I was short with Joe Felsenstein in the comments section of “Stark Incompetence,” a post in which I address, well, um, the stark incompetence on display in a recent publication of Eric Holloway. I have apologized to Joe, and promised to make amends with a brief post on the topic that he wants to address. Now, the topic is a putative model that Eric introduced in “Mutual Algorithmic Information, Information Non-growth, and Allele Frequency” (or perhaps an improved version of the model). Here is a remark that I addressed to Joe:
Tom English: As you know, if a putative model is logically inconsistent, then it is not a model of anything. I claim that that EricMH’s putative model is logically inconsistent. You had better prove that it is consistent, or turn it into something that you can prove is consistent, before going on to discuss its biological relevance.
I will not have to go far into Eric’s post to identify inconsistencies. After explaining the inconsistencies, which I doubt can be eliminated, I will remark on why the “model” is not worth salvaging. The gist is that Eric’s attempted analysis puts a halting, output-generating simulator of a non-halting, non-output-generating evolutionary process in place of the process itself. An analysis of the simulator would not, in any case, be an analysis of the simuland.
Inconsistency
In the following passage from his post, Eric describes a randomized procedure, dubs it an evolutionary algorithm, and then asserts the existence of a halting program that implements the procedure.
The alleles are 1s and 0s, and the gene G a bitstring of N bits. A gene’s fitness is based on how many 1s it has, so fitness(G) = sum(G). The population consists of a single gene, and evolution proceeds by randomly flipping one bit, and if fitness is improved, it keeps that gene, otherwise it keeps the original. Once fitness(G) = N, the evolutionary algorithm stops and outputs G, which consists of N 1s. The bitstring that is N 1s will be denoted Y. We will denote the evolutionary algorithm E, and it is prefixed on an input bitstring X of length N that will be turned into the bitstring of N 1s, so executing the pair on a universal Turing machine outputs the bitstring of 1s: U(E,X) = Y.
The first thing to note is that the randomized procedure is not an algorithm: it halts with probability less than one. That is, by a simple inductive argument, for all natural numbers 
the probability is less than one that the procedure halts after 
or fewer bit flips. Indeed, the probability is greater than zero that the procedure performs bit flips without improving fitness. Thus if you were to turn the procedure into an algorithm by stipulating that it performs at most 
bit flips, there would be a nonzero probability that gene 
remains equal to input 
when the algorithm halts, no matter how large you make 
The problem is not the particular randomized procedure that Eric has chosen. Evolutionary procedures do not converge surely to local fitness maxima in discrete spaces (except in trivial cases).
It is gobsmacking for me, though probably not for you (I labored over “Stark Incompetence” in hope that everyone would be as astonished by something coming from Eric as I am by many of his claims), to see Eric flatly assert that the randomized “algorithm” is a program for a deterministic computer. To put it plainly, a universal Turing machine cannot flip bits randomly. Eric’s “model” is inconsistent, and thus is not a model. To get randomness, Eric must
- draw the deterministic bit-flipping program
randomly, or - endow the deterministic bit-flipping program with a pseudo-random number generator, and supply the program with a random seed as input (along with
).
).However, this will not produce consistency, because Eric has predicated, with the expression 
that the program surely halts with an output that is maximal in fitness — no ifs, ands, or buts. His subsequent argument requires it. But his prior specification of the random bit-flipping procedure does not allow it.
The map is not the territory
Computer programs used by scientists to model evolutionary processes commonly signal the occurrence of events of interest to the scientists, and halt when the scientists are uninterested in gathering further data. An exceedingly naive response, most prominently on display in the “evolutionary informatics” of Marks, Dembski, and Ewert, is to analyze a program, and claim that the analysis applies to the modeled process — as though the evolutionary process itself signals the occurrence of events and halts.
I am not going to dig into Eric Holloway’s attempt at analysis. You should be able to see for yourself, if you have any business discussing what he has done, that it is literally the program 
that enters into the analysis, and not the evolutionary process that is simulated by the program. It ought to be obvious that an evolutionary process does not “know” when fitness is maximized, let alone announce the genome for which fitness is maximal. The part of the program that detects and announces the event in which fitness is maximized, and subsequently halts, is not part of the (simulation of the) evolutionary process. It is a monitor of the simulated process, and can be decoupled from the simulation per se, even though it is usually tightly coupled with the simulation in practice. I advise against struggling to make Eric’s inconsistent “model” into one that is internally consistent, because the result would be a bogus, though consistent, “model” that mistakes the simulation software for the evolutionary process itself.
