Uncommon Descent Contest Question 10: Provide the Code for Dawkins’ WEASEL Program

_{Denyse O'Leary

August 26, 2009

Darwinism}

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

Special invitation for Richard Dawkins – but any civil person is entitled to enter.

There’s been some discussion here and elsewhere whether the the recent IEEE article by Dembski and Marks correctly characterizes Richard Dawkins’ famous METHINKS IT IS LIKE A WEASEL program.

Does the program ratchet correct letters or does it let them vary? One is a partitioned or stair-step search, the other a more realistic evolutionary search. From The Blind Watchmaker, where Dawkins describes the program, its performance suggests that it could be either of these options (though he doesn’t say).

On the other hand, from a (video-run of the program , go to 6:15), it seems to be the latter.

It’s easy enough to settle this question: Make the code for the program public. Perhaps Richard Dawkins himself or his friends at RichardDawkins.net can finally provide this code (apparently a program written in BASIC).

The prize is a copy of either Stephen Meyer’s new Signature in the Cell or Richard Dawkins’ soon-to-be-out The Greatest Show on Earth.

Should the winner choose the latter, I will ask Dawkins’s publicist to mail the copy. Given that at his site, he calls himself “the most formidable intellect in public discourse,” I would assume that if he signs the copy, it will be worth millions.

But wait. Let’s see that code first.

Comments

Joseph, A partioned search is [insert] one type of [/insert] a latching/ ratcheting process. Fixed that for you. Perhaps you meant to say "A latching process is (necessarily) a partitioned search (as modeled in eqn22)" But that would be wrong. Or you meant to say "Weasel is a partitioned search (as modeled in eqn22)" But that would be obviously wrong.DNA_Jock_{August 31, 2009
August
08
Aug
31
31
2009
08:00 AM
8
08
00
AM
PDT}

Joseph, I am in the process of explaining why you are wrong. So, can you give me a straight answer to my question: In the TBW Dawkins says that the strings are copied with random errors/mutations. Since he does not limit these mutations in any way to incorrect letters this means correct letters also mutate -> No ratcheting or latching at the mutation level. Agreed?Indium_{August 31, 2009
August
08
Aug
31
31
2009
07:42 AM
7
07
42
AM
PDT}

DNA Jock, A partioned search is a latching/ ratcheting process.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
07:28 AM
7
07
28
AM
PDT}

Indium, It is very telling that you cannot meet my challenge. It is also very telling tat all you have are desperate semantics. Again "cumulative" and reversals only make sense in an agency involved scenario. Non-telic processes would not see reversals as an improvement over the parent.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
07:22 AM
7
07
22
AM
PDT}

Joseph, it is not our fault that you seem to think that a cumulative process can´t also accumulate negative values. Weasel is a cumulative because the results from a search are the basis for the next search step. I have never seen such a desperate semantic argument before. So, maybe you can give me a straight answer to the following question, we can then proceed from there: In the TBW Dawkins says that the strings are copied with random errors/mutations. Since he does not limit these mutations in any way to incorrect letters this means correct letters also mutate -> No ratcheting or latching at the mutation level. Agreed?Indium_{August 31, 2009
August
08
Aug
31
31
2009
07:17 AM
7
07
17
AM
PDT}

No, Joseph, you are wrong. Dawkins' narrative is clear to 99% of the people who have read it -- "Random mutation" means random mutation as it is observed to occur in nature. In particular Dawkins refers to selecting the best "phrase" which makes little sense for a latching mechanism, and none for a partitioned search. "Cumulative" is used in contrast to the alternative "single-step", or perhaps "instantaneous". It is as telic as "In April, water accumulated in my wheel-barrow" which does not (to me anyway) deny the existence of evaporation. You are asking (I hope) a rhetorical question - "Where in TBW does it say There is no latching ?". I might as well ask "Where is the KJV Bible does it say 'Adam did not have three arms' ?". But, and this is the really important point, even if there were an explicit latching mechanism in Weasel, that would not make it a partitioned search as described by D&M. The first reported Weasel run in TBW CANNOT be the result of a partitioned search. Only two letters change in generation 2. Latching is irrelevant to the accuracy of D&M's citation. I am accusing D&M of mis-citation when they claim that eqn22 describes Weasel. I encourage everyone to go to EIL and play with Atom's adjustable Weasel, specifically the Proximity Reward Search, and contrast its behavior with that of the Partitioned Search.DNA_Jock_{August 31, 2009
August
08
Aug
31
31
2009
07:16 AM
7
07
16
AM
PDT}

