Annoying creationists

[delphi_ote]

Can someone bump the jukebox, please? I know I requested "Fitness Landscape", but I'd like to hear a new tune now.

Is a function that maps a genotype to the real numbers that hard to understand? Don't they teach domain and range in elementary school? Fitness landscape. A landscape of fitness values. It defines itself.

I'd like to propose the following function:

[latex]$kleinman:\mathbb{R} \rightarrow (\mathbb{C} - \mathbb{R})[/latex]

(What do you get when you take the real numbers away from the complex numbers?)

 

[Dr Adequate]

Why Lie #5 Is Funny

I thought I'd just explain this briefly for the non-mathematicians here, so we can all join in the laugh.

The wikipedia article fitness landscape^WP gives an example of an optimization problem which can be solved by an evolutionary algorithm. The problem is:

"A delivery truck with a number of destination addresses can take a large variety of different routes, but only very few will result in a short driving time."

This is a brief statement of what mathematicians call the Travelling Salesman Problem: we are given a number of towns (we'll call the number n) and the distances between them. The travelling salesman wishes to visit every town on the list and return to his starting point while minimising the distance travelled.

Now the number of possible routes is the factorial of n (written n!) where by definition n! = 1 x 2 x 3 x ... x (n - 1) x n. This grows rapidly with n. For example, 20! = 2432902008176640000

This means that if we want to find a good route, looking at all the possible routes is not practical. As the wiki article says:

"It is almost impossible to check all possible routes once the number of destinations grows to more than a handful."

Checking every route would be equivalent, in biology, to a creator creating life-forms by trying every possible genome and seeing which ones work. This would be wasteful and time-consuming, and so would checking every route in the Travelling Salesman problem (in computer science, such a method is known as a "brute force algorithm").

So instead of using brute force, computer scientists use an evolutionary algorithm: we simulate the process of evolution so that the "genomes" are potential routes, the "fitness" is the shortness of the route, and the "mutations" are supplied by the computer's random number generator.

As the wiki article says, evolutionary algorithms are "particularly effective" at optimising this and similar problems when the number of routes is too large for a brute force algorithm to be effective.

---

Now, let's look at the mess kleinman's made of this.

In the first place, he thinks, for some crazy reason, that the various alternative routes correspond to selection pressures, when in fact they correspond to potential genomes.

But this blunder, though gross, pales into insignificance next to his belief that a mere three potential routes would pose a problem for an evolutionary algorithm.

That's the stupidest statement I've ever heard anyone make about computer science. Evolutionary algorithms are, as the wikipedia article says, "particularly effective" at solving the problem in cases where brute force algorithms are impractical --- i.e. in cases where there are trillions, quadrillions, or quintillions of routes.

I don't know where kleinman got his delusions on this subject from, since they have no connection with the text of the article he keeps whining about. Perhaps he hears voices in his head. If so, they're as dumb as he is.

Add to this his endless pompous bragging about how he "immediately grasped the significance" of the wiki article, and we have a recipe for Krazy Kreationist Komedy with Kleinman the Klown.

 

[joobz]

Why Lie #5 Is Funny

Nicely worded. Thank you.


Couple this with his continual Assertions of his masterful mathematical skills and yet fully missing the nice code that Delphi wrote that showed how 3 selection pressures can be faster than 1 pressure....

 

[kleinman]

Annoying Creationists

But maybe I'm missing the nuances.

You are missing much more than the nuances.

Can someone bump the jukebox, please? I know I requested "Fitness Landscape", but I'd like to hear a new tune now.

You want to hear a new tune, Dr Schneider has moved on. What’s the problem? Are you having a little problem with the mathematics of mutation and selection?

Is a function that maps a genotype to the real numbers that hard to understand? Don't they teach domain and range in elementary school? Fitness landscape. A landscape of fitness values. It defines itself.

The basic concept is not that hard to understand if you realize that the mapping can only be done implicitly. You don’t have an explicit algebraic relationship and it is hard for some people to realize that you can solve equations that you can’t write down. It is the mapping of the fitness landscape that is difficult to do especially when you have multiple selection conditions. What Paul seems to think is that an optimum on the fitness landscape can be found more quickly by a series of random directionless steps then if a systematic search is done of the landscape. Evolution works by a series of random directionless steps. That is the slowest way of finding an optimum.

I'd like to propose the following function:

Delphi, you deserve your own equation as well:

Delphi + EtOH -> Wikipedia & fitness landscape -> organize sock drawer

Have you ever noticed that it takes longer to organize your sock drawer the more different color socks you have? It’s very easy to organize your socks if they are all the same color.

 

[Dr Adequate]

What Paul seems to think is that an optimum on the fitness landscape can be found more quickly by a series of random directionless steps then if a systematic search is done of the landscape. Evolution works by a series of random directionless steps.

S - E - L - E - C - T - I - O - N, retard.

That is the slowest way of finding an optimum.

