Dawkins’ WEASEL: Proximity Search With or Without Locking?

_{William Dembski
March 16, 2009

Darwinism, Evolution, Informatics

3}_{Categories
Darwinism
Evolution
Informatics}

Share: Facebook; Twitter/X; LinkedIn; Flipboard; Print; Email

On pp. 47-48 of THE BLIND WATCHMAKER, Richard Dawkins gives two runs of his WEASEL program (note that there were typos in both initial seeds — one had 27 characters, the other 29 whereas they should have 28; I’ve corrected that). Here are the two runs using the Courier typeface, which assigns equal width to each character (spaces are represented by asterisks):

WDL*MNLT*DTJBKWIRZREZLMQCO*P
WDLTMNLT*DTJBSWIRZREZLMQCO*P
MDLDMNLS*ITJISWHRZREZ*MECS*P
MELDINLS*IT*ISWPRKE*Z*WECSEL
METHINGS*IT*ISWLIKE*B*WECSEL
METHINKS*IT*IS*LIKE*I*WEASEL
METHINKS*IT*IS*LIKE*A*WEASEL

Y*YVMQKZPFJXWVHGLAWFVCHQXYPY
Y*YVMQKSPFTXWSHLIKEFV*HQYSPY
YETHINKSPITXISHLIKEFA*WQYSEY
METHINKS*IT*ISSLIKE*A*WEFSEY
METHINKS*IT*ISBLIKE*A*WEASES
METHINKS*IT*ISJLIKE*A*WEASEO
METHINKS*IT*IS*LIKE*A*WEASEP
METHINKS*IT*IS*LIKE*A*WEASEL

These runs are incomplete. The first, according to Dawkins, required 43 iterations to converge, the second 64 (Dawkins omitted the other iterates to save space).

As you can see, by using the Courier font, one can read up from the target sequence METHINKS*IT*IS*LIKE*A*WEASEL, as it were column by column, over each letter of the target sequence. From this it’s clear that once the right letter in the target sequence is latched on to, it locks on and never changes. In other words, in these examples of Dawkins’ WEASEL program as given in his book THE BLIND WATCHMAKER, it never happens (as far as we can tell) that some intermediate sequences achieves the corresponding letter in the target sequence, then loses it, and in the end regains it.

Thus, since Dawkins does not make explicit in THE BLIND WATCHMAKER just how his algorithm works, it is natural to conclude that it is a proximity search with locking (i.e., it locks on characters in the target sequence and never lets go).

Interestingly, when Dawkins did his 1987 BBC Horizons takeoff on his book, he ran the program in front of the film camera:

www.youtube.com/watch?v=5sUQIpFajsg (go to 6:15)

There you see that his WEASEL program does a proximity search without locking (letters in the target sequence appear, disappear, and then reappear).

That leads one to wonder whether the WEASEL program, as Dawkins had programmed and described it in his book, is the same as in the BBC Horizons documentary.

In any case, our chief programmer at the Evolutionary Informatics Lab (www.evoinfo.org) is expanding our WEASEL WARE software to model both these possibilities. Stay tuned.

Comments

Mr Kellogg, re 102: Mr Musgrave is misleading you. (Utterly unsurprising for PT.) Look at Apollos' code form the previous thread and you will see just how short, relatively speaking an explicitly latched Weasel is. I don't know if the EIL code is there at the site, as Atom can confirm. In fact, Dawkins states that within each generation the closest to target is preserved. That means that he code has to have in it an assessment of proximity, which is going to be based on relative locations of letters to the target in any case. You already have the information you need to latch, just to get to the champion per generation. Also, on the math at EIL, latched searches -- whether explicit or implicit makes no difference -- will as median run to 98 generations. This is consistent with "good" runs of 40 - 60. GEM of TKI PS: JT at 105: Do words like “of” have meaning on their own? Clearly they do: Without a much wider context,t eh glyphs or sounds implicated are meaningless. That wider communicative and indeed social context is massively based on intelligence and intentionality.kairosfocus_{March 20, 2009
March
03
Mar
20
20
2009
12:53 AM
12
12
53
AM
PDT}

