them the advantage. Human evolution is as strong now as ever.

You're doing the math wrong. There are probably a minimum of 200 genes needed for a one celled organism, even though Sean Carroll (a passionate Darwinist) says that the minimum number found so far is about 1560 (I thought it was 400). In order to calculate the odds of a beneficial mutation you have to compare homologous genes across species. They range in similarity from 60% all the way up to 98% (see attachment). So you take a 150 amino acid protein and you figure that maybe, if you're lucky only 60% of those sequences are important, though clearly there are some sequences in the other 40% that result in the being's immediate death. You then take 20 and raise it the 90th power and that's roughly the odds of forming ONE successful protein.

First off, my numbers were Based on the Weasel program, and only intentional made harder than NS odds to illustrate 20^90 or more is no big deal for evolution. At the DNA level there are only 4 letters, while Weasel used 109. Your 10^90 odds is pretending that a sentence (protein) which has a 1 in 20^90 chance of happening with one Random() selection can't happen. The sentences that Weasel can evolve to have far smaller odds than that of a full complete sentence.

So does NS need to create a completely functional protein just to get started? No. Look at the Nylon bug, which can *only* eat nylon which didn't exist before 1935. It is only 2% efficient at eating nylon. So as the mutations continue to occur, and improve the proteins (like a sentence for the Weasel program), it'll get better at eating nylon. When it does, those that are only 2% efficient will get out competed and die. Just like the Weasel program.

NS doesn't have to come up with a single perfect protein to work. There are billions of different proteins that can do the same job, some better, some not so good, as the one an organism presently uses to do that same job.

By the way, how did you come up with the idea that the odds of getting a beneficial mutation are one in 108? I noticed Calli thinks that too, clearly you two are reading the same source.

Don't know Calli's numbers. My numbers came from the number of different characters the Weasel program I wrote used, but I got the release version wrong. Not sure why I said 108. The release version defaulted to 57. You could change it to 500, a thousand, or any odds you wanted though. It's just the number character to 1. 26 upper case letters, 26 lower case letters, 1 blank space, 1 period, 1 question mark, 1 apostrophe, and 1 comma.

I don't think it's possible to define exactly in NS. If a mutation is harmful, it may still be beneficial by giving a resistance to a new disease. What is harmful in one organism and/or environment may be good in another organism and/or environment. If nylon hadn't been invented by people in 1935, the mutation that created the nylon bug would have been deadly (instantly extinct), because that bug can no longer eat carbohydrates. However, since it found the nylon we invented, it has an environment with no food competition. So it thrives even though it is only 2% efficient.

You can only define such odds if you know every possible combination, every possible function of those combinations, and every possible chemical in existence that is available for food. Your own numbers falsely pretended that one protein, and that one version of that one protein, was needed for that one protein function to work at all. Simply false.

You see, no amount of debating will convince you of anything, till you take the time to actually test these claims yourself.

Such statements cannot be proven and hence are useless in this debate. By the way, I could make the same unprovable statement about you, but I know better than to make statements that can not be supported with reason.

So are you are saying that since I can't prove that if you learned to make genetic algorithms yourself, as see for yourself that NS actually works as claimed, that it would convince you, so there's no sense in you seeing it work for yourself? I'll readily admit that if you find some new claim or argument, which I haven't already tested, that I'll need to test it to be convinced. I'll do that on my own, and just thank you for being bright enough to suggest something new, without requiring you to prove it to me. That's all I'm asking of you, to test you own ideas and see if you can prove yourself wrong, as I have proven myself wrong many many times.

Even had to publicly eat my word on some rare occasions, and will do the same for you if you can provide the justification.

Finding homologous genes in different species does not prove that those genes were built by NS and random mutation. Darwinists never cease to make this mistake.

This is the ultimate Darwinian fallacy:

Common ancestry is true therefore Natural Selection through random mutation is the only mechanism by which species change.

It's not the homologous genes that tell us much, it is precisely how inhomologous the genes are that gives us a tree structure. But if the amount of variation in one gene across species was all we had it would be meaningless. If the amount of variation in all the genes varied by a significantly different amount between the same pair of species, evolution would be dead. So what kind of outrageous accident does it take for every gene variation between every pair of species on the face of Earth to match the exact same tree structure?

