The Simulation Wars

_{William Dembski

March 26, 2009

Intelligent Design}

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

I’m currently writing an essay on computational vs. biological evolution. The applicability of computational evolution to biological evolution tends to be suspect because one can cook the simulations to obtain any desired result. Still, some of these evolutionary simulations seem more faithful to biological reality than others. Christoph Adami’s AVIDA, Tom Schneider’s ev, and Tom Ray’s Tierra fall on the “less than faithful” side of this divide. On the “reasonably faithful” side I would place the following three:

Mendel’s Accountant: mendelsaccount.sourceforge.net

MutationWorks: www.mutationworks.com

MESA: www.iscid.org/mesa

Comments

Joe, I know it is a complete waste of time to attempt to get you to think about anything objectively, but... Weasel did/does not latch because it does not need to latch. Run Zachriel's version. Letters can and do revert. But because the closest match to the target phrase is chosen in each generation, the target is found without recourse to latching. Latching is unnecessary. Dawkins did not use latching! I can ask him to confirm it if you like! Remember Doug Axe. He responded unequivocally to a clarification.Alan Fox_{April 7, 2009
April
04
Apr
7
07
2009
08:12 AM
8
08
12
AM
PDT}

kairosfocus, I'll relay something from Dr. Elsberry. You write:
But equally, from the statistics involved, as the population size grows enough relative to mutation rates, skirts will begin to tell, and the program will first quasi-latch, with occasional reversions,t hen eventually as the far skirt begins to dominate the filter on sufficient population per the per letter mutation rate, we will see multiple reversions and the like, i.e. latching vanishes.
Emphasis added: Dr. Elsberry responds as follows (the rest of this post is his with minor edits):
Ignoring the misuse of "latching" (I suppose I can always-link, too), the above is precisely wrong. Increasing N, the population size, does not make it more likely that the best candidate in a generation will show a loss of a correct base from its parent, it makes it less likely. Let's recap the math one more time for the terminally obtuse:
Probability of a candidate changing a parent's correct base to an incorrect base = PCandidate_C2I = (1 - (1 - (u * (K - 1) / K))^C) Probability that a population will have at least one candidate that preserves all the correct bases from the parent of the previous generation = PPopulation_C2C = 1 - (PCandidate_C2I )^N Notice the power of N in there. As the population increases, the chance that the best candidate in each generation will change a correct base to an incorrect one falls off sharply, achieving the teensy-tiny reaches of small probability otherwise beloved of IDC advocates very quickly. We don't see changes of correct characters in the output of best candidates per generation in Dawkins' "The Blind Watchmaker" because it is by far the most probable outcome of a run of an accurate (and thus non-latching) "weasel" with a reasonable population size and a reasonable mutation rate. For the case where N=50, u=0.05, and the best candidate from the previous generation had 27 correct bases, the probability that the best candidate in the new generation still has all those bases correct is = 1.0 - (0.73614)^50 = 0.999999777
For [kf]'s assertion to be true, it would have to be the case that the value being subtracted from 1.0 would have to become larger with increasing N. Given that the term is a probability raised to the power of N, though, anybody with a thimbleful of knowledge (that a probability lies in the range [0 .. 1]) will recognize that can never be the case. It is the case that it grows smaller with increasing N, and thus the likelihood that the best candidate in each generation retains all the correct bases from its parent increases as N increases. [kf], please do avoid the betting games based on probability estimation. It is evident that these would do your wallet severe harm.
David Kellogg_{April 7, 2009
April
04
Apr
7
07
2009
06:11 AM
6
06
11
AM
PDT}

KF said: 'The real problem is that Weasel is targetted search that rewards mere proximity to the target, without requiring achievement of realistically improbable functionality first before hill climbing is permitted to proceed. As a result, it begs the question of getting to shores of function in the sea of relevant but non-functional configurations, and in so begging the question gives the misleading impression that the issue is not serious. but, I think a few rumours are beginning to get out into the sanctum of the peer reviewed literature, on what is really going on...' That Weasel is a targeted search and as such a 'bit of a cheat' has been known since 1986, from the moment that Dawkins himself qualified the programme in the Blind Watchmaker. The functionality issue seems misplaced. In terms of Weasel, functionality is defined as closeness to the target, so clearly phrases that better resemble the target are more functional than others. I still think you are blaming Weasel for not fulfilling objectives it never was meant for. All it purports to do is refute the notion that 'evolution is improbable because it requires many changes to happen at once' (tornado in a junkyard). That was its aim and it achieves it nicely. It had no further goals.faded_Glory_{April 7, 2009
April
04
Apr
7
07
2009
06:03 AM
6
06
03
AM
PDT}