PS: As for antievo and pandas thumb, I refuse to have anything to do with these notoriously disruptive and destructive fever-swamp of evo mat sites; strong cases in point of precisely the kind of destructive selective hyperskepticism and rhetorical manipulation and championing of the morally indefensible that is ripping our civlisation apart. If there is any merit in their arguments, GLF must present them here -- without the nasty ad hominems, strawmen and red herrings, not to mention cherry-picked one sided analyses that major on minor distractive points to duck the main issue. (Indeed, the very Weasel program itself is a case in point: targetted search is precisely a begging of the question Hoyle et al have raised -- getting TO shores of functionality, before hill-climbing to optimality can proceed. rewarding target- proximity of confessedly non-functional strings is exactly what should not have been seen, but was.)kairosfocus_{March 20, 2009
March
03
Mar
20
20
2009
12:41 AM
12
12
41
AM
PDT}

Onlookers: Re, GLF,100 : Still creating strawmen to destroy?Perhaps it would have been better for you, overall, if you had just said way back when “why yes, George. The letters do appear to latch but that’s because of the small sample size I’m looking at and not any explicit latching code” Selective hyperskepticism, turnabout accusation form. And ecvidently he is broadcassting a twisted wversion of whart has happened here at UD to his co-advocates of evo mat. About pa\r for their course I'd say. So, let us correct, for those interested in the truth of the matter: 1 --> The above is in response to my: "Weasel is utterly unconnected to the real world, where the issue is that targeted search that rewards non-functional configs based on mere increments in proximity to a target, have nothing to do with the alleged designing capacities of chance variation plus natural selection across competing FUNCTIONAL sub populations." 2 --> My summary -- since Dec last -- is an accurate summary of Dawkins' own description as cited many times on Weasel from BW, ch 3. In short, there is no strawman. At least, on my part. For, targetted search that rewards mere proximity of non-functional configs ducks Hoyle's challenge of getting TO shores of islands of functionality, and is utterly irrelevant to the vaunted BLIND watchmaker of natural selection. this last, is said to select by differential success of functioning life forms. No function, no differential reproduction; no natural serelction. WEASEL is not about the blind watchmaker but instantiates the power of intelligent design, here pursuing a rhetorical ultimate purpose. So, we must not let turnabout rhetoric distract us from the plain truth of the matter on the merits, regardless of the outcome on explicitness of observed o/p latching circa 1986. 3 --> Next, the claimed "small" sample size constitutes 300+ letters that could change [count 'em above], with 200+ being latched on output; never does a correct letter revert in he sample; which is avery low probability event indeed if the Weasel of 1987 is more representative of output circa 1986. On inference to best explanation, the output as published is a likely "good summary" on the Weasel's performance, and it with high probability implies the outputs are latched. So, GLF is here distorting the record on the o/p of the 1986 Weasel pgm and the extent of the sample used form Mr Dawkins' published "good" results. GLF is refusing to accept that a sample size of 300+ in a conte4xct of "good runs" being published, is not a small one by the law of large numbers standard. 