Over on the The War is Over: We Won! thread at UD, The subject of phone numbers is brought up by Bob O:
lots of things are sequences. But they can be produced in lots of different ways. Frankly, I have difficulty seeing how phone numbers mate and recombine, especially when within a longer string of sequences.
Our beloved Mung is quick to retort:
Ah, the old “I cannot imagine” defense. I could write a program in which phone numbers mate and recombine. Incredulity is not an explanation.
We all love Mung having a go at programming. Come back and walk us through it.
A few of thoughts for discussion:
- How big is the state space for phone numbers and how much functional space (viable / live numbers) is there?
- What is the ‘evolutionary’ history of telephone numbers? How big was the ‘biogenesis’ phone number?
- Is there a stepwise evolutionary history for phone numbers?
- Or is this all evidence for Evolution by design!!!1111???
Here’s a link to Bob O’H’s comment if you want to see the context.
_________________________________________________________________
One thing is not like the other.
This is perhaps the greatest problem with IDists’ “arguments” in general. Their analogies compare very different things (they routinely “analogize” technological “evolution” and biologic evolution, ignoring the telling differences), and their “gotchas” are of the sort above, where one notes that numbers do not (of their own accord) act like organisms and their DNA, and it’s pointed out by some IDist that one could program them to (sort of) do so.
Logic–not something that is treated well by IDists.
Glen Davidson
This might be out last ever thread as we’ve lost….
I think we’re used to it by now.
We lose, the science blithely continues, then we lose again, the science still works, we lose again, and they still can’t even pretend to do science with creationism/ID.
I’m somehow still surprised at how much a lot of them actually believe in their victories, however. No matter how incompetent they seem, it’s still odd just how backward their entire perspective really is. It’s like, you can’t believe that, can you? Yet they do.
Glen Davidson
Glen,
I think the key lies in the formidable defenses they deploy against new (to them), threatening ideas. And they aren’t passive defenses — they’re more like an immune system that hunts down and annihilates threats to their ideology.
Consider the junkyard tornado arguments. Many of the folks making those arguments have been involved in the evo debates for more than a decade, and evolutionists have been explaining the central flaw to them that entire time. There is no way you can miss the point for ten years unless you are trying very hard not to get it.
I agree. That is an astute observation.
compare with Barry Arrington
Wow, what turn of events!
Skimmed the piece. Basically, X happens, therefore X is all that happens.
Hahaha
To take the phone number questions seriously, which I’m not sure I should, the most straightforward genotype space for them is just to treat them as a string of digits (10 of them in my country), each “gene” being one digit, and having 10 possible “alleles”.
Recombination is easy to do, though one has to choose a probability of recombination between successive digits. Mutation could either be to the adjacent digit (so 7 could mutate to either 6 or to 8, and 9 could mutate to either 0 or 8), or else mutation could be to any one of the 9 other digits.
If one had a genetic algorithm, the real question would be how fitness was assigned. 1 if the phone number belonged to someone, and 0 if not? Or 1 if it is one particular person’s phone number, near 0 if not. In the former case many phone numbers would have fitness 1, and it would be fairly easy to evolve to fitness 1. The latter case is the classic “needle in haystack” fitness function. It would be very hard to get anywhere near the correct number.
In such cases evolution won’t get much of anywhere, and we do indeed find that I can’t flap my arms and fly 500 miles per hour.
If instead we also have, in the needle-in-a-haystack case, phone numbers assigned fitness which is higher the closer they are to the right number, then evolution could climb to higher fitnesses easily. Say having fitness be
, where 
is the number of digits different from the needle-in-a-haystack phone number.
So it depends on what type of situation you want to model.
There’s also the question of how phone numbers “evolved”. My expert on this is Wikipedia, which says that in the original cases numbers from 1 to 200 were assigned to 200 users in a telephone exchange. As the number of users got bigger the numbers went to 4 digits, then to 5.
They don’t say how that was done but I bet it was by prepending digits, especially zeros, to the old users, to avoid confusion. So 401 became 0401 and then maybe 2-0401. Before all-digit dialing the exchanges had names like Cypress which became two-digit numbers in about 1919. So CY became 29 and one’s number became 292-0401. Then when automatic long-distance dialing was introduced, area codes like 206 were prepended. Around here that was apparently in about 1950.
All this was design, designing a “space” that was not designed to evolve but to avoid reassigning people’s numbers too often as more users needed numbers.
