… with Mung. In a recent comment Mung asserted that
If Darwinists had to put up their hard earned money they would soon go broke and Darwinism would be long dead. I have a standing $10,000 challenge here at TSZ that no one has ever taken me up on.
Now, I don’t have $10,000 to bet on anything, but it is worth exploring what bet Mung was making. Perhaps a bet of a lower amount could be negotiated, so it is worth trying to figure out what the issue was.
Mung’s original challenge will be found here. It was in a thread in which I had proposed a bet of $100 that a Weasel program would do much better than random sampling. When people there started talking about whether enough money could be found to take Mung up on the bet, they assumed that it was a simple raising of the stake for my bet. But Mung said here:
You want to wager over something that was never in dispute?
Why not offer a meaningful wager?
So apparently Mung was offering a bet on something else.
I think I have a little insight on what was the “meaningful wager”, or at least on what issue. It would lead us to a rather extraordinary bet. Let me explain below the fold …
Mung accepted that Weasel programs reach their goal far faster than random sampling. However Mung also said (here) that
Weasel programs perform better than blind search because they are guided. I didn’t think the performance was in dispute, nor why the performance was better.
and elsewhere characterized Weasels as succeeding because of “intelligence” as opposed to ignorance.
So let’s imagine what might happen if we took Mung up on the $10,000 bet. We would bet that the Weasel would succeed, because of cumulative selection. Mung would bet that (because of “intelligence” or being “guided”) the Weasel would succeed. The stake would be held by a house of some sort, which would not take a commission.
The Weasel would be run. It would succeed. So the house would declare that we had all won. The stake would be given to the bettors, in proportion to their bets. But alas, no one actually bet against the Weasel. So the winnings would be zero. Everyone, Mung and the rest of us, would get their stake back, and that’s all.
To bet against Mung, we have to come up with some event that distinguishes cumulative selection from “intelligence” (or being “guided”). That seems to be the issue on which Mung was offering a $10,000 bet, and declaring (here) us all to be “pretender[s]” because we would not put up or shut up.
So there it is. We’re all betting on the same side, and no one will win or lose a penny. Unless Mung can come up with some test that distinguished “intelligence” or being “guided” from cumulative selection.
Now I am possibly misunderstanding what the bet actually would be. I hope that Mung will straighten us out on that, so that we can understand what test is proposed, and place our bets.
I’ve been trying to crunch the numbers, let’s see if they make any sense….
if the algo is capable of hovering around a high enough average fitness, the probability of hitting the target becomes higher than random shuffling the characters.
For a mutation rate of 5% and a 28 long target, any string with fitness higher than 13 is more likely to produce a descendant that hits the target than random shuffling. If my algo is working fine, that can be achieved with just 4 descendants per generation
Perhaps colewd could write a version of what he’s talking about in Python? It’s not difficult, really.
My wager was that I could write an algorithm that performed better than the one by Dawkins, which I did.
You also have N = 1 children. My reading of the code is that its selection consists of choosing the best of the children, and not comparing them with the parent. So with N = 1 you have no selection at all. Hence they perform the same.
Try, say, N = 2. I bet it shows a bigger effect of selection.
Yep, of course you are right. With only one descendant I essentially had Bill’s implementation of “random search” with a few mutations per generation. No wonder it averaged the same fitness as the purely random one.
With two descendants it clearly outperforms the random search with an average fitness of 60
So what?
Both. Guided in selecting certain traits , unguided in the unintentional consequences of selecting for certain traits. Belyayev’s fox experiment.
My take: the phenotype selected in artificial selection changes in the way the farmer wants. All the details of how this is accomplished are not specified by the farmer. So we cannot, without much further information, predict what gene frequencies for individual loci will do — how exactly that phenotype is achieved. In this sense it is not like engineering design, which controls most details. And it is not transmitting detailed design information into the genome.
And it is in that sense that it is analogous to natural selection, so that Darwin could devote the first chapter of the “Origin” to it.