It looks like my challenge cannot be met. No surprise there. I challenge anyone to find a passage or passages in TBW that would show that cumulative selection is not a ratcheting process. That reversals can happen and be selected. Without that all you anti-IDists are blowing smoke. Not only that you are accusing D/M of something that they didn't do.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
06:59 AM
6
06
59
AM
PDT}

my two cents: Cumulative selection -- increasing by successive additions -- is indeed a ratcheting process so long as the string, chosen from the present generation, most closely representing the target phrase is closer to the target phrase than the parent string. The chances of reversal would depend on the size of the population and the mutation rates. Thus, within specific parameters, Dawkins algorithm will provide a ratcheting process ("implicit" latching since it is based on probability of reversion to a string further from the target) and within other parameters it will not provide a ratcheting process. So long as we see a steady closing in on the target, we are viewing a ratcheting process (whether it happens as a result of probability or explicit programming to achieve ratcheting behavior).CJYman_{August 31, 2009
August
08
Aug
31
31
2009
06:58 AM
6
06
58
AM
PDT}

BillB:
I challenge anyone to find a passage or passages in TBW that would show that cumulative selection is a ratcheting process.
That part has been presented already. It has to do with Dawkins talking about "slight improvements". As soon as I get the book back I will provide the quotes.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
06:57 AM
6
06
57
AM
PDT}

Indium, The way Dawkins describes cumulative selection as "slight improvements" can only mean it goes in one direcetion- towards the goal. Reversals as an improvement only work with agency involvement because agencies can plan ahead.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
06:56 AM
6
06
56
AM
PDT}

Joseph, ratcheting, latching, implicit, explicit, so many arguments are just based on semantics. What do you mean by ratcheting? A ratchet can only go in one direction, the algorithm Dawkins describes can go in both (just try it yourself with Atoms suite). In the TBW Dawkins says that the strings are copied with random errors/mutations. Since he does not limit this mutations to incorrect letters correct letters also mutate -> No ratcheting or latching at the mutation level. Agreed? This is of course in line with what Dawkins says, what is demonstrated in a video and with biological reality. And even with kairosfocus allways linked appendix 7. And with the EIL software.Indium_{August 31, 2009
August
08
Aug
31
31
2009
06:48 AM
6
06
48
AM
PDT}

I challenge anyone to find a passage or passages in TBW that would show that cumulative selection is a ratcheting process. That reversals can not happen and will not be selected.BillB_{August 31, 2009
August
08
Aug
31
31
2009
06:22 AM
6
06
22
AM
PDT}

KF, Please could you at least try and understand what the debate is about. When Dawkins describes a mechanism that does not lock letters, and Dembski and Marks describe one that does, and both produce different behaviour, and Dawkins published behaviour is consistent with the algorithm he describes ... and Dembskis is inconsistent with Dawkins output ... and Dawkins specifies a population ... and so on and so on ... Why would any sane person still try and maintain that these are the same algorithm and that the citation is appropriate? Playing semantic games with words like latching doesn't help your case at all, it just smacks of arrogant desperation and a total inability to acknowledge your own mistakes.BillB_{August 31, 2009
August
08
Aug
31
31
2009
06:19 AM
6
06
19
AM
PDT}

DNA Jock, If we go by what Dawkins wrote in TBW then "weasel" is a partitioned search- ie ratcheting occurs. And we know ratchets can move in both directions but force is only applied in one.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
06:14 AM
6
06
14
AM
PDT}

I challenge anyone to find a passage or passages in TBW that would show that cumulative selection is not a ratcheting process. That reversals can happen and be selected. From the book only.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
06:12 AM
6
06
12
AM
PDT}

BillB, There isn't anything in your quote that satisfies my challenge. No mention that letters, once matched, can be changed. That is what I am looking for and I made that clear. Also I have said that intellignet agencies can use "cumulative" in a different way. So if you are saying that evolution is guided by an intelligent agency then you would have a point. Otherwise "cumulative" along with the description of "slight improvements" only means ratcheting is taking place.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
06:10 AM
6
06
10
AM
PDT}

BillB - It has been pointed out to kf that latching is irrelevant to the original question, which is "Are D&M accurate when they call Weasel a partitioned search, and use eqn 22 to describe it?" We have been explaining this to kf in simple, plain english since post 34, where I said
Forget about latching, you cannot have a partitioned search that changes only one element in its first iteration, and reaches a solution within 43…
kf, OTOH, has resorted to re-defining the word "mechanism" (@232) in his increasingly desperate attempts to avoid admitting any mistake. Humpty Dumpty would be so proud.DNA_Jock_{August 31, 2009
August
08
Aug
31
31
2009
06:07 AM
6
06
07
AM
PDT}