All of which proves what? That biological evolution did not choose our phone numbers.
Joe Felsenstein,
I’m not advocating it seriously 😉
A ‘fit’ number could be:
a) The right length
b) assigned to someone (anyone)
c) assigned to someone specific.
We go from trivial to needle in a haystack.
1919? Growing up in Forest Hills, Queens in NYC my phone number was BO3-9853 where the BO stood for “boulevard’ This was almost 50 years ago.
When talking about IDers I dont think it helps to just say ‘they’ Theres a huge difference between the anonymous commenters on UD and Meyer,Behe and Dembski.
I grew up in Syracuse, NY, in the 1950s, where 446 was GI-6 or Gibson 6.
Yes, as a mnemonic the letters were still in use then. But they were printed on the rotary dial in groups next to figure. Since about 1919.
Dearborn 5-2750. ‘Nuff said…
There is selection for phone numbers.
Vanity phone numbers are bought and sold. Many numbers can be presented as “meaningful” words or memorable strings.
There’s a law office down the street from me with the phone number 777-7777.
Now here’s a cute problem: suppose we had a Weasel program that had no knowledge of the potential meaning of number strings, but was subject to feedback based on the price of vanity numbers. What would we expect to happen?
So when you dialled DE on a rotary-dial phone, or when you put a touch-tone phone into “pulse” mode, what happened? Did it transmit ASCII characters? No, of course not.
It sent two chains of 3 pulses. Just what the dial position of D and E indicated. And that was true since about 1919 (in the U.S., anyway).
Before that you had to have an operator connect you to the Dearborn exchange. After 1919 you could dial it yourself.
‘Nuf said.
Still tacked to the wall in my mom’s garage:
petrushka,
In my area code it’s a taxi service.
My guess is not much. Weasel needs hills to climb, and I don’t think you’d get that in a vanity number fitness landscape.
Does it still work – does the color change to reflect the weather?
Joe Felsenstein,
Very good summary:
Go from 10 digits to 13 digits and now both the single phone number and multiple number cases become a needle in the haystack and both need a target or Mung’s new project;
colewd,
telephone Weasel 🙂
Seems like a fair question.
RB,
It does, actually!
The letters were an abbreviation of the name of the Central Office switch where the numbers originated.
True that they were the names of the switching offices.
However I’m not sure what it means to say that the numbers “originated” there. Surely they came from ancient India and the Middle East. 🙂
Anyway I think you are not disagreeing with the number having been assigned about 1919.
So it’s agreed then? The UD post and Mung’s prattling are so breathtakingly detached from reality that talking about our childhood phone numbers is more rewarding? Makes sense to me.
If Richardthughes will provide his phone number I will in fact write a program to evolve his phone number.
If Stormfield will also provide his phone number I will in fact write a program to evolve Richie’s phone number into Stormfields phone number.
It probably can’t be done, but I bet no one here would bet on it.
I bet that not even Joe F. would bet $100.00.
Here’s a challenge that no Weasel lover could possibly resist.
Write a program that demonstrates “the power of cumulative selection” using Richie’s phone number as the target.
For extra credit, pretend that you don’t actually know his phone number.
Here’s a challenge for Joe F., who has at least 100 bucks to spare.
Write a program that can evolve Richie’s phone number into Stormfield’s phone number given that Stormfield’s phone number is not known
Write a program that can evolve Richie’s phone number into Stormfield’s phone number given that Richie’s phone number is not known.
What observed biological process would this model? Why do you think EAs would be suitable for these “needle standing in a desert” problem domains?
I fail to see how this is relevant unless you believe all evolution is biological evolution. Is that what you believe? You believe that it is not possible to model evolution in silico? Nah, you don’t believe that. Go away Patrick.
Dawkins Weasel.
Apparently Mung did not read my earlier comment in this thread about which ways of assigning fitnesses to phone numbers would lead an evolutionary algorithm to work well.
It is just like the earlier 100 dollar bet I offered. Mung had been badmouthing the Weasel program and calling for a “test suite”. It was obvious that trying to agree on a test suite was a waste of time. So I simplified matters by offering a 100 dollar bet that a simple Weasel program would work.
Mung did not take this up, but instead loudly dared everyone to bet 10,000 dollars. Mung was not, however, betting that Weasel would not work. But betting that Weasel would work!