4 --> From the previous thread, 346 - 7, it will be evident that I proposed two alternative models for that observed o/p latching I noted on since Dec last: T2 -- explicit latching, and T3 -- implicit latching. On the 1986 o/p alone explicit latching is the best -- most parsimonious -- explanation. 5 --> On the principle of charity on testimony, I have accepted that per recent reports first published here in this thread above, Mr Dawkins did not EXPLICITLY latch. But that simply means that on further best explanation, he implicitly latched. The observed latching is the dominant feature of the o/p circa 1986, and in fact the implicit latching thesis has the problem that the letters that flick back circa 1987 then rapidly revert, hence the "winking" effect. On mere probability, that should not be likely. 6 --> Further to this, Apollos has shown that explicitly latching code can also revert temporarily. That is, only credible code is decisive, though on the evidence we can draw some fairly suggestive inferences. 7 --> E.g. as discussed above in this thread, the co-tuning of mutation rates [about 5% will give 1 - 2 letters per member of a generation] with an optimised population size will give effective latching, as 24% of all members will be no-change at that mutation rate with a string length of 28 characters. then,w e see that on observation the reported runs take 40 - 60+ gens to reach target for "good" runs. This means that > 50% of time, no-change wins in a generation. 8 --> Choose a population scaled so that we will get some typicality, perhaps 20 - 30 at base [just off law of large numbers as a pop size to start from], and small enough so that out in skirt cases of 1, 2 or 3 correct letters do not dominate outcomes, and you will [a] latch correct letters, and [b] progress to the target in the sort of range as published instead of within 30 gens. 9 --> So, the evidence is that the primary issue is settled independent of the debate [and I almost never use this in a positive sense; this is no exception] we have had over whether or no Weasel explicitly or implicitly latched the o/p circa 1986. Weasel is simply not blind watchmaker, period. And, with a sample size of 200 latched letters in 300 chances for letters to change, we are seeing a dominant pattern of o/p latching. On the current balance of evidence [giving heavy weight tot he reported testimony from Mr Dawkins] I am willing to go along with implicit latching, though I would like to see an explanation of the winking effect on flickbacks of correct letters circa 1987. ____________ GLF's turnabout attempt backed up by refusal to address the brute fact of a sample size of 300+ letters with 200+ showing the secondary phenomenon he has chosen to debate over in order to get away from the implications of his wager argument attempt, fails. So, let us note the distruptive pattern of his threadjacking behaviour, and not reward him by followig the schools of wriggling red herrings lead out to ad hominem soaked strawmen ignited to cloud and poison the atmosphere required for progressive discussion. GEM of TKIkairosfocus_{March 20, 2009
March
03
Mar
20
20
2009
12:29 AM
12
12
29
AM
PDT}