Methinks it is like a Weasel.
Dawkins’ little programme did not model evolution, it didn’t even model natural selection. It just illustrated the power of cumulative selection over random change. Eric’s programme does the same. We can only hope that it helps him, and his acolytes, understand how cumulative selection ensures that evolution is nothing like a tornado in a junkyard.
faded_Glory,
At what point in the junkyard scenario do you see selection taking place?
Come on Bill, read more carefully:
Noticed now?
You apparently don’t understand the problem here. To model evolution selected steps have to be identified. Without selectable steps between the tornado and the flying object evolution faces the tornado in a junkyard analogy.
Where are these selectable steps when we are dealing with sequences? Where are the selectable steps between prokaryotic cells and a fully functional ubiquitin system in multicellular organisms. This transition is like the city of New York forming from a hurricane.
No one who is impartial is going to buy something as vague as fitness as a selectable step moving a sequence consistently toward function let alone a ubiquitin system that involves thousands of sequences and requires precise operation for a multicellular animal to form.
I see Bill still hasn’t grasped the concept evolution doesn’t have pre-specified targets. He also doesn’t understand we don’t have to identify every last step in a billion year process to know the general path the process took.
Just Bill being Bill.
We all understand the real problem is your willful scientific ignorance and those Bible blinders you refuse to take off.
Very well said.
Can you tell us all of the steps that you took to eat dinner last night?
If you can’t, I guess that means that you didn’t eat and are about to starve to death. Or, at least, that seems to be the creationist argument that you are making.
Neil Rickert,
I know they exist and I know the mechanism that caused the arrival and consumption of my dinner.
I am not making a creationist argument I simply showing there is no natural theory explaining the cause of major evolutionary transitions at the cellular level.
So much for hoping…
The junkyard/tornado scenario doesn’t contain selection, let alone cumulative selection. That is one of the problems with it. It is an entirely absurd analogy to biological evolution.
Leaving aside colewd’s usual misunderstandings …
As Tom explains, issues like when to stop running the program are not important in investigating models interaction of evolutionary forces. It is reasonable to investigate only a few evolutionary phenomena, as long as you know what you are doing and why. At issue in the ASC mishigos(*) is whether anything can be proven (so far, not), and whether it puts any constraint on what natural selection can achieve.
I will leave this discussion alone for now, but hope in a few days to put up a post at Panda’s Thumb to investigate the issue of whether conservation laws on mutual information can be used to show limits on what natural selection can do. Shannon-style mutual information, not ASC. I think I have a relevant point to make there, using a situation put forward by EricMH himself.
In the meantime I am busy with implementing a useful improvement at Panda’s Thumb.
(*) mishigos, a Yiddish word cognate with the English “mixed-up-ness”.
We know they exist and we know the mechanism that cause the increase in information as species evolved over time. So what’s the problem?
Actually all you’re showing is your willful ignorance because we do have a natural theory explaining the cause of major evolutionary transitions.
Maybe to save time you could just type “BDUISBDBI” –
Bill Doesn’t Understand It So Bill Doesn’t Believe It.
Excellent. A remark I considered making in the OP, but decided to suppress, is that Eric has a hammer, and every problem looks like a nail to him. Classical mutual information is the natural choice when addressing a stochastic process.
Yes, this argument is flawed, and I’ll have to rework it. My sincere apologies. If you are interested in how Levin’s non growth theorem applies to evolutionary algorithms, stay tuned.
They’re indeed similar, but also different in some substantive ways.
That’s a correct characterization of Dawkins’s use of the program(me). Dawkins referred to the monkey/Shakespeare model of cumulative selection, and emphasized that there was not a target in cumulative natural selection as in his model.
However, the “Weasel” program can be used to approximate a special case of the Wright-Fisher model of population genetics. (The special case would arise only in breeding, so the cumulative selection still is not cumulative natural selection). See Joe Felsenstein’s “Wright, Fisher, and the Weasel.”
An important point to grasp, and which I have not learned to convey well, is that it generally is impossible to infer the use of a program by reading its source code. In Dawkins’s model, there indeed is a target string. However, in the Wright-Fisher model, it is simply a mathematical convenience to designate alleles that contribute to fitness as 1s, and alleles that do not contribute to fitness as 0s. The fittest genotype is designated by the string 111…1. But that does not imply that evolution has a target under the Wright-Fisher model.
Dawkins made a valiant effort to diagnose, and respond to, the misconceptions of evolution. However, he did not recognize that people generally do not understand what a model is. Having browsed the responses to your comment, I suspect I’ll be saying more on that matter.