Joseph:
[Weasel] . . . begins by choosing a random sequence of 28 letters … it duplicates it repeatedly, but with a certain chance of random error – ‘mutation’ – in the copying. The computer examines the mutant nonsense phrases, the ‘progeny’ of the original phrase, and chooses the one which, however slightly, most resembles the target phrase, METHINKS IT IS LIKE A WEASEL
Challenge met - no mention of a rule that prevents letters reverting. When you implement this you get a mechanism that works towards the target, revisions occasionally occur, as KF has demonstrated, and all published data is consistent with this. This is also consistent with the use of cumulative as paying attention to movement in one direction and not in the other, for example cumulative height or accumulated wealth.BillB_{August 31, 2009
August
08
Aug
31
31
2009
06:06 AM
6
06
06
AM
PDT}

Onlookers: Re BillB: This concession over implicit latching is a fantasy of yours, in Dawkins algorithm the fittest member appears to latch some of the time without ever needing a latching MECHANISM . . . Sadly, this reveals that BillB is till well out beyond the pale of civil discourse: 1 --> In the App 7, the always linked [dating to about April] we may see from point 12 on, after a presentation on what is to be explained, Weasel 1986:
12 --> To explain the latching more realistically, we may have an explicit latching algorithm based on letterwise search . . . . 13 --> Letterwise partitioned search is also a very natural way to understand the Weasel o/p in light of Mr Dawkins' cited remarks about cumulative selection and rewarding the slightest increment to target of mutant nonsense phrases. As such, it has long been and remains a legitimate interpretation of Weasel. However, on recently and indirectly received reports from Mr Dawkins, we are led to understand that he did not in fact explicitly latch the o/p of Weasel, but used a phrasewise search. 14 --> Q: Can that be consistent with an evidently latched o/p? ANS: yes, for IMPLICIT latching is possible as well. 15 --> Namely, (i) the mutation rate per letter acts with (ii) the size of population per generation and (iii) the proximity to target filter to (iv) strongly select for champions that will preserve currently correct letters and/or add new ones, with sufficient probability that we will see a latched o/p. (This effect has in fact been demonstrated through runs of the EIL's recreation of Weasel.) 16 --> in a slightly weaker manifestation, the implicit mechanism will have more or less infrequent cases of letters that will revert to incorrect status; which has been termed implicit quasi-latching. This too has been demonstrated, and it occurs because an implicit latching mechanism is a probabilistic barrier not an absolute one. So, as the parameters are sufficiently detuned to make reversions occur, we will see quasi-latched cases. Sometimes, under the same set of parameters, we will see some runs that latch and some that quasi-latch. 17 --> In a litle more detail, we will see reversions in a case where the odds of mutation per letter are sufficiently low that in a reasonable generation of size N, there will be a significant fraction of no-change members, and so the proximity filter will select no-change to be next champion, or a single step increment in proximity; or, in some cases a substitution where one correct letter reverts and one incorrect letter advances . . . . 19 --> In short, there is almost no chance that such a mutant generation population will have no unchanged members. In that context, with the proximity filter at work, no-change cases or substitutions or single step advances will dominate the run of champions. Indeed, the Weasel 1986 o/p samples show that runs to target took 40+ and 60+ generations, i.e. no-change won about half the time and single step changes dominated the rest. No substitutions were observed in the samples, suggesting strongly that there were none in the showcased runs. (Double step advances etc or substitutions plus advances will be much less probable. But in principle, per sheer "luck," we could see the very first random variant going right to the target. Just, the odds are astronomically against it. As, probabilistic barriers may be stiff, but are not absolute roadblocks.) . . . . 24 --> In turn that means that a further key issue on champion selection per generation is the specific action of the proximity filter, in a context where -- continuing the concrete example -- the expected number of zero change cases in a generation is about 17, that for single step advances is about 1, and that for substitutions ranges from 1/2 down to 5/1000, depending on degree of proximity already achieved. That turns out to be a fairly complex issue:
2 --> Given subsequent debates, I would adjust phrasing for clarity, but the point is plain: explicit mechanisms latch letters explicitly, and implicit ones do so out of the interaction of population size, mutation rate and filter characteristics. remember o/p is what we directly observe, and mechanism is based on the i/p's and processing that give rise to it. 3 --> That is, the issue of observed behaviour and its explanation by alternative mechanisms was long since addressed. It is not a latterday "concession," but an alternative EXPLANATORY MECHANISM, one that -- per inference to best, empirically anchored explanation -- is in fact preferred on the balance of the evidence; ever since some time in March or so, when the information came to our attention at UD. (Remember, it is a DEMONSTRATED mechanism, as is for instance reproduced and/or linked above.) 4 --> Therefore, it seems that BillB has regrettably failed to do his homework before criticising, and has indulged in pejorative mischaracterisation and misrepresentation. 5 --> This pattern of uncivil rhetorical conduct is to be corrected on his part. GEM of TKIkairosfocus_{August 31, 2009
August
08
Aug
31
31
2009
06:05 AM
6
06
05
AM
PDT}