No, the slowest method is the "systematic search" you suggest. If you tried applying that to the Travelling Salesman Problem with a mere 20 towns, your wait for an answer would be terminated by the heat death of the universe. To quote from the wikipedia article on fitness landscapes^WP:

"It is almost impossible to check all possible routes once the number of destinations grows to more than a handful."

This is why computer scientists use evolutionary algorithms instead, which, as the wiki article says, are "particularly effective".

You've got a real gift for being wrong, haven't you?

 

[Dr Adequate]

Delphi, you deserve your own equation as well:

Delphi + EtOH -> Wikipedia & fitness landscape -> organize sock drawer

You don't know what an equation is?

Sheesh.

Anyway, please return to your pompous, vaccuous bragging about your mathematical abilities.

 

[kjkent1]

Alright. I guess I'll start organizing my sock drawer, then.

Concur. I will get out the Roundup and the sprayer and take care of the weeds which appeared by magic in my yard during the past few months.

 

[delphi_ote]

[latex]$\mathbb{R}[/latex]

[latex]$\mathbb{R}^c[/latex]

I must re-think my theory. He doesn't always map the real to the imaginary.

 

[Dr Adequate]

Okay, I think it's time to update the kleinman FAQ.

This FAQ will demonstrate two things.

First, that the stuff kleinman recites is nothing more than the retarded droolings of a monomaniac with a grudge against reality.

Second, that his grotesque blunders have been pointed out to him so clearly that he knows that the stuff he's reciting is rubbish; and that therefore he is not merely contemptible for his stupidity and his vanity, as are all creationists; but also for his deep habitual dishonesty.

To prove the second point, the FAQ will consist of posts which have already been made and which prove the first of these points.

Enjoy.


* How kleinman screwed up with ev

The mistake Kleinman has made, or one of them, is to take a realistic value for p (the probability of a point mutation for a given base) but not for n (the population). This gives a totally unrealistic value for the probability that a given substition will occur in the gene pool per generation, which is given by:

q = 1 - (1 - p/3)ⁿ
If, for example, we take a realistic value for p of 10^-8, then for a measly million organisms, q is 0.3%. For a lousy billion, it's 96.4%.

If we use a more realistic order of magnitude for the bacteria, say something like the 10¹⁴ present in a single human gut, then my calculator isn't accurate enough to tell us the difference between q and 1.

Schneider is forced by practical constraints to take n to be small, and has compensated for this by using an unrealistic value for p to give himself a realistic value for q. This is eminently sensible, since it is the amount of variation within the gene pool, rather than the variation between individuals per generation, that determines the rate of evolution.

Kleinman, on the other hand, has chosen his numbers so that the value for q is wildly unrealistic; this is why his estimate of the time the process would take is also wildly unrealistic.

* Kleinman's interpretation of data from ev:

G=1000, mutation rate = 1 mutation per 1000 bases per generation, gamma = 16, binding site width = 6:
Population \ generation for convergence
2 \ failed to converge
4 , 66547
8 , 15916
16 , 17257
32 , 16416
64 , 9082
128 , 9378
256 , 4078
512 , 3685
1024 , 2793
2048 , 2080
4096 , 2565
6000 , 1541
8192 , 1798
16384 , 1001
32768 , 743
65536 , 633
131072 , 483
262144 , 702
524288 , 642
1048576 , 438

ev demonstrates decreasing rates of convergence with increasing population.

* Multiple selection pressures:

There are a number of news reports today about the verification of a gonorrhea "super bug" resistant to all but one of its useful classes of antibiotics. Seems it has aquired resistance to all four (4) previous classes, as explained in the following quote:

Over the years, gonorrhea has become resistant to a number of antibiotic classes starting with sulfa, then penicillin and the tetracyclines before fluoroquinolones.

linky

That's 1) Sulfomanides; 2) Beta-latcams; (3) Tetracyclines; (4) Fluoroquinolones.

...or, two successful adaptations past Kleinman Impossibility.

The gonorrhea super bug Neisseria gonorrhoeae is now on its fifth antibiotic class -- the only one we now have to treat it. Now, Dr. Kleinman has repeatedly claimed that three is the threshold of mutations against selection pressures for microevolution, at which point the life form can no longer survive, as exemplified by the HIV virus. He's also asserted that the threshold between 2 and 3 adaptations is the threshold between micro and macro evolution, and that macroevolution is mathematically impossible.

Dr. Kleinman's mathematics must therefore be incorrect, since the real world is not consistent with his prediction.

In other words, he's busted again.

* New genes:

Evolution of novel genes

The origin of new genes

Novel genes derived from noncoding DNA in Drosophila melanogaster

* The de novo origin of genomes: RNA species from a bucket of chemicals:

Evidence for de novo production of self-replicating and environmentally adapted RNA structures by bacteriophage Qbeta replicase.

Template-free generation of RNA species that replicate with bacteriophage T7 RNA polymerase.

Template-free RNA synthesis by Q beta replicase

De novo DNA synthesis by yeast DNA polymerase I associated with primase activity.