I personally want to write a program that attempts to reconstruct an entire text using only random incremental additions of words from that text. It seems obvious that should work though. That would be a complete waste of time- obviously it would work.JT_{March 19, 2009
March
03
Mar
19
19
2009
09:48 PM
9
09
48
PM
PDT}

ericB - I don't want to be an unequivocal advocate until having more time to review all this. I personally want to write a program that attempts to reconstruct an entire text using only random incremental additions of words from that text. It seems obvious that should work though. You seemed irritated or antagonistic towards these ideas, but to me they were at least interesting if not compelling.JT_{March 19, 2009
March
03
Mar
19
19
2009
09:34 PM
9
09
34
PM
PDT}

ericB [107]: The problem is that Zachriel presumes to treat any word it finds in the dictionary as having functional value in isolation. What are you saying? That supposing some hypothetical word is difficult to interpret as "functional" that invalidates the entire illustration? Dictionaries are constructed from usage within full blown English taken as a whole. I honestly don't know what you mean. Some people communicate quite adequately with a vocabulary of a couple of thousand words or so (I suspect - maybe less) - children and toddlers with even far less. Also are you saying there was never a time when the vocabulary of English was not far far less than it is today? Did words have any meaning back then? What about the limited vocabuluary in the vocalizations of animals. Meaningless as well? I expect you know that proteins that have function in the whole human body may nevertheless be useless when taken in isolation, apart from the support of certain cooperating proteins. Obviously subcomponents have to be functional enough to survive before being incorporated into larger functions. I would tend to think in early stages of life there wouldn't have been a lot of competition, so a lot of niches. A lot of neutral genetic info squirelled away in various corners not hurting anybody and slowly multplying. It is plain that if Zachriel selected one of these in isolation, based merely on its appearance in a dictionary of functional protiens, that this would be cheating as an analogy to biological evolution. It selects based on future function in the eventual whole body. I don't know anywhere Zachriel goes to a dictionary of functional proteins and tries to establish a strict parallel with the English Language. If you're just stating that its not proven yet nature is like human language no one's said otherwise. But the idea that human language would be more powerful and more expressive and more flexible than nature itself seems hard to believe. The fact that Zachriel uses not just one but a whole dictionary of many available selectable targets does not save it from the plain fact that its relevance to biological evolution is an illusion So now you're just asserting that its relevance to biology is an illusion. That's a type of argument from ignorance as well - "Until conclusively proven I will assert its an illusion." One of its core problems, just as with WEASEL, is the fact that it makes no discrimination between present and future utility. This is just categorically false. English words do have utility in isolation - and utility in many sorts of combinations. No one has purported to prove as of yet that nature is identical to human language, however. BTW - it occurs to me to ask whether you've actually reviewed those websites I mentioned previously [57], since you're discussing this.JT_{March 19, 2009
March
03
Mar
19
19
2009
09:18 PM
9
09
18
PM
PDT}

Arthur Smith, it seems at least that JT is not yet convinced. I suspect there are others who do not realize that these programmed word games are irrelevant as an analogy to biological evolution. But if we sincerely only wanted to show that cumulative selection that is unrelated to biological evolution is more effective than random selection, we wouldn't need a program for that. To find a password of N letters by chance is to choose from 26^N possibilities (or 52^N if case sensitive). If instead I can discover them one at a time in isolation and cumulatively select them, even a brute force search need only consider 26*N possibilities (or 52*N for case insensitive) to completely examine all possibilities. As N grows, the difference becomes huge, and as you know proteins are quite long. The advantage of non-biological, cumulative selection (where present function is not a concern) is beyond dispute. However, that is irrelevant to evolution of proteins, due to the nasty problem of needing present selectable function on the whole protein, combined with the predominance of amino acid sequences that are not functional as proteins.ericB_{March 19, 2009
March
03
Mar
19
19
2009
08:38 PM
8
08
38
PM
PDT}

David Kellog: JT [105], I’ll add to that list “A” — the title of an important long (800 page) poem by Louis Zukofsky. What about the flavor enhancer, "Accent". That's not even a character.JT_{March 19, 2009
March
03
Mar
19
19
2009
08:25 PM
8
08
25
PM
PDT}

JT, I've never claimed there are no single words that might be justified as having functional value in isolation. The problem is that Zachriel presumes to treat any word it finds in the dictionary as having functional value in isolation. Dictionaries are constructed from usage within full blown English taken as a whole. This means that Zachriel is illegitimately basing selection upon a word's value in the context of the entire English language. That is cheating by looking at future utility, whereas as Arthur Smith said, "evolution is an untargeted process, blind to the future". You wrote:
"Let’s start just with english (and not consider biology for the moment)..."
If you think Zachriel has any relevance for biological evolution, why not consider biology? That is where its error becomes the most obvious. I expect you know that proteins that have function in the whole human body may nevertheless be useless when taken in isolation, apart from the support of certain cooperating proteins. It is plain that if Zachriel selected one of these in isolation, based merely on its appearance in a dictionary of functional protiens, that this would be cheating as an analogy to biological evolution. It selects based on future function in the eventual whole body. Do you disagree with that analysis? If so, where is the biological case mistaken? Yet this is exactly what Zachriel does with English words within the English language. The fact that Zachriel uses not just one but a whole dictionary of many available selectable targets does not save it from the plain fact that its relevance to biological evolution is an illusion. The problem is not removed by having more targets to select. One of its core problems, just as with WEASEL, is the fact that it makes no discrimination between present and future utility.ericB_{March 19, 2009
March
03
Mar
19
19
2009
08:20 PM
8
08
20
PM
PDT}