So no!!! It's not true just because it is a mechanism that explains evolution. It is true because if it wasn't true then each and every gene in a given species would not produce the exact same tree of life as any other gene in that species. Every part of the gene sequence shows the same tree as every other part, which is the same tree as the whole sequence shows, which is the same tree as was reconstructed from the fossil record alone, is the same tree endogenous retroviruses show, etc., etc. And all you have to do to show otherwise is find a gene sequence anywhere, in any species on Earth, that does not show its evolutionary path in that *same* tree.

The same time scale carbon (etc) dating gave us in the fossil record, is the same time scale from mutation rates, which we directly measure in the laboratory without using carbon dating. We sequence the genes of a species, wait so many generations, sequence again and see how many mutations occurred. We repeat this over and over again on many different species and time scales. We get a general mutation rate, without carbon dating, apply it to the variations in the tree, and lo and behold, it matches what we got from carbon dating.

The ideas Intelligent Design is based on has been used by real scientist to try to falsify evolution for years. Only problem is evolution keeps winning in every test. Hence that claim is the ultimate fallacy of the anti-evolution crowd. That is why I keep encouraging you to figure out how to test these claims yourself, instead of thinking debating them is a meaningful way to learn.

[spinoza99]

NS doesn't "intentionally" mutate amino acids to make the species better at catching food. The Weasel program doesn't "intentionally" mutate letters to make it closer to the food sentence. Mutations happen randomly.

correct

How does NS 'decide' which mutations was the right ones? Because the wrong mutations died. How did Weasel know which ones were the right ones? Because the wrong mutations died. We did not tell NS what was food because we didn't have to put the food their for it.

correct

For example, let's say we have an organism with 200 genes and each gene is 100 amino acids long. With 100 amino acids that's about 10^120 possible sequences (but I forget the exact number), but let's cite Douglas Axe's work who suggested that of the 10^120 sequences only 10^46 are useful which makes the odds of getting a possible sequence one in 10^74 (since you just subtract the number of correct sequences from the total number of sequences)(I haven't read Axe's actual paper because the scientific community ridiculously charges exorbitant rates for these papers). You then multiply 10^74 by 200 genes and get about 10^14000, even though it's more than that because you have left-handed and right handed amino acids (though I think there are some organisms that form using right handed amino acids and there are also peptide bonds that do not link amino acids). So, yes, hidden within those 10^14000 possible combinations of genes there are winning combinations but NS does not know what they are.

Did you not understand when I explained that the Weasel program can sort through even more food combinations than that without changing the program? I did not even put the food sentence in the program at all. I put it in a separate file for the Weasel program to try to eat. So I could change food sources as often as I wanted without changing a single byte of the Weasel program. The Weasel program is "Universal" for *all possible* alphabetic food sources.

You're not reading what I write. You made no argument to disprove that I used wrong math. Your whole program lies on the assumption that the odds of landing a beneficial mutation is 1 in 109 which is outrageously far from reality. The real odds are well beyond one in 10^10000. Carl Sagan, who was no nitwit, put the odds at one in 10^40,000, however, I have to point out that in spite of this he considered himself an agnostic. And no, my friend, you can't just say: we have all the time in the world because we don't have all the time in the world. We have only had 10^60 Planck times since the Universe began (a Planck time is 10^-43 seconds) and we only have 10^80 atoms in the universe. So even if every atom tried to make a successful sequence during every Planck time the odds would still be (using Sagan's odds) one in 10^40,000 minus 10^140 which is one in 10^39860.

You can change this food source to any alphabetic food source, a variety far larger than your 10^14000, and it'll still evolve to catch that food source. I even used random sentences 500 letters long, and it still worked.

I don't think you're understanding what's necessary for these computer programs. What the Dawkins program does is during each period each character chooses a new letter and if a correct letter is chosen that letter is preserved. So basically the Dawkins' program in each period has at most a 1 in 26 chance of finding a beneficial mutation, which is nothing compared to reality.

So I'll say it this way: It works the same whether it's one or 10^1,000,000.

I don't see how you can seriously believe that. If you roll a dice with 10^1,000,000 sides for each second, you will have to roll it 10^1,000,000 times before the odds are 1 that you will get the right number. Seeing as there are only 10^17 seconds in our universe, it is impossible that you could score the right number. If you really believe this then I'm going to have to stop debating with you.