Joseph, I think you are confused. Dembski and Marks are using a different thing to explain a different thing but they cite Dawkins algorithm and make out it is the same, when as we have seen, it is actually different. They should have referenced a similar thing when explaining their thing rather than referencing a different thing than the one they explain. It is all very simple really!BillB_{August 31, 2009
August
08
Aug
31
31
2009
06:01 AM
6
06
01
AM
PDT}

I challenge anyone to find a passage or passages in TBW that would show that cumulative selection is not a ratcheting process. That reversals can happen and be selected. From the book only. Good luck...Joseph_{August 31, 2009
August
08
Aug
31
31
2009
05:47 AM
5
05
47
AM
PDT}

The facts are that Dawkins used cumulative selection as a ratcheting process. He descirbes it as "slight improvements". The way cumulative selection is described and illustrated in TBW it is a ratcheting process. And the only way around that is to contort what Dawkins actually wrote.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
05:45 AM
5
05
45
AM
PDT}

Interesting, onlookers, how a concession that implicit latching is real can be made to look like a claim of victory for the Darwinists!
KF, the onlookers aren't stupid, they can see that this whole "lets define latching to mean whatever we need it to mean in order to claim victory" is your own tactic, one you keep employing again and again. The onlookers who have been following this long enough will remember how you invented the ideas of implicit/quasi/semi-latching in order to get around the fact that al the published output of WEASEL can be explained without it requiring an EXPLICITLY DEFINED LATCHING MECHANISM - something Dembski and marks explicitly define for their algorithm but which Dawkins doesn't. This concession over implicit latching is a fantasy of yours, in Dawkins algorithm the fittest member appears to latch some of the time without ever needing a latching MECHANISM. We all knew this from the start and pointed it out to you, you then invented the term "implicit latching" to describe this behaviour, and claimed that you were right about there being a latching mechanism all along. You are making a fool out of yourself, and of this website.BillB_{August 31, 2009
August
08
Aug
31
31
2009
05:36 AM
5
05
36
AM
PDT}

Indium, D/M are explaining something that Dawkins' wasn't. So yes two different sets of people can use one thing to explain two different things. The confusion comes from yet another set of people who can't understand that. And here you are...Joseph_{August 31, 2009
August
08
Aug
31
31
2009
05:19 AM
5
05
19
AM
PDT}

PPS: I need correct Indium and DJ, that: 1 --> The observed Weasel 1986 behaviour [o/p] requires a mechanism [i/p & processing] -- i.e. the old I--> P --> O elements of a program -- that explains its dynamics; of which explicit and implicit mechanisms are the two best candidates. [Is this a case where StephenB's "causeless events" problem for Darwinists is surfacing? I firmly believe -- on good grounds -- that unless necessary causal factors are present an event CANNOT happen, and unless sufficient ones are present it WILL not happen. So if it happens, we can identify causal factors and how they work -- i.e mechanisms. And given the issue of synergy -- effects due to interaction -- mechanisms do not have to be explicitly built and labelled as such. I do not believe in magical poofery!] 2 --> In terms of Atom's adjustable weasel, explicit latching corresponds to "partitioned search" and implicit latching will show up with cases under his "proximity reward search." 3 --> The decision that implicit mechanisms best explain the observed showcased runs of champions c 1986 is based in large part on Mr Dawkins' subsequent testimony; understood as denying that Weasel explicitly latched. 4 --> Once ratcheting behaviour exists, the eqn 22 from p. 1055 of the IEEE paper, suitably understood, applies. For instance -- and here I took a correction on the interpretation of Q -- with generational clustering, queries [= mutant pop members] come in generational bunches of size S: Q = G*S. And the latching-ratcheting o/p will apply to the run of generational champions; an observable behaviour. 5 --> So, for instance with pop size = 50, queries to date will go 50, 100, 150 etc. corresponding to 1, 2, 3 etc times 50. 6 --> Interesting, onlookers, how a concession that implicit latching is real can be made to look like a claim of victory for the Darwinists! (Hint: rhetoric is the art of persuasion, and too often works by making he weaker case seem stronger than it is. So, let us attend to the merits and facts, and follow back to the points where it was being ever so stoutly insisted that here was no evidence of latching in the Weasel o/p c 1986, or that latching only meant explicit latching, or that implicit latching did not happen or that implicit latching is a triviality etc etc.)kairosfocus_{August 31, 2009
August
08
Aug
31
31
2009
05:14 AM
5
05
14
AM
PDT}