FWIW, one could propose a bet on the converse, namely:
Now regarding sharp wagering, when possible take weighted bets where you can bet on both outcomes and come out with some profit whatever the outcome. That may sound outrageous, but Alan Woods bet on horse races and probably all the horses with weighted amounts. He amassed 2 billion dollars in the process.
Those are called arbitrage bets.
Don Johnson used implicit arbitrage bets in Atlantic city to win millions because he essentially suckered the casino through negotiations to give him loss rebates!
https://www.theatlantic.com/magazine/archive/2012/04/the-man-who-broke-atlantic-city/308900/
What does this have to do with the OP? Maybe nothing. I would suggest however not taking Mung up on his bet, not because he’s right or wrong, it’s just a bad bet because you’re counting on Mung honoring his side of the deal if you win. Not that he’s dishonest, but do you think you could persuade him he lost the bet and thus pony up the money?
Secondly who would you trust to function as the house (the casino broker) with your money? Yikes!
Not to mention, that could be a violation of federal regulations (remember, brick and mortar casinos don’t like online gambling because it cuts into their profits. They’ll be quick to notify the feds about the evil online gambling going on.)
You can probably get away with such a bet if the winnings were given to charity.
I would say artificial selection is natural selection. So is genocide.
That is one of the problems with the whole “fitness” concept. If a bunch of poachers decide to kill off all elephants, so they can sell their ivory, and the elephants go extinct because of this, then natural selection didn’t favor them. Likewise, if we turn a wolf into a chihuahua, that is also natural selection.
I guess nature favored it, for its cuteness, and ability to fit into a socialites purse.
In respose to Sal:
First, nobody imagines that they will make a lot of money this way. The point of offers of a bet is that one can (1) make the opponent be very clear about what they think, (2) be able to say “I told you so” in a very clear way, (3) have a test that is a program that everyone can run to verify which side won.
Under those conditions no one will bet simultaneously on both sides. No one will expect the betting to continue until substantial amounts of money changes hands. Not a good way to make a living.
No one need hold the money, because if one side loses and won’t pay up it should be very apparent, and the loss of money would be more than compensated for by people seeing that.
If a program is agreed upon and run by everyone, there should be no issue of persuading anyone that they lost.
And since the stake is more likely to be less than $100 than more than that amount, law enforcement may not be much interested.
Actually, I think that once we get close to a bet, the losing side will realize that they are going to lose and will withdraw under some pretext, so no money will change hands.
The one complication is that if one or both sides are anonymous, one does have to find a person trusted by both sides who can pass on the money. But probably that will never happen anyway.
I’d be happy do de-anonymize with Joe some time. 🙂
I’ll donate $150.00 to charity if Salvador defines cumulative selection.
Mung,
Something I’ve asked you before: Do you understand that the amazing thing would be if the numbers of offspring that organisms leave did not depend on their heritable traits?
I’ve never encountered an ID proponent who questions Malthus, and I doubt that you do. In a population tending to exponential growth, with bounds on the resources needed for reproduction, small differences of types in their average numbers of offspring can lead to big differences in their representation within the population.
(Feel free, Joe, to clean up what I’ve said.)
I’d just add, Tom, that even when a population is well below the limits set by its environment, differences in phenotype can result in differences in survival and fertility.
Darwin cast the argument in turns of Malthusian limits, but that is not really necessary for natural selection to act.
Joe Felsenstein,
I hadn’t lost sight of that, and I should have mentioned it. I was trying to home in on something simple, and hard to deny.
The main point, in any case, is that differential reproduction is the default. It seems to me that Mung is turning it into something for which a special account is required. It would be a miracle if differences in heritable traits did not result generally in differential reproduction. (Yes, I know that differences in heritable traits do not guarantee differential reproduction.)
Feel free to tweak again. I’m not pretending to be an expert in evolutionary biology. I’m learning.
Yes, there are a bunch of ID supporters here whose default comment on that is “yes, but do you have any evidence that there are differences in fitness?” (For vast numbers of traits that have been studied, yes, they have evidence). And then “but natural selection is vacuous since you can’t know what fitness is until after you see the changes”. As if our lack of knowledge makes fitness differences absent.
I guess that makes me the first!