JT [105], I'll add to that list "A" -- the title of an important long (800 page) poem by Louis Zukofsky. Also, Zukofsky's collected critical essays are divided into three sections, entitled "for," "with," and "about."David Kellogg_{March 19, 2009
March
03
Mar
19
19
2009
08:01 PM
8
08
01
PM
PDT}

Let’s start just with english (and not consider biology for the moment): Do words like “of” have meaning on their own? Clearly they do: If… - movie title; “One” - song “Yes” - group. (I once heard some old-timer mention the above titles from some bygone era.)JT_{March 19, 2009
March
03
Mar
19
19
2009
07:51 PM
7
07
51
PM
PDT}

SteveB [65]:
wsoldin_je_d_swodx Now, consider the following “mutations” to the original, A and B. A wsocdin_je_d_swodx B wsoldin_je_d_srodx You, representing natural selection, do not know the target—in fact, you don’t even know what a target is, as you have no goals. Additionally, you cannot predict, cannot reason, and certainly cannot write computer code; you don’t have any lifelines and you can’t buy a vowel. All you can do is evaluate the “fitness” of each variation in the here and now. Your options are: select A, B, both, or neither. Which do you choose and why.
Let's say the original "wsoldin_je_d_swodx" is an eye. A) distorts the lens incrementally resulting in incrermental increase in image resolution., B) results in a corrosive acid being introduced into the eye fluids. -------------------------- Patrick [79]:
That may be doable for this GA but what about “Pneumonoultramicroscopicsilicovolcanoconiosis” (360 informational bits) or, better yet since it’s more relevant Dembski’s work (UPB), the word “Lopadotemakhoselakhogaleokranioleipsanodrimhypotrimmatosilphiokarabomelitokatakekhymenokikhlepikossyphophattoperisteralektryonoptokephalliokigklopeleiolag?iosiraiobaph?traganopterýg?n” (1464 informational bits). I’m not going to even try and look for functional intermediates. And I’d add that none exist, although the entire word consists of functional components. So someone could argue that an indirect pathway could duplicate all of them from other words and somehow assemble them into a coherent whole.
Those words above would never be specifically targetted themselves. There's only incremental increases in word length while maintaining legal words. Any conceviable word of the length above would be just as amazing as any other. And actually the way words of that length can be directly parsed and understood from their latin components to discern their meaning, a lot of the words randomly formed from such roots would be acceptable as descriptive of something, whether or not anyone had actually coined them yet or not (IMO). So when you get to a word of that length - its really like a sentence only with much less strict syntax. ---------------------- EricB: Let's start just with english (and not consider biology for the moment): Do words like "of" have meaning on their own? Clearly they do: If... - movie title; "One" - song "Yes" - group. Think of album titles, advertizing slogans and campaigns and so on. Phrases and sentence fragments are used all the time to convey meaning. Infants and toddlers also use words and sentence fragments to convey meaning and are understood. In normal conversation, people use fragments as well. Poets will slice and dice language. It is really an arbitary convention that legal sentences have to have a subject and verb for example. Groups of words and even single words most assuredly do have meaning in English. And complex words and sentences can be built up at random very easily from these smaller components in English - with no planning involved for future use. This results from the fact that smaller words and phrases can be used in millions of different ways, so that utilizing one phrase isn't locking us in to only one future target. (And just to reiterate, all the individual words and small phrases are useful on their own as well). So the applicablity to nature would be is if it as well had elementary components that were functional on their own and were functional in many other contexts as well. If this implies nature would have to be much like a human language itself, then there's your teleological argument right there, and so be it.JT_{March 19, 2009
March
03
Mar
19
19
2009
07:14 PM
7
07
14
PM
PDT}

