Dawkins’ WEASEL: Proximity Search With or Without Locking?
| March 16, 2009 | Posted by William Dembski under Darwinism, Evolution, Informatics |
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.
Patrick
Thanks.
GEM of TKI
PS: DK, your linked blogger lost me in the first sentence on his incivility, with condescending and slanderous language: if he does not know — or more likely refuses to know — that design thought is not to be equated with creationism, he has no more credibility with me. Full stop.
It is very plain from the text of BW, inclusive of descriptions as cited by Joseph and by Wiki, that Weasel circa 1986 shows ratcheting, and that the letters as published in a sample that shows 200 instances of latched letters as the o/p moves towards its target inexorably, with no reversals, in a space of 300 characters, that Patrick’s bottomline on this secondary issue is correct. On the report that Dawkins didf not EXPLICITLY latch, we have conclusded implicit latching as the best explanation.
As to the primary question, Mr Dawkins himself settled at the outset in 1986, that Weasel exhibits proximity based reward without reference to functionality. It thus cannot be the blind watchmaker at work or a reasonable illustration of it. It begs, as opposed to answers the Hoylean challenge of getting to functional, complex bio-information.
PPS: DK, As for “IDiots” as a reference to people who differ with him, sorry. Whatever possessed you to send me to a site that STARTS with that kind of language, DK?
Patrick [331], no. As Patrick May dmonstrates in the blog to which kairosfocus objects, a person using Dawkins’s text as a guide to coding would arrive at a version without. Also, as I have pointed out for some time, the examples of iterations in the text are nonrandom: they are heavily biased, being selected from among the projeny.
kairosfocus, you’re wrong: the blog post does not use the term “IDiots” (or “IDiot”), either on that page, or on his site at all. It no more uses that term that Weasel uses latching.
Non-latching implies that mutation is random in respect to fitness.
Latching implies that mutation is not random in respect to fitness.
kf and others: do you agree with those statements, yes or no.
And if not, why not?
And I’d like to encourage you to focus on just this question.
Hazel:
Since there is an implicit latching [then quasi-latching, then detuned . . . ] case, your question is mis-framed.
That is, there is no simple yes-no.
GEM of TKI
P.S.: kf, i do not see the word IDiot on the blog page that David linked to. I see IDCists: perhaps you misread.
In the non-latching case, the mutation function can mutate both incorrect and correct letters. That is, the mutation function has no knowledge of the fitness function nor is it influenced by the fitness function as it chooses which letters to mutate.
It is in this sense that we say that mutation is random with respect to fitness.
Do you agree with this sentence. If not, why not.
Hazel:
In the implicit latch and/or quasi-latch cases, the trick is that the multi-mutation skirts of the population are sufficiently excluded that the proximity filter locks out regressions, as 0-mutations and one or so good mutations will be caught with practical certainty.
“Don’t look behind the curtain” does not work here. The concern is the program as a whole.
On the EXPLICIT latch case, the point can be argued that “all” that is being done is letting what reached to the correct value by random changes plus filtering stay there as we “know” optimal mutations are preserved by the envt.
But the decisive objection is that the proximity reward on the non-functional invalidates the whole exercise from the start. Latching, etc are very much secondary issues, as I have pointed out from December on. Such, do however point to the key problem: targetted foresighted, non functionality constrained search.
Weasel is misleading, at its root; begging the very Hoylean question of origination of com,plex, functional bio-information it was supposedly set up to answer.
GEM of TKI
PS: I will not go back to that page to see if these old dyslexic eyes misread or not (which is possible . . . ); but the difference in context between the two is a distinction without a difference. NEITHER is materially better; BOTH are equally slanderously false, uncivil and condescending.
“I see that a further discussion launched after this point, but this seems to be a silly argument to make.”
The whole discussion is silly. “To latch or not to latch, that is the folly.”
“The most exquisite folly is made of wisdom too fine spun” Ben Franklin
kf: I notice that you didn’t answer the question, but rather repeated points you have made about some other issues.
Is there some difficulty in just looking at the relationship between the mutation function and fitness function and assessing whether the mutation is random in respect to fitness?
Let me repeat, more explicitly: the overall net effect of the process each generation or over generations is NOT the issue I am asking you about, and even more so the overall relevance of this exercise to the real world is not what I am asking about.
I am asking about a much simpler issue: when a phrase mutates in the non-latching situation, is it true or not that the mutation routine does not reference the fitness routine, and is thus random in respect to fitness.
Can you focus on, and answer, just this one question?
Hazel:
Pardon; you can properly ask me a question, but you cannot justifiably demand that I answer to fit a flawed, probably rhetorically loaded framing.