Joe, would you care to say a word or two about the relevance of Malthus to population genetics?
ETA: Just so you know this isn’t an attempt at a ‘gotcha’ let me say up front that Malthus gets nary a mention in your book. So I’d also like to hear your comment about why not.
I suppose you aren’t counting the average mutation as a trait, and refer here only to phenotypically relevant ones. True? Otherwise, not only would Larry Moran be annoyed — he’ll be annoyed even if you mean phenotype — but anyone who accepts the idea of junk DNA would be too.
Tom, I don’t have any quarrel with some organisms leaving more offspring than others due to differences that are heritable. I would argue that is not cumulative selection. That’s just plain old selection.
Cumulative selection of the Weasel variety clearly requires a target. Even Dawkins himself admits this. He also admits that life isn’t like that.
I’ll donate $200.00 to charity if John Harshman defines cumulative selection.
I have already defined cumulative selection. Where’s my money?
Selection is cumulative if each generation starts from the results of the previous generation. That kind of, sort of, tends to happen when organisms, um, reproduce. As, being organisms, they, um, tend to.
Or does Mung have something else in mind?
Mung quoth:
Darwin and Wallace were both inspired by reading Malthus. In the meantime there has been over 100 years of work on ecology and ecological models. Most biologists know about carrying capacities, density-dependent population growth, and logistic models for population growth. And that supersedes reading Malthus.
In my online book, the relevance of this, or irrelevance, is discussed in section II.2, which argues that if relative fitnesses of genotypes remain constant we can avoid bringing the ecology in (at least for models with infinite populations). What happens to selection equations when fitnesses are density-dependent is covered in section II.9. There is some invocation of density-dependence when the Wright-Fisher model is introduced (section VII.4) and more when selection in that model is discussed (section VII.7).
It would be fairer to complain that I make too little reference to mathematical ecologists such as Vito Volterra, Robert MacArthur, and Robert May.
You didn’t provide a definition, you described two conditions under which you thought cumulative selection might occur.
And your proposed “definition” doesn’t say anything at all about accumulation. Wouldn’t that be an important part of the definition?
If also fails to describe the necessary relationship between how fitness is determined and the target.
So how much money do you think you deserve?
Every generation starts from the results of the previous generation. You may as well be saying that genetic drift is cumulative. If it’s cumulative with selection, and it’s cumulative without selection, what’s left? It’s all cumulative!
Talk about a term that adds nothing to the term to which it is attached. Cumulative genetic drift!
What’s true, I see, is that there’s a serious problem with the way I’m putting things. I have in mind genotypes or phenotypes that are defined in terms of heritable traits. But when I simplify, and go straight to “heritable traits,” I’m not excluding individual bases in the genome, which are, of course, heritable traits. So it’s not a good way to simplify — particularly when I’m trying to make the point I am.
Thanks for the nudge.
No one that I know of has ever spoken of individual mutations as being ‘traits’.
It depends on whether I can add a fee to compensate me for dealing with all the useless word-lawyering. In that case it might get kind of pricey.
So 20 bucks then? I didn’t take you for a cheap scotch kinda guy. 😉
There seems to be so little interest in defining cumulative selection, or making it quantifiable. Are the stakes just not high enough?
Let’s see if we can’t identify and agree upon the elements of cumulative selection, using Weasel as our guide.
It requires a target phrase.
It requires a population of candidate solutions.
It requires the generation of new candidate solutions.
It requires a means of judging how close to the target a given candidate solution is.
It requires an accumulation of [something] over many generations.
It requires that the population of candidate solutions become more and more like the target over many generations.
It requires getting rid of stuff that we don’t like.
It requires a favorable mutation rate.
It requires a favorable limitation on candidate solutions.
It requires a favorable means of eliminating undesirable candidate solutions.
What else am I missing?
That’s pretty funny coming from you Joe. I asked you whether artificial selection is guided. Your answer was yes, no, maybe, I don’t know, momma help me, if the shoe fits.
Is artificial selection guided? Yes, No, or I don’t know. Please.
