… 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.
In order for me to wager, you would need to define what you mean by cumulative selection and find a way to quantitatively measure it. Is that an unreasonable request?
As far as funds, instead of a wager, would you object to a donation to a charitable organization? I won’t mind losing so much. 🙂
So for example, if you can define cumulative selection and make it quantifiable (measurable) in such a way that it makes no reference to a target or goal, and show that it’s operation in a WEASEL program is responsible for the success of the program, I’d put up $1,000.00. And gladly.
To be even handed shouldn’t you need to do the same for ” intelligence” ?
You’re complaining about a freeroll? Really? You can’t lose.
Mung,
I would wager100 dollars that Weasel with 1000 characters and a mutation rate of 5% would not out perform a random search. The total search time limit is 2 days. With 5 to one odds I would bet 100 dollars that it will not out perform a random search with 1000 characters and a 2% mutation rate.
Good point. We might need some way to measure search performance.
Can I suggest a Taylor Swift bet of one dollar. The kudos from winning is worth so much more.
Not complaining at all, I got no skin in the game.It is in your interest to define and quantify intelligence.
If you look at Tom English’s recent “Evolution is not search” thread (pinned as the top thread in TSZ at present), it is clear that with 5% of characters mutating every generation, and 1000 characters in all, the equilibrium distribution of matches would be too far from the target ever to come up with a complete match. That is because matching characters would mutate away from matching too often. So no, of course i would not take that bet. But if you look at another measure of success, say how many characters match the phrase after a long(ish) run, then the Weasel will beat random search — hands down. So I would happily take such a bet with that as the criterion
Cumulative selection is present when two conditions hold:
(1) Genotypes differ in their fitnesses.
(2) Each generation starts from the results of the previous one. (That condition is present in all living organisms and in models of living organisms).
So now maybe Mung can tell us when “intelligence” or a process that is “guided” is present. Then we can see if there are some processes that we can bet on.
Maybe I should just offer to donate $1000.00 if someone can define cumulative selection and provide a way to measure it.
Okay. Will you do that, then? Let’s get an independent judge we can both accept, and then you put 1000% dollars on the line. If I come up with a definition and a way to measure cumulative selection, to the satisfaction of the independent judge, you pay me 1000% dollars! Deal?
My understanding was that Mung’s assertion was that Weasels succeed because of intelligence (or because of “guidance”). And perhaps Mung is also saying that they do not succeed because of cumulative selection.
Rumraket, if you can get Mung to be clear on this I will pay you $10.
Do you believe that you have here provided a definition of cumulative selection?
Do you think she’d go along with it?
Well, I’ve provided enough for us to use in any bet. But I suspect that for it to be a “definition of cumulative selection” that would satisfy Mung, well, that may be beyond the capabilities of any mortal. Not sure why we need a terribly-precise definition anyway.
From the OP:
Don’t you think it’s important to the wager, if you are going to bet that the Weasel would succeed because of cumulative selection that you should be able to define cumulative selection?
How do we measure success?
I didn’t have time to read this. Is this about Sal starting a new website?
Is it not not obvious by now that Weasel succeeds because it was intelligently designed to be guided to find the one solution to the problem? LoL
You dirty rat.
Which is it, a weasel or a rat? I realise this is an evolution blog, but this is getting ridiculous. Vermin!
Mung:
Joe:
I’m willing to bet that this thread will devolve into an exercise in word weaseling by Mung. It appears to be headed that way already.
Safe bet for sure
I KNOW A HAWK FROM A HANDSAW
— Hamlet 2.2.403
(28 characters, just like the Weasel phrase)
QED
P.S.–It’s more interesting when you realize that Hamlet is alluding to his earlier “confusion” about shapes in the sky (camel, weasel, whale, whatever).
It, like all simulation programs, is designed. (Like all of them, it being designed is no evidence that the natural process that behaves like the simulation is designed).
It is designed to have a process of selection, analogous to natural selection. It was designed to show that such a process could get towards its goal far faster than “random” processes, thus giving the lie to creationist debaters who endlessly tell their audiences that evolution is a theory of “random” change. The audiences are expected to reject evolutionary processes, because everyone knows that biological adaptations could not have evolved by pure random wandering in genotypic and phenotypic space.
As for “guided”, I await hearing from Mung what Mung means by that, and whether natural selection processes are, according to Mung, also “guided”.
I’ll donate $100.00 to charity if you define cumulative selection. 🙂
Yes, if all success is caused by cumulative selection then we can be sure that whatever we look at was successful and was caused by cumulative selection.
With natural selection being a process analogous to intentional selection. What a tangled web we weave …
Since you didn’t answer my question, how to we measure success?
What would indicate success? It would be a shame to actually get something going only to have whether the program was successful or not left open to debate. 🙂
From the OP:
Shouldn’t be too hard for all of you. You have keiths on your side.
The alternative being that phenotypic space contains functional targets (like the vertebrate eye) that evolution (or cumulative selection) is aiming for.
But Dawkins is quick to assure is that is not the case.
Are you or are you not saying that natural selection is unguided?