I can change the odds of a beneficial mutation were 1 in 1000 and it still works. In fact I did, as it was only a variable in the code, only the one I released on RD.net defaulted to a 109 character set.

Try one in 10^150 and let me know if it works.

Suppose every person born in the last 5000 years lived, and all their kids lived. The Earth would be so full of people you couldn't find a place to put your foot on the ground that didn't already have somebodies foot there. Imagine if England expanded its per capita population as much as the US did after we colonized here. England would have to be a large multistory apartment complex to house them. This means most of us will not successfully have descendent's in the future, just as most from back then don't. The few who do are the few with the successful mutations that gave them the advantage. Human evolution is as strong now as ever.

Since the beginning of earth's history there have been about 10^40 individual bacteria. That's nothing compared to the number of possible sequences in a protein made of 150 amino acids.

20^90 or more is no big deal for evolution.

See above.

At the DNA level there are only 4 letters, while Weasel used 109. Your 10^90 odds is pretending that a sentence (protein) which has a 1 in 20^90 chance of happening with one Random() selection can't happen. The sentences that Weasel can evolve to have far smaller odds than that of a full complete sentence.

Weasel can do whatever it wants, life can't. The smallest bacteria have 1600 genes. Perhaps there are simpler life forms out there but to believe that there are is to believe in something for which there is no evidence. Maybe there's something out there with 800 genes but probably not less than that.

So does NS need to create a completely functional protein just to get started? No. Look at the Nylon bug, which can *only* eat nylon which didn't exist before 1935. It is only 2% efficient at eating nylon. So as the mutations continue to occur, and improve the proteins (like a sentence for the Weasel program), it'll get better at eating nylon. When it does, those that are only 2% efficient will get out competed and die. Just like the Weasel program.

I looked up the Nylon bug in wikipedia and it's true that it gained this ability by making one mutation on one gene. This is the standard Darwinian fallacy:

random mutation can cause big changes through mutating one position on one already-built gene, therefore random mutation can design a gene of 200 amino acids.

NS doesn't have to come up with a single perfect protein to work. There are billions of different proteins that can do the same job, some better, some not so good, as the one an organism presently uses to do that same job.

You read Sean Carroll, he said the homologous genes are on average 60% identical. That's 60 necessary amino acids on a sequence of 100 amino acids (and there are plenty of genes made of a 1000 amino acids). That's a one in 10^78 chance of it forming through chance. Moreover, proteins often work in teams of about 5 to 10. The number of proteins needed to perform translation and transcription, I think is around a 100 (I can get the source if you want it).

You can only define such odds if you know every possible combination, every possible function of those combinations, and every possible chemical in existence that is available for food.

This is true. After all there are 10^150 sequences and the only way to find out how many of those sequences are correct is to test one individually which is impossible. But it's reasonable to conclude that when you see the same genes doing the same thing in different animals and they are on average 60% identical that they must be a certain way to do a certain job.

It's not the homologous genes that tell us much, it is precisely how inhomologous the genes are that gives us a tree structure. But if the amount of variation in one gene across species was all we had it would be meaningless. If the amount of variation in all the genes varied by a significantly different amount between the same pair of species, evolution would be dead. So what kind of outrageous accident does it take for every gene variation between every pair of species on the face of Earth to match the exact same tree structure? So no!!! It's not true just because it is a mechanism that explains evolution. It is true because if it wasn't true then each and every gene in a given species would not produce the exact same tree of life as any other gene in that species. Every part of the gene sequence shows the same tree as every other part, which is the same tree as the whole sequence shows, which is the same tree as was reconstructed from the fossil record alone, is the same tree endogenous retroviruses show, etc., etc. And all you have to do to show otherwise is find a gene sequence anywhere, in any species on Earth, that does not show its evolutionary path in that *same* tree. The same time scale carbon (etc) dating gave us in the fossil record, is the same time scale from mutation rates, which we directly measure in the laboratory without using carbon dating. We sequence the genes of a species, wait so many generations, sequence again and see how many mutations occurred. We repeat this over and over again on many different species and time scales. We get a general mutation rate, without carbon dating, apply it to the variations in the tree, and lo and behold, it matches what we got from carbon dating.