De Novo synthesis of DNA-like molecules by polynucleotide phosphorylase in vitro

De Novo Initiation of RNA Synthesis by the RNA-Dependent RNA Polymerase (NS5B) of Hepatitis C Virus

Template-free generation of RNA species that replicate with bacteriophage T7 RNA polymerase

De novo RNA synthesis by a recombinant classical swine fever virus RNA-dependent RNA polymerase

De novo RNA synthesis catalyzed by HCV RNA-dependent RNA polymerase

Template-free, polymerase-free DNA polymerization

Template-Free Primer-Independent DNA Synthesis by Bacterial DNA Polymerases I Using the DnaB Protein from Escherichia coli

* Kleinman's inability to understand evolutionary algorithms:

Why Lie #5 Is Funny

I thought I'd just explain this briefly for the non-mathematicians here, so we can all join in the laugh.

The wikipedia article fitness landscape^WP gives an example of an optimization problem which can be solved by an evolutionary algorithm. The problem is:

"A delivery truck with a number of destination addresses can take a large variety of different routes, but only very few will result in a short driving time."

This is a brief statement of what mathematicians call the Travelling Salesman Problem: we are given a number of towns (we'll call the number n) and the distances between them. The travelling salesman wishes to visit every town on the list and return to his starting point while minimising the distance travelled.

Now the number of possible routes is the factorial of n (written n!) where by definition n! = 1 x 2 x 3 x ... x (n - 1) x n. This grows rapidly with n. For example, 20! = 2432902008176640000

This means that if we want to find a good route, looking at all the possible routes is not practical. As the wiki article says:

"It is almost impossible to check all possible routes once the number of destinations grows to more than a handful."

Checking every route would be equivalent, in biology, to a creator creating life-forms by trying every possible genome and seeing which ones work. This would be wasteful and time-consuming, and so would checking every route in the Travelling Salesman problem (in computer science, such a method is known as a "brute force algorithm").

So instead of using brute force, computer scientists use an evolutionary algorithm: we simulate the process of evolution so that the "genomes" are potential routes, the "fitness" is the shortness of the route, and the "mutations" are supplied by the computer's random number generator.

As the wiki article says, evolutionary algorithms are "particularly effective" at optimising this and similar problems when the number of routes is too large for a brute force algorithm to be effective.

---

Now, let's look at the mess kleinman's made of this.

In the first place, he thinks, for some crazy reason, that the various alternative routes correspond to selection pressures, when in fact they correspond to potential genomes.

But this blunder, though gross, pales into insignificance next to his belief that a mere three potential routes would pose a problem for an evolutionary algorithm.

That's the stupidest statement I've ever heard anyone make about computer science. Evolutionary algorithms are, as the wikipedia article says, "particularly effective" at solving the problem in cases where brute force algorithms are impractical --- i.e. in cases where there are trillions, quadrillions, or quintillions of routes.

I don't know where kleinman got his delusions on this subject from, since they have no connection with the text of the article he keeps whining about. Perhaps he hears voices in his head. If so, they're as dumb as he is.

Add to this his endless pompous bragging about how he "immediately grasped the significance" of the wiki article, and we have a recipe for Krazy Kreationist Komedy with Kleinman the Klown.

* Kleinman on the definition of "macroevolution" (thanks to Dr Richard and Mr Scott):

If you had read this thread you would already have known that I accept that microevolutionary processes occur. I have always acknowledged this.

I am working with the evolutionist definition, which is there is no difference between micro and macroevolution

With respects to macroevolution, we can work with your definition that there is no distinction between micro and macroevolution

I happen to like the terms micro and macroevolution to distinguish between evolutionary events which occur and those which don’t occur

let’s work from the assumption that macroevolution is simply the sum of a series of microevolutionary steps

The goal post for microevolution is the transformation of a gene from some initial function to a new and completely different function and the evolution of a gene from the beginning. There is/are no selection process(es) that can accomplish this.

The goal post for evolution is the transformation of a gene from some initial function to a new and completely different function and the evolution of a gene from the beginning. There is/are no selection process(es) that can accomplish this.

I posed the cases of the evolution of a gene from the beginning and the transformation of a gene from one form to an entirely new form as examples of macroevolution

I’m accepting for the sake of discussion that there is no distinction between micro and macroevolution

I have not presented the definition for macroevolution

Paul holds the position that a series of microevolutionary changes can lead to a macroevolutionary change. This is why I set up the goal post of the evolution of a gene from the beginning and the transformation of a gene from one form to another

My view on this issue is that once you get beyond a single point mutation you are already starting to enter the realm of macroevolution.

The goal post for macroevolution is the transformation of a gene from some initial function to a new and completely different function and the evolution of a gene from the beginning. There is/are no selection process(es) that can accomplish this

* Department of "did he really say that?":

Kleinman on natural selection:

I am interested if you can explain why a beneficial mutation can be selected for.

Kleinman on "macroevolution":

My view on this issue is that once you get beyond a single point mutation you are already starting to enter the realm of macroevolution.