Now Mung bravely challenges me to bet 100 dollars. Where I would be betting that my statement about finding phone numbers when fitnesses were needle-in-a-haystack fitnesses was wrong. And Mung would be betting that my statement was right.
This is nothing if not creative.
It is good to see that Mung is so willing to bet that Weasel-like genetic algorithms work when fitnesses rise gradually toward the optimum. And so willing to bet that Weasels won’t work when there are needle-in-a-haystack fitnesses.
Mung, it turns out, is a great admirer of the Weasel.
It’s relevant if you intend to make the argument that the failure of an EA to solve the problem you pose demonstrates some failure of biological evolution. If that’s not where you’re going with this, I don’t see your point.
That’s a different kind of problem. Dawkins’ Weasel’s fitness function rewards partial matches and differs significantly from most EAs in that it has the solution hard coded into it. As has been repeatedly explained here, it is only intended to demonstrate cumulative selection.
Your problem does not reward approximations to the goal. Either you have a fundamental lack of knowledge of EAs or you are looking to try to score cheap rhetorical points in your own mind.
This is factually incorrect, but even so, who ever doubted that a simple Weasel program would work?
A $100.00 bet over something not in dispute. I offer that same bet.
Joe, you’re better than the typical riff-raff here at TSZ, so why on earth do you think the bets you’ve proffered address anything I’ve ever claimed? I’d really like to know.
A guided search performs better than a blind search. Can we agree on that? Do we need to wager on it?
So you claim that evolution in silico must necessarily be different from evolution in vivo?
You must either support your claim or retract it. Unless you’re a hypocrite.
You have an unfortunate tendency to fail to include context and post non-sequiturs.
The context here is that you’re proposing a needle-standing-in-a-desert fitness landscape. EAs require a smoother fitness landscape. That is what we observe in biological evolution.
If your goal is to argue that the failure of an EA to solve the problem you pose demonstrates some failure of evolutionary mechanisms in general, you’re simply wrong. If that’s not your goal, I fail to see what point you are trying to make.
I have made the reasons for that conclusion clear. Which is it?
@Mung:. So if you always acknowledged that the Weasel program works, why all those demands you made for a “test suite”?
Was that your way of agreeing with folks here?
A program can work without meeting all the design requirements.
That’s true, right?
You’ve written some software. Have you ever written software tests for your code, or do you test it all manually? Do you just hope it works? Or if it appears to do what you think it ought to do you just accept that as good enough?
How do you validate your code? If you make a change to your code, how do you validate that you haven’t broken anything? Or do you?
Are you one of those folks who releases your code and has the customer do the testing?
If the point of the Weasel program was to demonstrate that guided search performs better than blind search, surely that was known already and was news to no one.
You forgot creationist debaters, who often make a concerted effort to persuade their listeners that evolution by natural selection “is random”. They want their audiences to reject natural selection as an explanation for adaptations, because everyone knows that purely random changes cannot be the explanation for nonrandom phenomena like adaptations.
The whole point of Dawkin’s Weasel example was to refute the notion that natural evolutionary processes are just making changes “at random”.
Mung,
The point of the Weasel program is to demonstrate the power of cumulative selection. Despite your best (albeit puny) efforts, you’ve been unable to cast doubt on its ability to do so.
Your ass has been kicked and rekicked by a mere Weasel.
Is it possible to test that claim?
But Weasel does make changes “at random.” If you don’t believe me just talk to keiths. Randomness is at the very heart of “natural evolutionary processes.”
Not that it matters, since the Weasel program is not natural. It’s designed.
Those two rather creaky old talking points?
OK, I’ll bite:
1. All computer simulations of anything are designed. If I do a computer simulation of landslides, does that mean landslides are designed? Does the fact that landslides aren’t designed mean that a designed computer simulation cannot help us investigate their behavior? A Weasel simulation, designed as it is, is designed to imitate evolutionary processes. You don’t understand that?
2. So you actually agree with the misleading points made by creationist debaters who accuse evolutionary biologists of putting forth a theory which accounts for adaptation by pure random change? Those creationist debaters are misleading their audience in a major way. You mean you don’t understand the distinction between a simulation that wanders randomly and one that has selection? You mean you miss the point of the Weasel?
Mung, still auditioning for that QA1 gig…
What does the Dawkins Weasel program simulate?
No.
How do you know that landslides are not designed? You assert it as a fact. How do we test your claim?
How do we test the claim that your simulation actually simulates a landslide?
What does the Weasel program simulate, and how do we test that claim?