Kudos for acknowledging that your argument is flawed. But it seems that you do not understand the latter section of my post.
I’ve always supposed that the theorem is relevant to evolutionary algorithms, which humans use to search for solutions to problems, not to model biological evolution. Evolutionary biologists are not saying that biological evolution is an evolutionary algorithm executed by the Universe. What technologists like you and me do with an evolutionary algorithm is to use a simulated evolutionary process to sample the space of possible solutions to a problem. The component of the evolutionary algorithm that transforms the sample into a solution to a given problem is logically distinct from the component that generates the sample. There’s nothing “evolutionary” about the transformation of the sample. That is, there’s no analog of it in biological evolution.
Any analysis you do of an evolutionary algorithm, in its entirety, will be irrelevant to the theory of evolution by natural selection. An analysis of the samping component might be relevant.
I recall reading a compelling, logical and detailed explanation, written around Ben Franklin’s time, of how heavier than air flight was impossible and would forever remain impossible. If you didn’t know better (and nobody at the time did), you could hardly help but regard these conclusions as obvious and inevitable.
Today, of course, with enough air traffic to alter weather, most people would be able to find problems with this old essay, and understand that the author’s case rested on assumptions which have turned out to be incorrect in special cases. But imagine if someone’s religion said heavier than air flight is not possible. In that case, someone would STILL find the essay compelling and jet airliners be damned.
Now, let’s substitute “phoodoo” for “someone” and substitute “evolution” for “heavier than air flight.” And sure enough, surrounded by rampant evidence of historical and ongoing evolution, phoodoo will find a post based on demonstrably false assumptions and refuted by every living organism, to be “very well said.” Once again, none so blind as those who WILL not see.
Categorically false.
All that you’re doing here is to demand that someone identify an extremely long evolutionary pathway. You ought to know that term from your reading of Behe. In fact, Behe accepts that natural selection sometimes does account for change. He does not reject models of evolution by natural selection, but instead argues that there are biological systems and structures that the models cannot account for, because there are no evolutionary pathways to them from precursor systems/structures. (Actually, what I’ve summarized is his original notion of irreducible complexity. Midway through Darwin’s Black Box, he switches from claiming nonexistence of a pathway to claiming extreme improbability of pathways.)
Ha. That’s your argument?
Bill, you won.
Your comment is categorically erroneous.
If we accept what you are saying as right, then one can call anything a modelling of evolution. I could put a puddle of silly sludge in a bowl, and say this is modeling evolution. And if one said, well, but its not really modelling evolution, I could refute that by saying, its changing right? See!
It seems to me the only thing you are saying to Bill is that he should be complaining that it is a poor model of evolution (not a non-existent one), if it doesn’t model the steps involved.
I think saying its a poor model, and saying it is not a model, is a pointless semantic disagreement.
Do tell…
This is part of the skeptic bible playbook deception, section three I believe. They hate being called on this one, because its so time consuming to deflect. In section four it covers that though, just say, “go read a book” and then just throw out any title you can think of that sounds convincing.
It seems to me that it would have seemed to you that I said considerably more to Bill if you had processed more than the first two words that I wrote.
Well, not really Tom. You said Bill is asking for an explanation about a very long evolutionary process. The rest of what you said concerns what Behe thinks, which is not really relevant to this discussion.
Yea, it’s supposedly a very long evolutionary pathway. So what? So then we don’t need to explain it, for your theory to have any veracity? I don’t think so, bills complaint is valid. The theory of small changes which confer small fitness benefits that can accumulate into more significant benefits and complex systems doesn’t fit well with the notion of a code for making those outcomes also experiencing the same small benefits adding up over time. Accidental mutations to codes, especially codes which must be completely coherent in order to produce completed outcomes doesn’t make a very convincing theory.
You can’t just randomly change parts of a computer program, hope the change is not too destructive, and then hope those changes one day turn not to be useful. That’s a whack theory. I think skeptics too often expect rational non skeptics (the real kind of skeptics in my opinion) to ignore the bottom line of what your theory proposes.
Unfortunately, I think it’s the greater public, who is not really paying much attention to the details who is ignoring this obvious flaw.
Phoodoo, you are being misled by the analogy between DNA and a compter progamme. Although useful in some aspects, it fails in others. DNA does not, and is not, completely coherent, yet it routinely produces completed outcomes.
There are plenty of mutations that can break it, but there are also plenty of mutations that make little to no difference, and some that cause improvements (in phenotype fitness).