Kleinman thinks he has a clue:

We do have clues. Most mutations are harmful, ask anyone with a genetic disease.

The purpose of Kleinman's life, according to Kleinman:

I have a purpose in life Adequate, it is to annoy you.

 

[BPScooter]

I simply have to speak here, for the sake of future suffering readers that may not speak English as the natal tongue, and to help my own understanding:

"ev" is a computer program designed to probe certain areas in possible design space, with constraints and possibilities set up in advance, right?

*really smart people made this program to get some insight about certain esoterica regarding information and its development and transfer, right?

*an equally smart person has seized upon aspects of this to further an alternate theory: evolution is "mathematicallly impossible" and in the doing so refute "dumbass" theories, and quiet "silly" "whining" people with the contention that they are ignorant of the "math".

In terms of rhetoric, I think Kleinman is missing a great opportunity to convince, and seeking opportunities to offend. I admit, the term "liar" has been applied to him. This tends to raise the ante.

Perhaps all of you will be happy to know that I have been reading Dawkins' Blind Watchmaker for the first time and giggled at the quaint references to BASIC and Pascal in it.

cheers to you all, keep it decent if you can.

 

[Dr Adequate]

I simply have to speak here, for the sake of future suffering readers that may not speak English as the natal tongue, and to help my own understanding:

"ev" is a computer program designed to probe certain areas in possible design space, with constraints and possibilities set up in advance, right?

*really smart people made this program to get some insight about certain esoterica regarding information and its development and transfer, right?

*an equally smart person ...

Ah, let me stop you there.

Kleinman has expended thousands and thousands of words reciting a handful of very, very stupid lies over and over again to a tiny audience of people all of whom know that he's lying, and whose main reason for participating this thread is to laugh at his buffoonish antics. Which aspect of that strikes you as smart?

In terms of rhetoric, I think Kleinman is missing a great opportunity to convince ...

No, not really. Demosthenes himself couldn't make kleinman's crap convincing. It's not the schizophrenic raving about cheese and alchemy, nor the stupid made-up words, nor the screaming, twitching temper-tantrums that really let him down. It's the fact that everything he says is bollocks.

Rhetoric is, as it were, the final polish we give to our thoughts to make them shine --- but one cannot polish a turd.

 

[Ichneumonwasp]

Oh, ok, sorry I misrepresented your position.

That I can respect. Thank you.

Do you think that multiple selection pressures applied simultaneously can accelerate evolution and if so do you have any mathematical or real examples of this.

Accelerate evolution? That depends on how we define evolution. Mathematical or real examples -- depends on how you take my explanation below concerning the involved principles.

Multiple selection pressures (holding potency of each pressure constant) will decrease variability more than one selection pressure of the same potency. This follows from the definition of "selection pressure". If we define evolution in terms of the number of variants (which represent a form of change over time), then evolution will be slowed in this instance. Actually, at this stage, any definition of evolution means that the process is slowed -- again, at this stage.

However, there is another sense of the word "evolution", referring to the accumulation of changes over time.

So, for instance, if there were no selection pressures, variability in organisms would reach maximum. But we couldn't sepeak of information in the same way that we do now. We would see a blinding array of different organisms, but any changes could occur and disappear with no consequence one way or the other. By applying selection pressures, we would slow this process of change (one meaning of evolution). Multiple selection pressures (holding potency constant) would slow this process even more.

Since information would basically be inconequential, however, we would not see what our world holds. Selection pressures obviously determine what sorts of changes survive and leave behind more copies of themselves (by definition, again). Multiple selection pressures applied simultaneously impact the process more. While they would slow the process of increasing variability and slow the evolutionary process (at that stage), they actually would be more profound at spurring cumulative changes over time.

For HIV, a single selection pressure would be likely to result in resistance to a single drug (or drug class). Multiple relatively weak (weak compared to current practices) pressures would spur resistance to multiple drugs. So, while multiple selection pressures (holding potency constant) will slow the process of change (defined as increasing variability) in the initial stages, they will ultimately result in greater change if the organisms survive.

Like most philosohpical debates, in other words, the issue revolves about the particular definitions of the words used.

 

[Paul C. Anagnostopoulos]

Your scheme for changing the weights is far from being realistic. If you increase the weights uniformly, you get the same generations for convergence for each case. If you want to model selection pressures realistically, your selection pressures need to be tied to the dead of creatures. A hundred fold increase in the selection intensity would cause far more deaths.

The mistake counts were not added to provide uniform selection pressure increase, but to allow the relative selection pressures to vary. You were not involved in the discussion with Cristi Pavel that resulted in this feature.

That fits perfectly with his string cheese theory of evolution.

You're really avoiding this issue, aren't you?

The way you model the variance in selection pressure by varying the weights neglects a very important realistic effect. If the selection pressure represents the concentration of an antimicrobial agent, increasing that value 100 fold probably would markedly impair the microbes’ ability to reproduce. Selection intensity is a highly nonlinear variable. If the intensity is high enough, it causes extinction. You have not included this effect in the model.