Did you read what I wrote? I wrote just because common ancestry is true does not mean NS is the only mechanism that changes species. I believe in common ancestry. So basically you're still committing the classic Darwinian non-sequitur: common ancestry is true, therefore NS is true.

The ideas Intelligent Design is based on has been used by real scientist to try to falsify evolution for years. Only problem is evolution keeps winning in every test. Hence that claim is the ultimate fallacy of the anti-evolution crowd.

No, evolution has not been passing every test, common ancestry has been passing every test. common ancestry, by the way, doesn't matter, you can be an atheist and believe that all animals go back to 10 organisms, it doesn't matter. The only thing Natural Selection through random mutation can do is preserve species like the nylon bug which are the result of one mutation. Natural Selection has not been shown that it can build complete genes. You need intelligence to produce information.

[GenesForLife]

My dear chap, selection acts a posteriori on mutations.

Of course, at this juncture, it will be prudent to ask "When does selection begin to act?" Joyce et al investigated this in self replicating ribozymes and arrived at results that demonstrate that as long as differential survival/replicability is present with competition for resources selection will begin to alter the nature of the population wrt the members of said population.

Computer control of Darwinian evolution has been demonstrated by propagating a population of RNA enzymes in a microfluidic device. The RNA population was challenged to catalyze the ligation of an oligonucleotide substrate under conditions of progressively lower substrate concentrations. A microchip-based serial dilution circuit automated an exponential growth phase followed by a 10-fold dilution, which was repeated for 500 log-growth iterations. Evolution was observed in real time as the population adapted and achieved progressively faster growth rates over time. The final evolved enzyme contained a set of 11 mutations that conferred a 90-fold improvement in substrate utilization, coinciding with the applied selective pressure. This system reduces evolution to a microfluidic algorithm, allowing the experimenter to observe and manipulate adaptation.

Full paper at http://www.plosbiology.org/article/info ... io.0060085


Selection does not cause mutations, it eliminates detrimental mutants from the population.

[my_wan]

You're not reading what I write. You made no argument to disprove that I used wrong math. Your whole program lies on the assumption that the odds of landing a beneficial mutation is 1 in 109 which is outrageously far from reality. The real odds are well beyond one in 10^10000. Carl Sagan, who was no nitwit, put the odds at one in 10^40,000, however, I have to point out that in spite of this he considered himself an agnostic. And no, my friend, you can't just say: we have all the time in the world because we don't have all the time in the world. We have only had 10^60 Planck times since the Universe began (a Planck time is 10^-43 seconds) and we only have 10^80 atoms in the universe. So even if every atom tried to make a successful sequence during every Planck time the odds would still be (using Sagan's odds) one in 10^40,000 minus 10^140 which is one in 10^39860.

Let me us the same math you did using the Weasel program. The 10^40,000 odds you are quoting came from "Evolution From Space" by Fred Hoyle and N.C. Wickramasinghe (page 24). The 10^40,000 is *NOT* the odds of a beneficial mutation, it the odds of a twenty-amino-acid polypeptide chain, giving the odds of 10^20. Then, because an organism has about two thousand enzymes, or (10^20)^2000 = 10^40,000. So this is the odds of getting the *entire* organism, with every letter precisely predefined, from one single mutation!!!

This is exactly the same thing as claiming that I must get the *entire* sentence in the Weasel program with a single random mutation!!! That exactly the same thing as claiming a person came into existence from a chemical *accident* in a mud puddle one day!!!! Yet we know that some people have more beneficial mutations than others, so the odds of a beneficial mutation is not the same as the odds of getting a *whole* person from a single mutation.

Now let's do the same numbers for Weasel. The Weasel program actually defaulted to a 57 (not 109) letter code. DNA has a 4 letter code, A, G, T, and C. See the difference already? Now the *whole* sentence is the polypeptide chain. For "Methinks it is like a Weasel." is a 29 character polypeptide sequence. Hence the odds, as you defined and attributed to Sagan, is 57^29. Lets round this down to 10^29 just to match the NS number base. Yet already 10^29 is already larger than 10^20 from the (10^20)^2000 number. So, if Weasel can do this 2000 times in a row, then the odds are (10^29)^2000 = 10^58,000, which is much bigger than 10^40,000.