Mutations don’t have to generate major improvements all in one go to cause long term change. Small variations from step to step are adequate, as long as they are subject to bad outcomes being filtered out.
Although you think otherwise, you could actually generate working computer code that way, but the effort in doing so would be much larger than for a human to sit down and write code. That human effort would actually not have to be spent on the mutation side (which could be automated), but on the selection side of the process.
The filter you would apply to select good outcomes from bad ones would be human judgement – someone has to decide if the output from each mutated programme is of enough value to maintain the changes, or not.
That will be a major effort, and the time spent in that would be beter spent in writing new code from scratch.
Nature doesn’t work that way. The thing with biological evolution is that there exists a simple and ubiquitous filter, one that applies automatically, always and everywhere. It is called the risk of untimely death. In Nature, it comes for free with the package.
Function to do what? Does it have to do with survival and reproduction and other vague concepts intimately related to fitness, perchance?
I’ll look forward to that.
I hope you concentrate on modelling populations and selection. I think mutation is unimportant for capturing the essence of the selection mechanism. As was pointed out by ds at PT, you could start by modeling selection as deterministic. Also, only a single generation matters in order to capture the essence of selection.
I think most believe intuitively that the MI to be captured by the model is that between the aggregated genome of a population and the environment. It’s not, for example, any sort of MI between an aggregated genome and its successor in subsequent generations..
We’re all born with some 100 unique mutations, so your analogy fails.
What’s worse is that that explanation has been given to Bill before multiple times. He just doesn’t like the explanation and wants to replace it with “it was wished into existence” instead. If only he can make everyone else state that to be the truth, then finally, finally, when everyone else can be made to parrot what Bill wishes to be true, maybe he can believe it himself.
No it is not. One does not have to “identify selectable steps” to model the occurrence of evolution. That’s just plainly false.
Go and read Theobald’s 29+ Evidences for macroevolution article.
What code are you talking about? What are “those coutcomes”?
What does “completely coherent” mean with respect to a code? What code are you actually talking about? And what is a “completed outcome” here?
Also, what you happen to find “very convincing” is completely irrelevant. Nobody here is under any illusions about what you can be convinced of. You’ve proven many times before that, to you, the subject of evolution is emotionally and intellectually off-limits to you. Your mental blockade is so deep and overwhelming you can’t even get yourself to accept that natural selection occurs. If you happen to experience that people have stopped bothering to engage your inanities, then this is the reason. When speaking on the topic of biological evolution, something weird happens to you and you become impervious to logic and reason.
But you can with organisms.
Rumraket,
Why 23 pairs of chromosomes for humans? Will it one day be 34? One day can we see six nucleotide bases? Or nine?
Or we to believe that there was once many mutations to the genetic code, which was both redundant and irrelevant, but then one day it became essential and unbreakable?
And what about all those mutations to individual genomes, which showed no effect at all, over many generations, then one day caused a good result that turned beneficial? Is that in the 29 evidences deflection you pulled from the Skeptic Bible?
I see, you don’t need to explain all the crazy details, for skeptics, acceptance is usually the best policy. Why so many questions!
In a recent article, a doctor in China has identified a man who has 44 chromosomes instead of the usual 46. Except for his different number of chromosomes, this man is perfectly normal in every measurable way.
I predict one day we will have 48, and we will eventually be potatoes.
dazz,
The dumb luck theory of life.
Three siblings with 44 chromosomes reported back in 1984.
It’s delightful to watch a creationist retreat from “this is impossible” to “this is unlikely”, whether it is irreducible complexity or, here, chromosome fusion to explain the chimp-human difference in chromosome count.
Sadly, many people do not understand that analysis of human chromosome 2 provides independent confirmation of the shared ancestry.
From previous posts and comments I’m starting to think some of the arguments made by the major ID proponents are mutually incompatible.
Eric and BDemski’s proposal is obviously incorrect. Of course an evolutionary algorithm can generate complexity. The only questions is how likely it is. That depends on a bunch of things: population size, mutation rate and the size and connectivity of functional space which depends on lots of biological details such as the likelihood that an enzyme will catalyze a secondary reaction. It seems to me that the main arguments of Axe and Behe depend on this being true. They either explicitly or implicitly acknowledge that the above is true and then go on to argue details about the functional space.
This method of hitting all the bases reminds me of how they have endless snarky posts on ENV and UD attacking science in general. But just in case some of their supports dont quite agree they’ll have separate posts claiming ID is science.