Arthur Smith:
Pointing out the obvious, that it is not an analogy for natural selection, will not have any impact on the validity of evolutionary theory.
No but pointing out that "evolution is an untargetted process, blind to the future", does have an impact on its validity. And that is why no one can model "evolution". It is an untestable concept- "untargeted process, blind to teh future". It may make for interesting philosphy but it doesn't belong in a science classroom. You rock Arthur! Thank you.Joseph_{March 19, 2009
March
03
Mar
19
19
2009
05:16 PM
5
05
16
PM
PDT}

Does Ian Musgrave consider that locking takes place regardless of any specific coding given a target, a small enough mutation rate and a large enough sampling size? That alone ensures that at least one exact copy of the "parent" will be present in each generation. I have already gone over this. So what is so special about Ian?Joseph_{March 19, 2009
March
03
Mar
19
19
2009
05:09 PM
5
05
09
PM
PDT}

The Panda's Thumb post is interesting. Here's what Ian Musgrave says:
I’ve gone back and done a head to head comparison myself between a program with no “locking” (all letters in any given string have a chance to be mutated) and one with “locking” (where the matching letters are preserved against mutation). Trying to implement “locking” al la Dembski proved too hard. You have to keep indices of the letter locations and keep updating them. It is such a pain in the bottom to try and do this that I cannot imagine Dawkins even wanting to try and program a “Locking” implementation in GBASIC. Remember, Dawkins weasel was a quick and dirty program bashed out in a short time. To implement “locking” I just kept a copy of the parent string unmutated (after all, in the real world not every offspring has mutations in genes of interest).
His conclusions?
“Locked” runs finished earlier, on average [sic] but most of the trajectory of the run was determined by mutation supply. As you can see, runs done with locked and unlocked versions fell within the error bars of each other, for runs that set the Offspring number at either 100 or 30.
He says some other things, but I'll leave it to others to go there.David Kellogg_{March 19, 2009
March
03
Mar
19
19
2009
04:56 PM
4
04
56
PM
PDT}

KariosFocus
That is, Weasel is utterly unconnected to the real world, where the issue is that targeted search that rewards non-functional configs based on mere increments in proximity to a target, have nothing to do with the alleged designing capacities of chance variation plus natural selection across competing FUNCTIONAL sub populations.
Still creating strawmen to destroy? Perhaps it would have been better for you, overall, if you had just said way back when "why yes, George. The letters do appear to latch but that's because of the small sample size I'm looking at and not any explicit latching code" I mean. There is now a thread at Pandas thumb talking about how Dr Dembski is finally going to get round to publishing a "correct" version of Weasel, finally, on evoinfo. And of course we have the upcoming paper to look forwards to which it's suspected uses the "incorrect" (or latching) version of Weasel in some way. I notice you have not made an apperance at the thread discussing the actual math behind Weasel either at PT or ATBC http://www.antievolution.org/cgi-bin/ikonboard/ikonboard.cgi?s=49c2d3d791798009;act=ST;f=14;t=6034 http://pandasthumb.org/archives/2009/03/weasles-on-para.html I look forward to seeing you at either of those places Kariosfocus. If you want to discuss the math further, go there. If you want to repeat yourself ad nauseum, stay here. Under discussion is the probablity of a correct letter appearing and not changing. If you are a true believer in knowledge, in taking the chance that you might be wrong, you might learn something. And on the subject of ad nauseum, FSCI, instead of talking about it and typing tens of thousands of words defending it to people asking the same questions over and over where you give the same answers over and over, why don't you do something with it instead? You know, use it? Put your money where your mouth is. Publish.George L Farquhar_{March 19, 2009
March
03
Mar
19
19
2009
04:22 PM
4
04
22
PM
PDT}

@ Gpuccio I agree with your assessment of Zachriel's intellect. I should have mentioned it.Arthur Smith_{March 19, 2009
March
03
Mar
19
19
2009
12:12 PM
12
12
12
PM
PDT}