Joseph,
It appears that they (D/M) are combining the outputs to get one.
If it means what I think it means (combine the correct letters of several parallel partitioned searches to get to the next phrase) then this is wrong. Just look at Eq22 again. Or at the text, which explains it will enough. And even IF it would be the way D+M were constructing the next parent string it would still be very different from Weasel. I thought you understood the difference earlier in this thread but it seems I was mistaken. How can one use the same algorithm differently anyway without using in fact a different algorithm? You could use different parameters (population sizes or mutation rates) for Weasel but you have no such choices for the partitioned search!Indium_{August 31, 2009
August
08
Aug
31
31
2009
05:03 AM
5
05
03
AM
PDT}

I would say they are using the same algorithm just using it differently. It appears that they (D/M) are combining the outputs to get one. Ya see the intention is different- as I said above. Dawkins was trying to show how the ratcheting properties of cumulative selection will find a target much sooner than a random step-wise search. D/M were trying to show how that type of search does its job.Joseph_{August 31, 2009
August
08
Aug
31
31
2009
04:47 AM
4
04
47
AM
PDT}

KF Yet again you need to be reminded of the basis for this debate, and O'Leary's question. The issue is whether Dembski and Marks describe the same algorithm as Dawkins in the section of their paper where they reference The Blind Watchmaker. It is a question of good scholarship, of correcting your errors when they are pointed out. The issue of WEASEL's biological plausibility, of whether it explains how life began, is totally irrelevant to this issue of scholarship - WEASEL could be a total failure at whatever Dawkins was trying to achieve, but it is still wrong to misrepresent it in a peer reviewed paper. Please can you try and stay on topic. If you want to discuss the interesting and worthwhile question of how WEASEL applies to biology, or the technicalities of Dembski and Marks paper then I'm sure Clive would oblige with a separate thread. The question we are trying to get a clear answer on is purely about why Dembski and Marks cite the WEASEL algorithm in The Blind Watchmaker, but then describe a substantially different algorithm. You are jumping backwards through oily hoops to try and make out that the differences between these two clearly different things don't amount to anything. It is both tragic, and very entertaining. Moving on... When you said this:
this implies that when metric falls to distance to target = 0, there is a latching action imposed.
You appeared to be trying to claim halting as a latching mechanism relevant to the debate on latching mechanisms, I'm glad that this was just poor communication on your part, but why mention halting with reference to latching at all? - it seems irrelevant to the debate but on closer inspection it highlights the differences between WEASEL and a partitioned search. Without an explicit stop condition for the software Dembski and Marks algorithm will stop searching when it hits the target and loop forever doing nothing. WEASEL will continue to produce generations, each with mutated members and, depending on mutation rates and population sizes, the winning phrase will continue to show occasional reversions even after the target is found. So when you say this:
Weasel type algors will lock up as they hit home.
You are in error. I also agree with nephmon about your comment at 222 - what was the point, you are just demonstrating what we already know - that a latching mechanism, an explicit piece of code to lock correct letters out of the search, is not required. You seem to have a blind spot about this issue and the way it impacts the two algorithms, I'll try and explain (again): Dembski and Marks describe a series of searches for individual letters - each letter is ranomised and checked against a target letter, if it matched then a halting condition is achieved. Proximity to target with repect to the whole phrase is determined by the number of letters being searched for - when there are zero the target is found. Dawkins describes a process where the number of correct letters are summed to produce a score, a proximity, for the phrase as a whole, not for individual letters. The search is NOT partitioned up into individual letter searches, each with its own halting condition. The search mechanism does NOT halt for an individual letter when it matches, it halts when the sum of matching letters is equal to the number of letters in the phrase. These are important differences between the two, even if they sometimes produce similar behaviour. Glossing over this by trying to pretend that it is irrelevant to the content of the paper misses the point. It trivialises the whole subject of search algorithms. These are two different processes at work, it is inappropriate to describe them as being the same.BillB_{August 31, 2009
August
08
Aug
31
31
2009
04:40 AM
4
04
40
AM
PDT}