Recall, onlookers: all of this back-forth over latching etc is in light of a rhetorically very loaded context, GLF’s attempt to discredit the undersigned by raising a distractive point on a thread on the self-referentially absurd implications of selective hyperskepticism. GLF’s foolish wager boast gambit still stands exposed for the hollow rhetorical tactic it was.
And, onlookers, note how there has been a silence on the point that there is such a thing as the law of large numbers, which makes the o/p’s of 1986 credibly good sampling data on the latching of the 1986 o/ps as published. You will search above in vain for a retraction on the former loudly announced claims that the published samples in question were too small to be likely to be truly representative of the o/p. Likewise, you will search in vain for the apology or correction of violation of my privacy. Not to mention, for the explanation of why a denunciation and rhetorical dismissal of me by name by someone indulging in blood slander against Christians and associated enabling behaviour for lewdness in public in the Jamaican media [and for which the newspaper in question was forced to publish a corrective; though they damage controlled even that . . . ], was latched on to and trumpeted, here and at Anti Evo.
Hazel, do you see why I HAVE to treat this as a rhetorically loaded context, not a quiet collegial afternoon exchange over sipped cups of tea in a Departmental Seminar Room?
Now, back on your point: I pointed out already, that the root issue on the secondary question of latching is not just module action but module interaction in light of parameter settings etc. Hence, IMPLICIT latching.
Emergence of — sometimes unexpected — behaviours through such interactions is a well-known characteristic of systems.
(Indeed, this is one reason why the commonly met with co-optation argument against irreducible complexity is so simplistic — it grossly underestimates the challenges of coupling, interfacing and interaction; as any experienced system designer and developer can tell you. Join that to the key-lock fitting, wholistic behaviour of proteins on folding and agglomerating based on their AA sequences, and you see that the co-optation story is a little less than realistically credible.)
So, you can allow at-random mutation among members of the population of a generation all you want. But, once pop size, mutation rate and — most importantly — filtering based on mere proximity without a realistic functionality criterion are imposed, under relevant co-tuned conditions [probably accessible through trial runs], the program will credibly lock or quasi-lock letters as they reach their individual targets.
Next, you can try to make a fine distinction between at random mutations that affect the string as a whole all the way through to the final version — and, how does the pgm know to stop mutating at that point, is that “random” or “blind”? — and the possibility of explicitly latching correct letters by partitioning the search so that once a letter reaches its functionality target that is recognised by locking. Indeed, one can always search for lawyering emanations of penumbras of hints in the text of TBW, circa 1986, to one’s heart’s content.
But all of that is immaterial.
For, on a plain, direct and reasonable, objective plain meaning reading of the text [as Joseph has underscored in recent days], the explicitly latched version of the program is directly and strongly supported by the statements and o/p excerpts published in 1986. The example of Monash university over in Australia — which supports “your: side of this exchange — shows that beyond reasonable doubt. They had to be “corrected” by Mr Elsberry to see that hey had “missed” the “correct” evolutionary materialist interpretation of Weasel circa 1986. 9And, yes I am using “scare quotes” to highlight the point.)
This is therefore what is relevant: is there evidence that explicit latching is a reasonable interpretation of Mr Dawkins’ description and published o/p at that time?
ANS: Plainly, yes. It is on further reports that Mr Dawkins states that he did not use explicit latching, that we have explored how else he could have programmed Weasel AT THAT TIME. And, on the balance of the evidence, the best explanation is IMPLICIT LATCHING. (In that context, the 1987 o/p is most likely a de-tuned for video impact version.)
Back to the original frame: The latest rabbit trail over which is more or less random is a third level distraction from the core issue with Weasel.
For, once mere proximity to target is rewarded without reference to realistic function, then Weasel begs rather than answers the Hoylean question of origin of complex bio-functional information.
For, it is not a BLIND watchmaker at work.
Weasel is not a proper illustration of claimed evolutionary processes, and begs the question of getting TO biofunction. It should therefore be withdrawn as a yet another flawed icon of modern evolutionary thought.
And, on the question of latching, which was raised as a distractor in a previous thread, in an attempt to discredit the undersigned in the first place [and by extension once his name came up Mr Dembski], once we see that the text supports this as a reasonable interpretation, and that here are two credible mechanisms to achieve such, which are both consistent with the published results circa 1986, the exchange should have ended at that point.
That it has not, simply reveals to the astute observer, that the evident latching of the o/p in Weasel circa 1986 — cf the original post in this thread — is a strong hint on what was fundamentally wrong with this icon of evolution.
So, clouds of rhetorical squid ink notwithstanding, we know that there is a squid trying desperately to get away behind all the distractive clouds of contention and debate points.
Hazel, you have raised some significant points that have contributed to a clear enough conclusion. For that, I give you thanks.
GEM of TKI
kairosfocus, did you ever define “implicit latching”? What would it mean for a letter to latch “implicitly”? I can’t find such a definition.