Correct.

Of course you don’t evolve the same final creature when you change the selection pressures. What is meaningful is that it takes huge numbers of more generations to evolve the three selection conditions in ev than evolving any single selection condition. It is much more difficult to find an optimum on the fitness landscape when you are trying to satisfy three selection conditions than when you are trying to satisfy a single selection condition.

No argument there, at least as far as Ev is concerned. But you're not extrapolating this to all possible selection pressures in the real world, are you?

That’s ridiculous Paul, what are you going to do, change the code in ev so that you can’t set a selection pressure weight equal to zero? You do that, I like it when you have to hide the real behavior of your model in a lame attempt to make your case.

As long as you stop making any claims about evolving "perfect creatures" when one or more mistake counts are zero, I'm happy to drop this subject. When mistake counts are zero, the creature is not "perfect" in the original meaning of the term.

That’s easy to explain, you have three selection conditions, one condition where it is a mistake if a binding site is not located where it should, the second condition is that it’s a mistake if a binding site is located in the gene region of the genome and the third condition is that is a mistake if a binding site is located outside of the gene region. Interestingly, it does not take all three selection conditions to evolve your binding sites. Setting any one of the selection conditions to zero simply stops selection for that condition. Setting all three of the weight factors to zero is equivalent to turning selection off in the model. What is the function of the genome when you set one or two of the selections to zero? That function is simply to satisfy the selection conditions imposed on your creatures.

Excellent, so we agree that the creatures have different evolved functions, depending on the mistake counts. It is therefore quite dangerous to compare the number of generations to evolve these different functions. A vague generalization about Ev requiring more generations when all mistake counts are nonzero is fine. Extrapolation to the real world is not.

~~ Paul

 

[kleinman]

Annoying Creationists

Alright. I guess I'll start organizing my sock drawer, then.

Concur. I will get out the Roundup and the sprayer and take care of the weeds which appeared by magic in my yard during the past few months.

Another example of a single selective pressure, watch out that you don’t select for resistant weeds, it’s much easier with a single selective pressure.

I must re-think my theory. He doesn't always map the real to the imaginary.

Just how many socks do you have in your drawer?

I simply have to speak here, for the sake of future suffering readers that may not speak English as the natal tongue, and to help my own understanding:

We plan on having this thread translated into Vulcan.

"ev" is a computer program designed to probe certain areas in possible design space, with constraints and possibilities set up in advance, right?

Yes, these are the areas of the design space Dr Schneider intended to investigate:

Variations of the program could be used to investigate how population size, genome length, number of sites, size of recognition regions, mutation rate, selective pressure, overlapping sites and other factors affect the evolution.

*really smart people made this program to get some insight about certain esoterica regarding information and its development and transfer, right?

I don’t know what you mean by esoterica.
Dr Schneider said the following about his model:

Actual biological materials were used to determine the original hypothesis. Read the literature: Schneider1986

And he said this about ev.

The simulation was of phenomena in the "real" world.

*an equally smart person has seized upon aspects of this to further an alternate theory: evolution is "mathematicallly impossible" and in the doing so refute "dumbass" theories, and quiet "silly" "whining" people with the contention that they are ignorant of the "math".

Oh yes, I seized upon genome length, mutation rate, population size and selection pressure, all esoteric features in the ev model that Dr Schneider thought should be investigated. And each of these esoteric features show your theory to be mathematically impossible.

In terms of rhetoric, I think Kleinman is missing a great opportunity to convince, and seeking opportunities to offend. I admit, the term "liar" has been applied to him. This tends to raise the ante.

What you don’t understand BPScooter is that the simple act of challenging the theory of evolution is an offense to evolutionists. Evolutionists use a similar strategy that lawyers use in court. If you have the law on your side, argue the law, if you have the facts on your side, argue the facts, if you have neither, attack your opponent. The evolutionist variant of this principle is if you have the mathematics, argue the mathematics, if you have the facts, argue the facts and since evolutionists have neither, they call me a liar. This shows how weak the evolutionist case is.

Perhaps all of you will be happy to know that I have been reading Dawkins' Blind Watchmaker for the first time and giggled at the quaint references to BASIC and Pascal in it.

Maybe Dawkins needs Turbo Basic.

cheers to you all, keep it decent if you can.

And Adequate, if you can’t keep it decent, at least don’t be dull and boring.

Do you think that multiple selection pressures applied simultaneously can accelerate evolution and if so do you have any mathematical or real examples of this.

Accelerate evolution? That depends on how we define evolution. Mathematical or real examples -- depends on how you take my explanation below concerning the involved principles.

In ev, this is very easy to define, it is the reduction in the number of generations necessary to evolve to selective conditions. In real examples, the examples are analogous. Consider the case of treating HIV again. Monotherapy leads to the emergence of resistant strains more quickly (fewer generations) than combination therapy.