DNA_Jock,
I said more chromosomes, not less. I am well aware that mutations can destroy.
You understand the difference, right?
Maybe one day all kids with Down’s syndrome will have a reproductive advantage, like when low flying cars chop everyone else’s heads off who are tall?
Did you read this part:
DNA_Jock,
Hey Jock, why do you think it is that the news never reports about all the good mutations that happen to humans, you know, like when some child is born with a mutation from some drug the mother took, that causes superhuman eyesight, or gives them limbs that are webbed and helps them glide out of trees?
Is it just because the news these days is just too cynical, they never like to report happy stories? I mean, these mutations have to be happening some of the time right? A lot actually. They can’t only all be confined to a million years ago, when we didn’t have the internet, yea?
Tip: If you break a cookie, you get TWO cookies.
They can also create.
Ausiannikava D, Mitchell L, Marriott H, Smith V, Hawkins M, Makarova KS,
Koonin EV, Nieduszynski CA, Allers T. Evolution of Genome Architecture in
Archaea: Spontaneous Generation of a New Chromosome in Haloferax volcanii. Mol Biol Evol. 2018 Aug 1;35(8):1855-1868. doi: 10.1093/molbev/msy075.
The process was fairly straightforward and obvious: One chromosome with multiple Origins-of-replication, splits into two at two loci that doesn’t break existing genes. Both fragments retain functional origins-of-replication elements.
There are now two intact functional chromosomes where before there was one. And now we get to hear the excuses. Oh but it’s not really “new” new, in the special creationistic sense of new where new means something never defined. Nothing was “created” bla bla bla where’s the information bla bla bla random is impossible bla bla bla. And yet there it is, it evolved by mutation.
Poor creationism, just can’t catch a break.
You’ve never heard of gifted individuals? There’s a large ensemble of mutations associated with positive effects on all sorts of measures of performance. Particularly in sports.
Literally less than 30 seconds on pubmed:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3993978/
You are only ignorant of these things because you choose to be.
Do tell, what’s the mutation? Super eyesight!! I can’t wait to hear.
I think you need to watch less cartoons and read fewer comic books, and instead read more about actual genetics.
Ahh yes, “super eyesight”. If we can’t find THE WOLVERINE then evolution is false.
Kids, phoodoo is your brain on creationism. Just say no!
The chromosome fusion did not destroy phoodoo, it only fused two chromosomes into one.
I see. So you’re just an idiot. Good to know.
So we should be worried, that we have a chromosome fusion that has no advantages over having the two separate chromosomes? Why?
Rumraket,
and
Some people are good marathon runners, others are better sprinters. Who knew?
Which is more fit? Can they glide from trees?
Which is more fit?
Rumraket,
Woman ..circa 470 B.C.
“Doctor, doctor, can you help me! My child, he was born a long distance runner. Was it because I drank wine when I was pregnant? I think I may be the first one to have ever had such a child. Do you think give him a special name, and me too?”
Mrs. Pheidippides LCA of Long Distance Runners.
Tom English,
This is because the mechanism that is offered is not capable of the task at hand. We know how information is generated. Selection like fitness is a vague concept relative to generating information. If you could model how these could generate information then you would not need to know every step. The problem is that there appears to be no free lunch.
When you offer something as vague as selection as a generator of information without the detail you essentially have nonsense. Behe’s claim about natural selection assumes pre existing information as a starting point.
The theory of evolutions grand claims are being destroyed by the reality of cellular mechanisms. As with the origin of matter we have no good answer for their origin.
Tom, I recently did a sequence comparison of skeletal muscle actin called alpha actin which has 377 amino acids in mammals. That is 20^377 possible arrangements. The alignment is between rats mice pigs and humans. The alignment is 100%. Given there should be billions of fixed neutral mutations since mice and rats split how do you explain this other that the original sequence was designed?
Yes, those genes variant (aka mutants) are not absolutely required to perform well on those tasks. Nobody said they were. But they help, which is why they are significantly correlated with high performance on those tasks.
So there are mutations with beneficial effects on various measures of performance. What you implied earlier there was not. So you were wrong.
And now you’re just flailing around with your comic book understanding of the subject.
Not you, apparently. You seem to be stuck with this silly idea that one must be the best at everything simultaneously. But that’s clearly not the case.
That new chromosomes can’t evolve. But they can.
That beneficial mutations aren’t known. But they are.
Thats you, that’s phoodoo, always wrong. Always ignorant. Always proud of it. Always the laughing stock of any discussion.