As for the rest, your paragraph beginning “And, onlookers, note how there has been a silence” is full of what I view as errors and misrepresentations. I have to do an errand now, but it’s likely I’ll still be in the moderation pile when this gets posted. Such delays have made it possible for you to ignore (or perhaps be unaware of) various points that I have made in this thread. In any event, you have not responded to them.
“Implicit latching” sounds to me like being “a bit pregnant”. Either a correct letter locks in place until the whole target phrase is reached or it does not. If correct letters can be deselected for whatever reason then there is no latching.
I take it you aren’t going to answer my question. I realize that of course you don’t have to, but since you once again just spent 1000 words repeating things you have said many times before, and have made it clear that you’re not interested in a conversation on specifics, I gather there is no sense in my repeating myself and asking you again, so I won’t.
The above was to kairosfocus: probably obvious, but I should have been clearer.
Kariosfocus,
At this URL
http://www.uncommondescent.com.....ent-300338
You said
Concentrate on the first sentence. I include the second as your usual response here is to accuse me of quotemining. The second sentence is irrelevent at the moment.
Now you say:
No, it is you who have attempted to distract with your “hoylian challenge, islands of functionality etc” strawmen.
The issue is quite clear.
I offered $100,000 if you could provide a quote from Richard Dawkins that backed up your original position.
You have not been able to do so.
You did not response to my question at the time. You claim that 200 out of 300 sample points that show no reverting proves your point.
I asked you at the time – what was the total population of letters? It was not 300 was it?
You ignore relevant questions and then claim victory? You ignore the fact that people are talking about this exact issue in mathmatical terms and yet proclaim you are right, and it’s been proven?
Here
http://tinyurl.com/cytao2
The probablity of candidate changing a parent’s correct base to an incorrect base is discussed.
If you are so sure you are right, why not prove it with maths? Who could argue with that?
This is simply untrue, as has been documented in detail. Simply ignoring when people point it out does not advance your case.
Why have you ignored so many questions addressed to you?
Quite right. And Richard Dawkins has said that he did not use latching, that latching would have been against the point he was trying to make and he did not even consider using latching.
Therefore your interpretation is incorrect.
Accept it. And then tell me how you intend to pay me my $100,000 for winning our bet.
Hazel:
You seem a little slow, Hazel. Over how many weeks, across multiple threads, has KF circumlocuted questions regarding latching? Sure, he inserts a few “quasis” and “implicits”, but at this point it is safe to assume he isn’t going to be painted into a corner over the issue of Weasel’s latching. When it become obvious even to a relatively uneducated onlooker like myself, I think the point has been made. Persisting further is only gilding the lily. Or perhaps gilding the shadow of the lily on the wall of Plato’s Cave.
kairosfocus @343
KF, I used to think you were one of the most decent, gentlemanly participants here, but your behavior has gone beyond the ridiculous on this topic.
The only reason you won’t answer Hazel’s simple, non-loaded question is because you are constitutionally incapable of admitting even the slightest error. It is painfully clear that your claims of explicit latching are unfounded. Rather than admit this rather minor mistake, you spew copious amounts of unrelated verbiage in an attempt to distract from that core, essential, point and treat polite correspondents like Hazel rudely.
I expected better of you.
JJ
kairofocus [343], hello again. I want to point out that the claims you have made are subject to alternate reading. I don’t expect you’ll agree on any of this, and I don’t claim that my views are uncontestable. Moreover, I put this in a separate comment in the hope that you can deal with the latching issue in its own post without digression or accusation.
Wesley Elsberry has made argued that the law of large numbers works against you. You have not responded to mathematical argument on its merits.
I have pointed out that the data are highly non-representative because they are the products of a large selection bias (the best sampple from each generation provided). This was obvious from the text of TBW.
The only person who keeps talking about your name at UD is you. I have mentioned that spam does not go to a proper name but to an email, so your claim of more spam from your proper name being put on another board is spurious. As far as AtBC, that’s not my board. I’m not going to ask them to change a policy that has nothing to do with me.
The person in question was responding to something that you signed and published. There was no “blood slander.” There was a claim about the evangelical community that is subject to dispute.
That’s a non-issue for me.
David
It isn’t 1986 any more. Many independently written and tested implementations of Dawkins’ simple “Weasel program” teaching aid are now publicly available, It’s a simple program and a working script can be written by any reasonably bright teenager in a few minutes.
The program gives a simple demonstration of the process which Darwin termed “Descent with modification” and with which all animal breeders are familiar.
It’s a simple filter with two components. Firstly variation is generated: Dawkins chooses a copying algorithm that randomly introduces “mutations”–imperfections–into the copy. Secondly a selection is made: Dawkins uses a predefined target and a simple distance metric suited to the search domain to select the “most fit” copy from which to make the next generation. Anyone who has bred pigeons, dogs or other animals for show where published breed standards apply will recognise the technique.