Multiple selection pressures (holding potency of each pressure constant) will decrease variability more than one selection pressure of the same potency. This follows from the definition of "selection pressure". If we define evolution in terms of the number of variants (which represent a form of change over time), then evolution will be slowed in this instance. Actually, at this stage, any definition of evolution means that the process is slowed -- again, at this stage.

That’s interesting that you would describe the affect of selection pressure this way. Alleles that would allow a creature to survive and reproduce when there is no selective pressure could be a disadvantage when a particular selective pressure is applied. Are you saying that selective pressure reduces information in the gene pool?

However, there is another sense of the word "evolution", referring to the accumulation of changes over time.

This is why you need to familiarize yourself with ev. Dr Schneider has attempted to put stringent mathematical definitions to the concept of mutation and selection. Until you have some understanding of how he did this, you will not have systematic way of discussing how each of the variables affects the evolutionary process. Putting this mathematical framework on the debate gives a way of testing the hypotheses.

So, for instance, if there were no selection pressures, variability in organisms would reach maximum. But we couldn't sepeak of information in the same way that we do now. We would see a blinding array of different organisms, but any changes could occur and disappear with no consequence one way or the other. By applying selection pressures, we would slow this process of change (one meaning of evolution). Multiple selection pressures (holding potency constant) would slow this process even more.

Selection pressures always act. The variability (divergence of genes) is always going to be limited. If a gene diverges to the point that the gene no longer functions, there will be a selection pressure against that creature. This is why I have problems with Delphi’s concept of gene duplication as a mechanism for creating new genes. The transformation of the gene to a new function must take a path on the fitness landscape that never causes selection against that creature.

Since information would basically be inconequential, however, we would not see what our world holds. Selection pressures obviously determine what sorts of changes survive and leave behind more copies of themselves (by definition, again). Multiple selection pressures applied simultaneously impact the process more. While they would slow the process of increasing variability and slow the evolutionary process (at that stage), they actually would be more profound at spurring cumulative changes over time.

The problem you have is that the mathematics of ev shows that the number of generations required to accumulate these changes (by random point mutations) is huge. The number of generations required is far too large to support the concept of macroevolution.

For HIV, a single selection pressure would be likely to result in resistance to a single drug (or drug class). Multiple relatively weak (weak compared to current practices) pressures would spur resistance to multiple drugs. So, while multiple selection pressures (holding potency constant) will slow the process of change (defined as increasing variability) in the initial stages, they will ultimately result in greater change if the organisms survive.

In the early stage of treatment, even with weak selective pressures, the evolutionary process is slowed until an appropriate resistant allele can evolve and that allele gets selected for and becomes dominant in the population.

Like most philosohpical debates, in other words, the issue revolves about the particular definitions of the words used.

This is why I like debating ev. The definitions are mathematical and the assumptions used at arriving at these definitions can be easily scrutinized. If you are going to continue on in this discussion, you had better familiarize yourself with ev. This thread is about the mathematics of mutation and selection.

Your scheme for changing the weights is far from being realistic. If you increase the weights uniformly, you get the same generations for convergence for each case. If you want to model selection pressures realistically, your selection pressures need to be tied to the dead of creatures. A hundred fold increase in the selection intensity would cause far more deaths.

The mistake counts were not added to provide uniform selection pressure increase, but to allow the relative selection pressures to vary. You were not involved in the discussion with Cristi Pavel that resulted in this feature.

Modeling relative selection pressures more realistically will take much more than the strategy that you have employed. Even with your simple strategy, the relative selective pressures change as you lengthen the genome. With large genomes, the non-binding site region becomes the dominant source of spurious binding mistakes. This is why as you lengthen the genome, even with a mutation rate fixed to a number of bases, that the number of generations for convergence increases.

The way you model the variance in selection pressure by varying the weights neglects a very important realistic effect. If the selection pressure represents the concentration of an antimicrobial agent, increasing that value 100 fold probably would markedly impair the microbes’ ability to reproduce. Selection intensity is a highly nonlinear variable. If the intensity is high enough, it causes extinction. You have not included this effect in the model.

Correct.

Do you want to venture a guess whether including this effect would speed or slow the evolutionary process? I’ll stick my neck out and guess that this will slow evolution by reducing populations.

Of course you don’t evolve the same final creature when you change the selection pressures. What is meaningful is that it takes huge numbers of more generations to evolve the three selection conditions in ev than evolving any single selection condition. It is much more difficult to find an optimum on the fitness landscape when you are trying to satisfy three selection conditions than when you are trying to satisfy a single selection condition.

No argument there, at least as far as Ev is concerned. But you're not extrapolating this to all possible selection pressures in the real world, are you?

Yes I am extrapolating this result to all possible selection pressures. Increasing the number of selection pressures is simply increasing the number of sorting conditions. It doesn’t matter what the selection conditions are. Delphi’s Wikipedia link to fitness landscape discusses the same issue. This is a basic mathematical principle of optimization. When Delphi went to sort his sock drawer; the more colors of socks he has the slower the sort proceeds. If all his socks are of a single color, any two socks he grabs randomly out of the drawer gives a match.