Get this straight, 10^40,000 was *not* the odds of a single beneficial mutation, 10^40,000 is the odds of an *entire* organism, containing 2000 twenty-amino-acid polypeptide chains, from a single mutation!!! A single mutation of A, G, T , or C doesn't even always change the function of one-amino-acid polypeptide chain, much less 20*2000 of them!!!

I don't think you're understanding what's necessary for these computer programs. What the Dawkins program does is during each period each character chooses a new letter and if a correct letter is chosen that letter is preserved. So basically the Dawkins' program in each period has at most a 1 in 26 chance of finding a beneficial mutation, which is nothing compared to reality.

I understand perfectly, I wrote the program, with much higher odds constraints than Dawkins original. Yes his original was 1 in 26, but NS uses a 4 letter alphabet, not a 10^40,000 letter alphabet that number from above tried to pretend. "Reality" is that NS has a 4 letter alphabet, and each beneficial mutation doesn't have to randomly shuffle every letter to get a new person in order to get a beneficial mutation. Even a single letter change (1 in 4, not 1 in 26 like Weasel) can sometimes result in a beneficial mutation.

I don't see how you can seriously believe that. If you roll a dice with 10^1,000,000 sides for each second, you will have to roll it 10^1,000,000 times before the odds are 1 that you will get the right number. Seeing as there are only 10^17 seconds in our universe, it is impossible that you could score the right number. If you really believe this then I'm going to have to stop debating with you.

Easy, just change the length of the sentence in the Weasel program and do it X number of times, to get those odds. Recall how I explained that a beneficial mutation can be defined by a single letter change in a 4 letter DNA code, giving the odds of a mutation occurring there a 1 in 4 chance of getting the best beneficial mutation for that location.

The nylon bug was a frame shift mutation, which added 3 letter to the beginning of the sequence. There's a 1 in 64 chance that a 3 letter codon will have a particular sequence. The bug now eats nylon, which didn't exist before 1935 and can only eat nylon and nothing else. It only has a 2% efficiency, but it doesn't matter because no other organism is competing for the nylon food.

Try one in 10^150 and let me know if it works.

You still need to understand that 10^40,000 is the *whole* organism at once, not a single beneficial mutation.

Since the beginning of earth's history there have been about 10^40 individual bacteria. That's nothing compared to the number of possible sequences in a protein made of 150 amino acids.

Yet the odds of the *whole* sentence "Methinks it is a weasel." is one in 5.7x10^58,000. The sources you are using is *pretending* that a single letter change in a 4 letter code in NS cannot be a beneficial mutation, it can. Those sources are pretending that the entire sentence, "Methinks it is a weasel.", is needed to make a single beneficial mutation, it doesn't. Only a single letter change in NS, that only contains 4 possible letters, is needed to make a beneficial mutation. We don't know which ones are beneficial and not, but it's still a 1 in 4 chance of getting the best possible mutation, not 1 in 26 per Dawkins, or 1 in 57 per my version.

See above.

See above.

Weasel can do whatever it wants, life can't. The smallest bacteria have 1600 genes. Perhaps there are simpler life forms out there but to believe that there are is to believe in something for which there is no evidence. Maybe there's something out there with 800 genes but probably not less than that.

Life can, as I illustrated above that any given DNA (real life) mutation has a 1 in 4 chance of being the best possible mutation for that base letter. As a matter of fact the smallest life I know of is φX174, and has 11 genes. It's a virus that infects bacteria. You said smallest bacteria, but start with bacteria, except to make it sound like life requires more complexity than it does?

No evidence you say? The lab experiments are showing that trying "less" complex approaches are more successful.
http://www.wired.com/wiredscience/2009/ ... cleotides/
Here's we create something close to life. Though it doesn't even contain cells, it does self-replicate, compete, and evolves. In fact, while watching it evolve, the newer better version started making the originals go extinct.
http://www.livescience.com/strangenews/ ... -life.html

I looked up the Nylon bug in wikipedia and it's true that it gained this ability by making one mutation on one gene. This is the standard Darwinian fallacy:

random mutation can cause big changes through mutating one position on one already-built gene, therefore random mutation can design a gene of 200 amino acids.