I suspect that Dawkins gives a reasonably clear definition in Chapter 3, “Accumulating Small Change,” of The Blind Watchmaker. But I won’t be looking for quotes this evening. Perhaps tomorrow. What I can tell you for sure is that you cannot understand what he’s saying without keeping in mind his distinction of single-step selection and cumulative selection. He’s showing that the outcome of a process of cumulative selective can be highly improbable for single-step selection.
Dawkins most definitely made it clear that cumulative selection is a general mechanism, and that the monkey/Shakespeare model (“Weasel program”) is not a model of cumulative natural selection (his emphasis). He explains that biological evolution has no target when he transitions from the monkey/Shakespeare model to the biomorphs model, which does not have a target (though selection is still artificial, in the ordinary sense of the term). When you say that cumulative selection requires a target, you evidently do not recognize that the biomorphs model is his primary illustration of cumulative selection. Are you saying that because it is impossible to quantify cumulative selection in the biomorphs model, it does not illustrate the “power of cumulative selection” in a sense grasped by the general (non-word-lawyering) reader? As I’ve said to you before, The Blind Watchmaker is a work of popular science. Dawkins assumes a cooperative reader — someone who wants to understand what he’s trying to say — not an adversarial reader concerned with nothing so much as to poke holes in his “testimony” on “cross examination” (think Darwin on Trial).
Not to worry, I have it on Kindle now. I’ve just been holding it in abeyance. I’ll probably quote it myself and then all you’ll have to do is verify my quotes. 🙂
But if you do come across a quote don’t waste time typing it up. Direct me to where it is and I’ll gladly post it.
I couldn’t agree more, though I would say he assumes a gullible reader. Someone who never stops to question his story.
And I haven’t forgotten Biomorphs. That section is just full of references to search. No other section of the book contains so many references to search.
Mung,
I’ve got the Kindle edition. And I’ve already copied a bunch of passages into HTML, preparing for another post on David Glass’s paper. I’m just too tired at the moment to deal with going through it again.
I think you and I are closer to agreement that Joe and I. I believe there is a significant difference between artificial selection and natural selection. I also believe that the Weasel program is an example of the former.
If Weasel is not a model of cumulative natural selection, why can’t Joe (and so many others here at TSZ) simply admit it?
So when I ask for a definition of cumulative selection, the response should be that one needs to distinguish between cumulative natural selection and cumulative artificial selection.
I bet I could even agree with that!
Tom English,
Good response of yours on the “cumulative selection” issue. It is what Dawkins was trying to do that is important, and he was trying to show that selection in a genetic algorithm that started each generation from the results of the previous generation’s selection was much more powerful than just guessing “at random” which is how the disingenuous (or ignorant) creationist debaters were describing natural selection. That was the issue he was thinking of when talking about the power of cumulative selection.
Oh hooey.
“Guided” is your word, not mine. I was asking what you meant by it, hence my uncertainty.
I answered about artificial selection, without using the G word. The breeder has some control over the trait selected on, but not on most of the details of how it was achieved. They can see that (say) butterfat content of milk increased, but without further investigation, not the details of how this was achieved.
So you tell us whether that selection response is “guided”. Your word, not mine.
This issue was the subject of debate during the times of Darwin and Spencer. If you don’t see why this is an important objection to evolution, you don’t see much.
Anyway, the difference is this: In “guided” cases, such as breeding of domestic animals, the “target” of evolution is known beforehand and is consciously being aimed at. Once achieved, the breeding stops. This is artificial selection. In “natural” cases, there is supposedly nobody to set the target (hence “unguided”), so nobody knows in what direction to evolve to and consequently there is no conscious pressure towards it either. Given this, in what sense is natural selection a mechanism that drives evolution? In artificial selection, it’s clear that it drives the modification or adaptation of the organisms, because somebody has set the target and is working to achieve it, but there is nothing like this in natural selection.
PS. By the way, the co-inventor of the theory of evolution by natural selection – Alfred Russel Wallace – always made an important distinction between artificial selection and natural selection and defined the latter by means of this distinction. He also made a big deal about that the latter is UNGUIDED, while the former is guided. Come on JF, it’s your theory. Know it’s history!