Dawkins developed Weasel 30 years ago to demonstrate the difference between entirely random and cumulative selection. He said himself "it was a bit of a cheat" and, unlike Weasel, evolution is an untargetted process, blind to the future. It is a piece of history. Evolutionary theory does not rely on Weasel for support. Pointing out the obvious, that it is not an analogy for natural selection, will not have any impact on the validity of evolutionary theory.Arthur Smith_{March 19, 2009
March
03
Mar
19
19
2009
12:10 PM
12
12
10
PM
PDT}

Lots of ink has been spilled here about the goal/target of the weasel program. Make no mistake about it: the program would not work without a future target specified by Dawkins or his programmers. Consider this fact in light of the quotations below (my emphasis). IMO, what appears to be happening here is that goals, targets and the like are expressly not allowed when evolution/NS is discussed at the 30,000 ft-level, but then teleology gets smuggled into the bowels of the weasel program--by Dawkins, not by me--when it suits his needs.
Adopting this view of the world means accepting not only the processes of evolution, but also the view that the living world is constantly evolving, and that evolutionary change occurs without any ‘goals.’ The idea that evolution is not directed towards a final goal state has been more difficult for many people to accept than the process of evolution itself. (Life: The Science of Biology by William K. Purves, David Sadava, Gordon H. Orians, & H. Craig Keller, (6th ed., Sinauer; W.H. Freeman and Co., 2001), pg. 3.) The ‘blind’ watchmaker is natural selection. Natural selection is totally blind to the future... Humans are fundamentally not exceptional because we came from the same evolutionary source as every other species. It is natural selection of selfish genes that has given us our bodies and brains … Natural selection is a bewilderingly simple idea. And yet what it explains is the whole of life, the diversity of life, the apparent design of life.(Richard Dawkins quoted in Biology by Neil A. Campbell, Jane B. Reese. & Lawrence G. Mitchell (5th ed., Addison Wesley Longman, 1999), pgs. 412-413.) Nothing consciously chooses what is selected. Nature is not a conscious agent who chooses what will be selected... There is no long term goal, for nothing is involved that could conceive of a goal.(Evolution: An Introduction by Stephen C. Stearns & Rolf F. Hoeckstra, pg. 30 (2nd ed., Oxford University Press, 2005).)
SteveB_{March 19, 2009
March
03
Mar
19
19
2009
08:52 AM
8
08
52
AM
PDT}

EricB: Did you think that anyone doubts this? Yes. Did you read SteveB's comment to which I responded? (my apologies to StephenB whom I originally thought had written that post)Hoki_{March 19, 2009
March
03
Mar
19
19
2009
08:16 AM
8
08
16
AM
PDT}

Instead of using a dictionary of English words, a program such as Zachriel could have used a dictionary of proteins found in the human body represented as character sequences. However, the fundamental flaw in the approach is the same. Supposing that it discovered a sequence that matched one of the entries in the dictionary of proteins. Then, since this is one of the "words" in the dictionary, Zachriel would preferentially preserve it. However, proteins have function in the human body in the context of cooperation with other proteins. Many of these proteins would be useless if taken in isolation, without the cooperating proteins they are interdependent with. By selecting for any word in the dictionary, Zachriel is selecting according to potential value -- value as it will be in the context of the future whole, once one has the complete language or the whole body. It is not selecting according to current utility or real advantage conferred with regard to current pressures.ericB_{March 19, 2009
March
03
Mar
19
19
2009
06:12 AM
6
06
12
AM
PDT}

Pendulum: "The benefits of selection only appear when the population is set higher. What is surprising about Weasel is not that it reaches the goal, but how fast it reaches the goal, even for very small population sizes." On the contrary, it is not surprising at all once one realizes that WEASEL selection is not based on true selection according to current function. In the second run in the original post, notice that even by the second entry in the sample we are given, the sequence "LIKE" is already being preferentially preserved, even though it is not yet functioning as a word. The WEASEL selection is cheating by taking advantage of recognizing and selecting according to the function it will eventually have. This is possible only through access to knowledge about the ultimate solution. Given that it can peek at the future utility, of course it is able to quickly converge.ericB_{March 19, 2009
March
03
Mar
19
19
2009
05:59 AM
5
05
59
AM
PDT}

hoki:
There is no real target as such (at least not one in the distant future). For natural selection there are current selective pressures.
No target = no search. Just the "necessity" of survival.Joseph_{March 19, 2009
March
03
Mar
19
19
2009
04:54 AM
4
04
54
AM
PDT}