This is getting a bit silly. This is like saying Weasel is a fallacy, because a random mutation of a single letter, thus it's only claimed that it can produce the whole sentence of 200 amino acids. It only takes one letter change at a time, with only 4 possible letter, for a person to evolve from an amoeba. That's a 1 in 4 chance each mutation takes it in the right direction to get a human. Yet there are other creatures, perhaps as much or more intelligent than us, which could have evolved if the mutations happened differently.

You read Sean Carroll, he said the homologous genes are on average 60% identical. That's 60 necessary amino acids on a sequence of 100 amino acids (and there are plenty of genes made of a 1000 amino acids). That's a one in 10^78 chance of it forming through chance. Moreover, proteins often work in teams of about 5 to 10. The number of proteins needed to perform translation and transcription, I think is around a 100 (I can get the source if you want it).

That's an overall average, but it homologous genes range from a few percent to nearly 100%. If your gene X is 90% homologous with gene X in species A, then your gene Y will be very near 90% homologous with gene Y in species A. No exceptions. It is exceptions like that you need to disprove evolution. Yet you can't do it. This 60% is only the averaging together of homologous genes between us and mosquitoes (38%), us and dogs (81%), and every other possible combination. It's meaningless. It's like the average amount of evolutionary change in 4.5 billion years, and pretending that the odds of that difference requires one single mutation to go from any one species to any other species in a single mutation. Bullshit.

This is true. After all there are 10^150 sequences and the only way to find out how many of those sequences are correct is to test one individually which is impossible. But it's reasonable to conclude that when you see the same genes doing the same thing in different animals and they are on average 60% identical that they must be a certain way to do a certain job.

The key to the 60% average homologous genes is that between us and monkeys it is never 60% homologous, it's always more. Between us and mosquitoes it's never 60%, it's always less. This is how we know what the tree of life looks like, and the tree looks exactly the same if we compare any one of the genes, or another gene in the same pair of organisms. or the homologousness of the entire organism.

Did you read what I wrote? I wrote just because common ancestry is true does not mean NS is the only mechanism that changes species. I believe in common ancestry. So basically you're still committing the classic Darwinian non-sequitur: common ancestry is true, therefore NS is true.

I was explaining about homologous genes like above. We know about other mechanisms that can lead to effects in the offsprings (epigenetics), which is not a change in the DNA code. Only a change in code expression. However, you cannot get speciation from it, or a tree of life. The only way to get the tree of life is to let the DNA code randomly mutate, and let the bad changes die. Yet if you have a living organism, then any given random mutation in a 4 letter code, A, G, T, and C, you have at least a 25% chance of getting the best possible per letter mutation. Not the 1 in 26 per Dawkins, or 1 in 57 per my version.

No, evolution has not been passing every test, common ancestry has been passing every test. common ancestry, by the way, doesn't matter, you can be an atheist and believe that all animals go back to 10 organisms, it doesn't matter. The only thing Natural Selection through random mutation can do is preserve species like the nylon bug which are the result of one mutation. Natural Selection has not been shown that it can build complete genes. You need intelligence to produce information.

Yes, common ancestry is distinct, but common ancestry can't exist without evolution. Even if there was 10 different ancestors, evolution would still require that you have a common ancestry with at least part of the animal kingdom.

So, what evidence do we have for just evolution, irrespective of common descent? We'll we have genetic algorithms that us the "Random()" function to illustrate how the 4 letter DNA code varies. We have accidental gene fusions and insertion. The nylon bug was a frame shift insertion, so that the functionality of every gene in the DNA sequence was rewritten in a single accident. Algorithms need more evidence though, so we have been tracking the step by step evolution of bacteria in the laboratory for years. We watch each an every mutation that occurs, builds up, and leads to quiet different organisms years later. Not a single letter in every step of the changes missing.

What else do you need, unless you are trying to force a brain into the "Random()" function? It doesn't require a brain, only random changes in base code to 1 of 4 possible letters.

Question:
1) Are you ready to admit the 10^40,000 is the whole organism sequence, not the odds of a single beneficial mutation?
2) Are you ready to admit that a beneficial mutation can occur by a single letter change in 1 of 4 possible letters, A, G, T, or C?