Not surprisingly, the simple filter converges quickly on the target, just as competitive breeding of animals produce examples of an animal that match the parameters of the breed remarkably faithfully.
Notice that I haven’t referred to natural selection anywhere here. The point is that variation exists in nature and can be selected. This was already well established–had been known by breeders for centuries–before Darwin was born.
The example doesn’t demonstrate natural selection because it isn’t intended to do so. It is as applicable to demonstrating the mechanism by which domesticated animals are bred as to anything else.
I don’t know why intelligent design advocates would have any problem with this. The Weasel Program demonstrates nothing remotely controversial and does so in a simple and straightforward way that is difficult to misunderstand.
Onlookers:
Re 344 on:
There is a successor thread, which answers to new objections.
In particular the idea that the observed latching on the o/p of Weasel circa 1986 is not real, has been long since addressed in terms not only of the fact that we have a more than adequate sample and direct statements by Mr Dawkins on the matter, bit also that we provided two possible mechanisms, with sufficiently detailed explanations. Onlookers, simply scroll up to the original post above and count: of 300+ places where letters could change, 200+ — well beyond where the law of large numbers weighs in on the likely representativeness of a sample — show letters that once they go correct remain so. A dominant feature of the Weasel o/p circa 1986, and which Mr Dawkins made clear reference to, lwayering over emanations of penumbras of the text notwithstanding.
With zero exceptions.
On preponderance of evidence, Weasel circa 1986 latches implicitly off per letter mutation rates interacting with per generation population size and most importantly a selection filter that rewards mere proximity in the teeth of non-functionality. Circa 1987, Weasel was evidently detuned so that we see a different pattern of behaviour: multiple reversions that occur fairly frequently.
This and other successive variations and versions of Weasel and neo-Weasels should not be confused with the issue that was raised in an agenda-serving, originally threadjacking attempt to discredit those who pointed out the obvious fact of o/p latching of Weasel circa 1986. (That was brought up in a thread that was discussing the pervasive problem of Cliffordian evidentialist form selective hyperskepticism, which continues to be a central intellectual challenge that the evolutionary materialists seem to have. Complete with shoals of red herrings led out to ad hominem- soaked strawmen, ignited to cloud and poison the atmosphere for discussion that might lead to inconvenient truth.)
Going all the way back to December last, it is that proximity – without- functionality filter that has been highlighted as utterly gutting Weasel of any proper didactic or illustrative credibility. Weasel always has been, a rhetorical exercise in question-begging and misdirection. That is why it has always been controversial, bland declarations to the contrary notwithstanding.
This can be seen both from what GLF cited adn twisted yet again from 111 in the December thread, and from what he did not cite from 107, e.g.:
As to GLF’s infamous $100k offer, it was plainly never a serious offer, and was clearly only intended to serve as a gambit on which he hoped to boast that no-one could take him up. In that, he has been sadly exposed for the hollowness of the rhetoric involved. As can be abundantly seen here and in the previous thread.
As to the recycling of long since adequately answered objections [apparently an evolutionary materialist advocacy strategy on the ID movement; probably on the idea that an often repeated claim, even if unwarranted, may often be perceived as true], you can simply scroll up and see how the cycle of selective hyperskepticism has played out over and over again.
In short, more than enough has been said, many times over, for any reasonable person to see that Weasel is not legitimate, and has never been, regardless of the issue of “latching.” On this secondary issue, it is plain that Weasel 1986 latches and that we have a reasonable mechanism — or two — to account for that.
On either explanation [explicit or implicit], the latching points straight back to the core problem: Weasel is targetted, foresighted, designed, active information based search, not a reasonable analogue of any BLIND watchmaker, especially the much vaunted natural selection.
GEM of TKI
PS: Onlookers will also observe that to date, there has been no responsible accountability over violation of my privacy, or over gleefully citing an abusive dismissal of me in the Jamaican media that the newspaper in question had to publish a corrective over. Note as well, that dismissal was in service tot he blood slander of plastering Evangelical Christians with the terrorism of IslamIST radicals, and to thereby enable public lewdness for profit (with amateur night also horrendously in play, including corruption of minors).
KF, you are incorrigibly long-winded, verbose, convoluted, and unable to stay on any one topic.
Kairos, be a gem and focus.
David #334
Did I not say just that?
“Or if the program is 100% recreated and then analyzed with various runtime parameters.”
Learn to read carefully before you respond.
That is the source of the problem. Dawkin’s unwitting mistake has led others into error. An error that has been noticed and dealt with, but people like you seem to have trouble realizing that.
In any case, this thread is degenerating into a mud-slinging contest so I’m ending it.