Also, take a look here https://rationalwiki.org/wiki/Unguided_evolution “Unguided evolution is part of the scientific theory of evolution by natural selection. It is the idea that evolution is guided by one simple principle — living long enough to reproduce; that neither nature nor the process of evolution and natural selection are partial to what reproduces or how; that there is no ultimate form of life that is the “target” of all this reproduction, mutation, and selection.” It’s science, dude! And this process, being unguided and having no target, can be quite reasonably called fundamentally random. Perhaps not in the statistical or mathematical sense, but certainly in the metaphysical sense.
I’ll have to disagree with Alfred Russel Wallace regarding selection, then. The process of biological selection is the same, whether styled natural, artificial or sexual. It’s biased sampling of alleles (leading to a change in allele frequency) due to differential reproductive success in the niche. The environment is the selecting (guiding, if you like) element. Plant breeders and choosy mates just make up part of some niche environments.
Don’t want to add to the confusion, but in my mind “traits” are phenotypes, not genotypes. So there was was no problem with your intial statement.
Individual bases and mutations are not usually referred to as traits. Have to agree with Mung here.
Alan Fox,
You are actually pretty much agreeing with him. However, “plant breeders and choosy mates” are not merely “niche environments” for the following reason (according to A.R. Wallace): Natural selection selects for the survival (of the species), artificial selection for other things, such as certain pretty colors, shapes, or productivity of edible parts. In natural selection, shapes and colors and body parts of the organism are subjugated to inherent survival capacity because the organism must sustain itself while the environment wants to devour it, but in artificial selection the survival capacity becomes secondary because the breeder provides the comfiest environment for the organism where it does not have to fight for survival. These are the lines drawn by A.R. Wallace.
Never heard of the term, but hey, why not? Genetic drift can result in the build-up of gradual evolutionary change including fixation of genetic variants, and by definition it has no target.
Why do you see that “cumulative’ is a trivially obvious concept, but refuse to a accept the consequences?
There seems to be a guy calling himself Mung that dodged my challenge.
So would I.
Yes it is. What I’m missing is why you think this is a problem with the fitness concept.
Elephants having large ivory tusks that are attractive to poachers make them a target for being killed, so the frequency of large-tusk alleles in elephant populations go down, but so fast that eventually elephants are rendered extinct. So poaching is an example of natural selection against an unfit allele that used to have a very high frequency in the elephant population.
Yes, yes it is. Why is that a problem for the concept? It isn’t a problem, it makes perfect sense.
That looks like an operational definition to me.
If so, critics need to explain why that kind of definition is unsatisfactory in a scientific context.
Erik leaves out an important aspect of artificial selection (which I discussed a couple of times in this thread, one in the comment immediately preceding Erik’s). The breeder selects on the trait (say butterfat percentage in dairy cow milk) but does not have any knowledge of exactly how this will be achieved, which genes will change. Those details are indirect consequences. And that is why Darwin used artificial selection as a close analogy to natural selection.
Yup, all those slime molds and sequoia trees and algae and sponges are blissfully unaware of what is happening. If that disturbs you, my condolences.
Again, see above about how artificial selection changes all sorts of details that are not known about by the breeder.
So Erik thinks that natural selection isn’t really there? It is there just as much as the fitness differences unwittingly caused by the breeder in artificial selection.
I have heard of Wallace’s views on this. He was wrong to make such a big distinction.
It is important to convey to people how it is not random. Dawkins’s argument was countering the misleading arguments of creationist debaters, who endlessly drum into their audiences’ heads that evolutionary biologists are explaining adaptation by “random” change. They hope that their audiences will immediately conclude that evolutionary biologists have no mechanism to explain adaptation. The use of “random” by creationist debaters is deeply disingenuous.
The change may have no target or conscious goal but it is nonrandomly in the direction of higher fitness. You can be as metaphysical as you want, but just throwing around the word “random” is misleading those audiences. You really want to mislead them?
Yeah, Erik knows natural selection doesn’t work. Why? well, because no one is guiding it, but why does life need an invisible guiding force to evolve? because it wouldn’t work without it, obviously