NM: Plainly, you represent the "it doesn't matter . . . " faction! Again: the primary issue is that Weasel's superiority over "lucky noise" comes from its active information, manifested in targetted, proximity rewarding search that picks and promotes non-functioning "nonsense phrases." (This was actually inadvertently acknowledged and/or implied by CRD in BW.) The secondary debates over Weasel and explicit or implicit latching have to do with [a] how the results showcased in 1986 were best explained, [b] responses to attempts to dismiss that. Latterly [c] to the claim that here is a radical divergence between what Weasel 1986 did and the description ion Marks and Dembski's IEEE paper that the showcased results reflect partitioning due to ratcheting action [which enfolds latching of correct letters]. Unfortunately, we have also had to deal with a lot of Darwinist rhetorical games that drag distractive red herrings, lead off to strawman misrepresentations soaked in ad hominem mischaracterisations, which are then ignited to cloud, confuse, poison and polarise the atmosphere. (Apart from this factor, we probably would not have had several threads at UD since about March running over 1,000 comments on a plainly SECONDARY matter. There are threads where I have actually been rebuked for pointing out the primary issue, as being distractive from the "main" point!) I trust this sets the matter in clear enough context. GEM of TKI PS: Lest we forget, here arte the showcased runs of Weasel from 1986: _________________ We may conveniently begin by inspecting the published o/p patterns circa 1986, thusly [being derived from Dawkins, R, The Blind Watchmaker , pp 48 ff, and New Scientist, 34, Sept. 25, 1986; p. 34 HT: Dembski, Truman]: 1 WDL*MNLT*DTJBKWIRZREZLMQCO*P 2? WDLTMNLT*DTJBSWIRZREZLMQCO*P 10 MDLDMNLS*ITJISWHRZREZ*MECS*P 20 MELDINLS*IT*ISWPRKE*Z*WECSEL 30 METHINGS*IT*ISWLIKE*B*WECSEL 40 METHINKS*IT*IS*LIKE*I*WEASEL 43 METHINKS*IT*IS*LIKE*A*WEASEL 1 Y*YVMQKZPFJXWVHGLAWFVCHQXYPY 10 Y*YVMQKSPFTXWSHLIKEFV*HQYSPY 20 YETHINKSPITXISHLIKEFA*WQYSEY 30 METHINKS*IT*ISSLIKE*A*WEFSEY 40 METHINKS*IT*ISBLIKE*A*WEASES 50 METHINKS*IT*ISJLIKE*A*WEASEO 60 METHINKS*IT*IS*LIKE*A*WEASEP 64 METHINKS*IT*IS*LIKE*A*WEASEL _________________ These are runs of generational champions, used to seed the next successive generation. For over 200 cases of letters that go correct, there are no reversions thereafter that are seen in the sample -- and since we can easily see how persistent incorrect letters often are [E.g. cf the initially incorrect W, C Z and M in the first run and the Y, P , F and Q in the second] -- this strongly argues that there are none; especially given the gushing description of "cumulative selection" in BW.kairosfocus_{August 31, 2009
August
08
Aug
31
31
2009
04:34 AM
4
04
34
AM
PDT}

nephmon - you and I may already understand how a program that has no latching mechanism may still display latching behavior, but it is nice of kf to explain it to this audience, even if he is a bit long-winded about it. Let us remind ourselves that kfocus has defined Q as
kf @185 Q is — on second thought — number of generations multiplied by size of generations
And he has told us (@163) that Weasel is the same as the "Proximity Reward" search (this, at least, is true). I am surprised that kf (@222) would then showcase Weasel runs with Q = 999 x 21 and 500 x 31 As shown in the D&M paper, Q for a Partitioned Search has a median of 98, and should rarely go above 160. But according to kf, his didactic examples of Weasel runs have values of Q of 20,979 and 15,500. And he still wants to argue that D&M's use of eqn22 to describe Weasel is accurate.DNA_Jock_{August 31, 2009
August
08
Aug
31
31
2009
04:28 AM
4
04
28
AM
PDT}

Prev 1 … 3 4 5 6 7 … 13 Next

You must be logged in to post a comment.

Leave a Reply