I have no idea why I kept putting a % sign after the 1000. lol
Easy enough. For Weasels with a low enough mutation rate we could consider success to be arriving at the target phrase. But Weasels with a higher mutation rate never get there, because they mutate away from the phrase too often. So a better measure would be to simply count the average number of characters that match the target phrase. This is far higher, in the moderate run and the long run, than if there is no selection.
Now we await any insight from Mung into what is and is not “intelligence” or “guidance”. Mung’s words, so Mung gets to be more specific about them.
And if Mung won’t, and tries to push the question off on everyone else, then the next time Mung says people are “pretenders” because they won’t take Mung’s $10,000 bet, we just get to laugh.
Dawkins is right about that. Phenotypic space contains not targets, but fitnesses of the phenotypes (and thus of the genotypes, for genotypic space). There is no target, as shown by the development of other kinds of eyes than vertebrate eyes. There can be multiple ways “uphill” on a fitness surface, in different directions, depending on where the population is on that surface. We are trying to explain not why organisms have the best possible phenotype, but why they are as well adapted as they are.
I’m not saying whether it is or is not “unguided”, because in this discussion that is Mung’s word, not mine. You’re right to ask, but you asked the wrong person. I would be curious to know whether Mung would say that natural selection is “unguided”. I asked him exactly that question in a comment in this thread. So let’s await Mung’s clarification.
I predict you will get no guidance
I’ve stated my position a number of times, but I don’t mind at all saying it again.
I don’t know how we could possibly ascertain whether or not natural selection is guided or unguided. If you or anyone else here has a way to settle the question I am all ears.
Would you say that artificial selection is guided or unguided?
You don’t see any problem with this statement? Other kinds of eyes are also targets. So the existence of other kinds of eyes doesn’t mean that there is no target, it only means that the vertebrate eye is not the only target.
But now you’ve added an additional factor. It’s not just cumulative selection that is the cause of the success of the program. It’s cumulative selection plus other stuff.
You don’t say.
Joe Felsenstein,
Lets describe this as a filtering mechanism. Where does the raw material that is filtered come from? I think one area is genetic recombination. Would you consider this process random or deterministic?
So we need a target phrase. Add that to needing a fine-tuned mutation rate. And since the claim is that the success is caused by cumulative selection, we need a way to eliminate the mutation rate and the target phrase as being causes of the success of Weasel.
If we eliminate the target phrase Weasel won’t succeed. So we can’t say that it succeeded because of cumulative selection. Obviously, it succeeded because of the target phrase.
Likewise if we get rid of the fine tuned mutation rate. We already know it won’t succeed in that case either. Obviously, it succeeded because of the tuned mutation rate.
What else causes Weasel to succeed?
The length of the genotype string.
The possible values at each position.
The mapping of fitness to target phrase.
It’s hard to see how cumulative selection causes Weasel to succeed at all. Remove all the obvious design elements from the program and cumulative selection will succeed at nothing.
That pesky target phrase. But life isn’t like that.
I know a handsawfromahandjob.
—–Xaviera Hollander
2-69, 3- a2 +b2=c2
but his wager is that a random search would do better than the weasel algo.
A random search would have an average fitness of 35.7 = (1/28)*1000
I’m running this weasel algo with some extra printouts, (mutation rate = 0.05, number of children = 1)
https://github.com/stinaq/WEASEL/blob/master/python.py
… and the average fitness seems to converge at around 37. So in that sense it does do better than a random search
but his wager is that a random search would do better than the weasel algo.
A random search would have an average fitness of 35.7 = (1/28)*1000
I’m running a weasel algo I found in github (tried posting the link but apparently it was detected as spam) with some extra printouts, (mutation rate = 0.05, number of children = 1)
… and the average fitness seems to converge at around 37. So in that sense it does do better than a random search
Weasels have a target phrase, have selection, and have each generation start from the outcome of the previous one. The latter two are cumulative selection. How would a weasel work that does not have the selection be cumulative? Mung seems to think that the cumulative part is not essential.
I can think of two ways to make a weasel-like program that would have selection, but not have it be cumulative. One would be to choose a random starting string, and to have each “generation” start from that string. The other would be to choose a new random starting string for each “generation”. I’m putting quotes around “generation” because in these cases we do *not* use the current generation as parents of the next.
So Mung really thinks that this would be equally effective? Hmm, I think there’s a wager there …
What is a random search with a mutation rate? Ignoring fitness and skipping selection?
Oh, and fine-tuned mutation rates: mutation rates can’t be too high and have us get all the way to the target phrase. But a wide range of mutation rates would allow success by the other, more relevant criterion — an average fraction of matching characters that significantly, and substantially, exceeded 1/27.
Even if the criterion is “time to produce a perfect target sequence”, will a random search be expected to produce that sequence faster than selection with a high mutation rate? Don’t random mutations equate to a random search of sequence space for that fraction of sites that mutate, and won’t that fraction have a higher probability of hitting the exact target by chance than a random mutation of the entire sequence? Put another way, if selection gets the sequence most of the way there, won’t further random changes be more likely to produce the target than a random draw from sequence space, even if the former is still highly unlikely?
I got this wrong, it should be (1/27)*1000 = 37.03
So looks like they perform about the same