Pendulum Actually, if Weasel were to bump up mutation rate and population size, within the performance capacity of the PC, it would outrace deterministic search by getting double, triple and so on finds in each generation. (That is because configs that would be way out in the skirts of the population would begin to show up in "random" samples, and by the distance to target metric, would consistently win and advance.) That is, Weasel is utterly unconnected to the real world, where the issue is that targeted search that rewards non-functional configs based on mere increments in proximity to a target, have nothing to do with the alleged designing capacities of chance variation plus natural selection across competing FUNCTIONAL sub populations. GEM of TKIkairosfocus_{March 19, 2009
March
03
Mar
19
19
2009
04:04 AM
4
04
04
AM
PDT}

A key aspect of Weasel, et al. is that they are population based searchers. If you dial down the population size to 1, Weasel is the same as random search. The benefits of selection only appear when the population is set higher. What is surprising about Weasel is not that it reaches the goal, but how fast it reaches the goal, even for very small population sizes.Pendulum_{March 18, 2009
March
03
Mar
18
18
2009
09:06 PM
9
09
06
PM
PDT}

Hoki:
Considering how many people fail to understand the difference iterative selection can make to randomness, it is anything but pointless.
How many more people fail to understand the difference it can make for a program to illegitimately "assist" its selections by looking at future utility? You are right to point out that actual selection works according to current pressures. Did you think that anyone doubts this? Or that anyone doubts that this type of selection is real? The problem is that programs such as Zachriel (or WEASEL) create an illusion of heightened effectiveness that does not work within the confines of true natural selection. But the fact that real natural selection helps (when there are incremental advantages to select) is not questioned.ericB_{March 18, 2009
March
03
Mar
18
18
2009
08:27 PM
8
08
27
PM
PDT}

tribune7, good point regarding Hoyle. Even though he never accepted that the intelligence was God, he and Wickramasinghe were pioneers in pointing out the absurdity and the mathematical unreasonableness of undirected evolutionary explanations for biological life. It is an inconvenient truth that inference to intelligent agency, even among the pioneers in our time, was not confined to theists.ericB_{March 18, 2009
March
03
Mar
18
18
2009
08:19 PM
8
08
19
PM
PDT}

Hoki -- Remember that an old creationist argument was that evolution was equal to a hurricane blowing though a junkyard to produce a 747. Actually, that's an atheist argument. At least Sir Fred Hoyle was an atheist at the time he made it :-)tribune7_{March 18, 2009
March
03
Mar
18
18
2009
04:58 PM
4
04
58
PM
PDT}

tribune7: The only think definitively illustrated by Dawkins’ program is its pointlessness. Remember that an old creationist argument was that evolution was equal to a hurricane blowing though a junkyard to produce a 747. Considering how many people fail to understand the difference iterative selection can make to randomness, it is anything but pointless.Hoki_{March 18, 2009
March
03
Mar
18
18
2009
04:36 PM
4
04
36
PM
PDT}

Hoki–The purpose of the program is simply to show that iterative randomness + selection can come to a target faster than randomness alone. IOW, it is sort of like someone taking the time to write a program illustrating that a man will take fewer steps walking 10 feet than he would 10 miles. The only think definitively illustrated by Dawkins' program is its pointlessness.tribune7_{March 18, 2009
March
03
Mar
18
18
2009
03:02 PM
3
03
02
PM
PDT}

Prev 1 … 7 8 9 10 11 12 Next