That’s ridiculous Paul, what are you going to do, change the code in ev so that you can’t set a selection pressure weight equal to zero? You do that, I like it when you have to hide the real behavior of your model in a lame attempt to make your case.

As long as you stop making any claims about evolving "perfect creatures" when one or more mistake counts are zero, I'm happy to drop this subject. When mistake counts are zero, the creature is not "perfect" in the original meaning of the term.

It is your terminology which is causing confusion. What you call a “perfect creature” is nothing more than a genome with a sequence of bases that satisfies all three of your selection condition. Reducing the number of selection conditions by setting selection weights to zero gives a genome with a sequence of bases that satisfies the remaining selection conditions when you click the check box – Pause on perfect creature. You can’t use your confusing terminology to obscure what is occurring mathematically.

That’s easy to explain, you have three selection conditions, one condition where it is a mistake if a binding site is not located where it should, the second condition is that it’s a mistake if a binding site is located in the gene region of the genome and the third condition is that is a mistake if a binding site is located outside of the gene region. Interestingly, it does not take all three selection conditions to evolve your binding sites. Setting any one of the selection conditions to zero simply stops selection for that condition. Setting all three of the weight factors to zero is equivalent to turning selection off in the model. What is the function of the genome when you set one or two of the selections to zero? That function is simply to satisfy the selection conditions imposed on your creatures.

Excellent, so we agree that the creatures have different evolved functions, depending on the mistake counts. It is therefore quite dangerous to compare the number of generations to evolve these different functions. A vague generalization about Ev requiring more generations when all mistake counts are nonzero is fine. Extrapolation to the real world is not.

I don’t agree that you have creatures that have evolved different functions. The creatures have evolved to satisfy each of the selection conditions. In the case of three selection conditions your “perfect creature” has evolved a genome to satisfy all three selection condition. When you set some of the selection conditions to zero, you evolve a genome to satisfy the remaining selection conditions. Do you think when HIV develops resistance to a particular drug it matters whether this occurs during monotherapy or combination therapy?

What is dangerous is your extrapolation of these microevolutionary events to macroevolution. Your extrapolation is not an extrapolation to the real world.

 

[Mr. Scott]

Alan Kleinman Lie #3 Exposed

Dr. Alan Kleinman is wrong about how the three selection pressures affect the rate of evolution. What matters is the combined intensity of selection pressures, not their count, as explained in the article Patients whose therapy fails having used at least three classes of drugs.

using a number of such drugs together might have a cumulative benefit which outweighs the potential toxicity

Here's how it works:

1) A virus colony like HIV can be stopped if an anti-viral drug is used at sufficient intensity to stop all virus individuals before any of them have time to adapt.

2) Dosages of anti-viral drugs must nevertheless be low enough to avoid harming the patient, so it may not be wise to administer the concentration required to stop the entire colony before resistance develops.

3) Administration of three different anti-virals may roughly triple the pressure against the virus while not tripling the level of toxicity to the patient.

4) The advantage of multi-drug therapy is therefore to change the ratio of efficacy to patient toxicity. It's not to increase the count of selection pressures.

5) Application of multi-drug therapy below compliance would likely result in multi-adaptation to all the drugs (macroevolution in Kleinspeak).

6) Dr. Kleinman's HIV example of triple-drug therapy as evidence that macroevolution can't happen, is therefore utterly bogus.

Cancer treatments likewise can involve a similar approach. The best chance of cancer recovery comes from using more than one approach against the cancer colony, typically 1) physical removal of the cells, 2) radiation, 3) chemotherapy. This works because each of the three remedies will have an additive effect at fighting the cancer but not at harming the patient. The operation will remove as little good tissue as practical, the radiation and chemo likewise. Their effects are additive against the cancer but not against the patient.

Multi-drug therapy works on the same principle. It has nothing at all to do with "three point mutations are impossible therefore evolution is impossible." It's a way to launch a toxic attack on a virus that is not toxic to the patient.

Kleinman, you're busted again.

 

[kleinman]

Annoying Creationists

Dr. Alan Kleinman is wrong about how the three selection pressures affect the rate of evolution. What matters is the combined intensity of selection pressures, not their count, as explained in the article Patients whose therapy fails having used at least three classes of drugs.

Mr Scott, you are confusing several principles. Those principles you are confusing are selection pressures, intensity of selection pressures and extinction. The mathematics is clear; increasing the number of selection pressures slows evolution. The number of selection pressures and the intensity of selection pressures determine whether extinction occurs. When you treat patients with HIV and use selection pressures that are insufficient to cause extinction of the virus, you will get resistant strains of the virus to those selection pressures. The fewer the number of selection pressures, the more rapidly the resistant strains appear.

 

[Paul C. Anagnostopoulos]

Modeling relative selection pressures more realistically will take much more than the strategy that you have employed. Even with your simple strategy, the relative selective pressures change as you lengthen the genome. With large genomes, the non-binding site region becomes the dominant source of spurious binding mistakes. This is why as you lengthen the genome, even with a mutation rate fixed to a number of bases, that the number of generations for convergence increases.

I agree, a realistic model of selection pressures is way more complex. Thus I would not draw any conclusions about selection pressures in the real world. The feature was added for the same reason as the tie-breaking feature: at the insistence of a creationist.

Do you want to venture a guess whether including this effect would speed or slow the evolutionary process? I’ll stick my neck out and guess that this will slow evolution by reducing populations.

Well, extinction would certainly slow the evolution of the extinct species, yes.

Yes I am extrapolating this result to all possible selection pressures. Increasing the number of selection pressures is simply increasing the number of sorting conditions. It doesn’t matter what the selection conditions are. Delphi’s Wikipedia link to fitness landscape discusses the same issue. This is a basic mathematical principle of optimization. When Delphi went to sort his sock drawer; the more colors of socks he has the slower the sort proceeds. If all his socks are of a single color, any two socks he grabs randomly out of the drawer gives a match.

So you're claiming that there is no possible way we could have two orthogonal selection pressures to which an organism adapts in parallel?

Also, I can sort any number of sock colors at the same speed: I look at each sock and drop it in the correct pile. Perhaps you really mean to discuss selection speed?

It is your terminology which is causing confusion. What you call a “perfect creature” is nothing more than a genome with a sequence of bases that satisfies all three of your selection condition. Reducing the number of selection conditions by setting selection weights to zero gives a genome with a sequence of bases that satisfies the remaining selection conditions when you click the check box – Pause on perfect creature. You can’t use your confusing terminology to obscure what is occurring mathematically.

Yes, my terminology is causing confusion. Therefore, you should not employ my terminology in making your point. We all agree that different numbers of selection pressures will cause Ev to arrive at a zero-mistake creature in different numbers of generations. But to refer to all those various creatures as "perfect creatures" obscures the issue.

I don’t agree that you have creatures that have evolved different functions. The creatures have evolved to satisfy each of the selection conditions. In the case of three selection conditions your “perfect creature” has evolved a genome to satisfy all three selection condition. When you set some of the selection conditions to zero, you evolve a genome to satisfy the remaining selection conditions.

And thus, those two creatures have evolved different functions. One distinguishes binding sites from all other sites. The other does not.

~~ Paul

 

[Paul C. Anagnostopoulos]

Mr Scott, you are confusing several principles. Those principles you are confusing are selection pressures, intensity of selection pressures and extinction. The mathematics is clear; increasing the number of selection pressures slows evolution. The number of selection pressures and the intensity of selection pressures determine whether extinction occurs. When you treat patients with HIV and use selection pressures that are insufficient to cause extinction of the virus, you will get resistant strains of the virus to those selection pressures. The fewer the number of selection pressures, the more rapidly the resistant strains appear.

And the intensity of those selection pressures doesn't matter? Three mild pressures will always result in "slower evolution" than two brutal pressures?

~~ Paul

 

[Ichneumonwasp]

There's no sense in rehashing the rest of your replies, but......

This is why I like debating ev. The definitions are mathematical and the assumptions used at arriving at these definitions can be easily scrutinized. If you are going to continue on in this discussion, you had better familiarize yourself with ev. This thread is about the mathematics of mutation and selection.

The very issue of this thread is whether or not ev serves as a proper model for what you think it does. You cannot assume that ev properly models the mathematics of evolution (as it exists in the real world). You provided as support for this idea (that ev does model evolutionary change properly) HIV triple therapy as directly analogous to ev's functions. In your assessment of ev you have claimed that ev proves that evolution is so slow when three selection pressures are used that it could never have accounted for the bounty that surrounds us. However, if HIV triple therapy is analogous to the function of ev, as you have proposed, then your theory is wrong. Resistance develops in the presence of HIV triple therapy when the potency of this therapy is taken into account -- specifically when the potency is not as strong as the currently used treatments. And this resistance develops relatively quickly on an evolutionary time scale.

I am forced to conclude either: (1) ev does not adequately model three selection pressures as they relate to the development of resistance or (2) ev does not adequately model the development of resistance in a realistic time frame, which may simply be an issue of population size (which ev cannot handle).

Whatever the explanation, ev does not trump reality. Reality wins this debate. You can like ev as a model all you want, but where it fails, it fails. It served its purpose to demonstrate, with the use of Darwinian pressures, an increase in information. It wasn't designed to be used in the way that you are using it; and your own personal hand-picked example -- HIV triple therapy -- reveals that it does not correctly model reality.

 

[Paul C. Anagnostopoulos]

I am forced to conclude either: (1) ev does not adequately model three selection pressures as they relate to the development of resistance or (2) ev does not adequately model the development of resistance in a realistic time frame, which may simply be an issue of population size (which ev cannot handle).

I'll go with (1). If Ev's simplistic multi-pressure model happens to mimic real-world evolution of drug resistance, either that is a stunning coincidence or real-world resistance is trivial. Wait, or both.

~~ Paul