elsberry sez:
"Latching" would require an internal mechanism with knowledge of "correct" states and the ability to protect "correct" states from mutational processes.
That is false. Given a target, a small enough mutation rate and a large enough sample size latching will always take place. It is inevitable.
That would be counter to what we know of biology, and, indeed, Dawkins himself thought that ascribing "latching" would be didactically wrong.
That is why Dawkins admitted the program was not a relection of biology. Keep getting your "knowledge" through elsberry. You will never "know" much of anything.Joseph_{April 7, 2009
April
04
Apr
7
07
2009
05:59 AM
5
05
59
AM
PDT}

Alan Fox:
==> Dawkins wrote “Weasel” to illustrate a point in a popular book. Whether “Weasel” is flawed has no bearing on the case for the theory of evolution.
What point was Dawkins trying to illustrate?
2==> The “Weasel” program did not latch. Wesley Elsberry has categorically confirmed this on Dawkins’ behalf.
It does latch for all the reasons I have already provided: 1- Given a target 2- A small enough mutation rate AND 3- a large enough sample size you will NEVER see a reversal
3==> The “Weasel” program did not need to latch.
It just does it as a matter of programming.Joseph_{April 7, 2009
April
04
Apr
7
07
2009
05:47 AM
5
05
47
AM
PDT}

I am not sure what the objection is against implicit latching. Isn't the entire point of Weasel to show that the random initial phrase evolves over successive generations into the target phrase? How would that ever be accomplished unless over time more and more correct letters are found and preserved? Objecting to implicit latching equals to objecting that the software works as intended! Regarding the 1986 published samples, these are samples of the winners from several generations, not unbiased samples from the entire population of each generation. Clearly the winners will have a much higher probability to show correct letters than random members of the population. After all having a significant number of correct letters is what helps make them winners in the first place! Hardly remarkable, nor evidence of sleigh-of-hand.faded_Glory_{April 7, 2009
April
04
Apr
7
07
2009
05:43 AM
5
05
43
AM
PDT}

5] The fundamental flaw with weasel and kin . . . . .. IS NOT LATCHING! The real problem is that Weasel is targetted search that rewards mere proximity to the target, without requiring achievement of realistically improbable functionality first before hill climbing is permitted to proceed. As a result, it begs the question of getting to shores of function in the sea of relevant but non-functional configurations, and in so begging the question gives the misleading impression that the issue is not serious. but, I think a few rumours are beginning to get out into the sanctum of the peer reviewed literature, on what is really going on:
To stem the growing swell of Intelligent Design intrusions, it is imperative that we provide stand-alone natural process evidence of non trivial self-organization at the edge of chaos. We must demonstrate on sound scientific grounds the formal capabilities of naturally-occurring physicodynamic complexity. Evolutionary algorithms, for example, must be stripped of all artificial selection and the purposeful steering of iterations toward desired products. The latter intrusions into natural process clearly violate sound evolution theory [172, 173]. Evolution has no goal [174, 175]. Evolution provides no steering toward potential computational and cybernetic function [4, 6-11] . . . . At the same time, we have spent much of the last century arguing to the lay community that we have proved the current biological paradigm. Unfortunately, very few in the scientific community seem critical of this indiscretion. One would think that if all this evidence is so abundant, it would be quick and easy to falsify the null hypothesis put forward above. If, on the other hand, no falsification is forthcoming, a more positive thesis might become rather obvious by default. Any positive pronouncement would only be labeled metaphysical by true-believers in spontaneous self-organization. Those same critics would disingenuously fail to acknowledge the purely metaphysical nature of the current Kuhnian paradigm rut [178]. A better tact [SIC -- should be tack] is to thoroughly review the evidence. Let the reader provide the supposedly easy falsification of the null hypothesis. Inability to do so should cause pangs of conscience in any scientist who equates metaphysical materialism with science . . . . While proof may be evasive, science has an obligation to be honest about what the entire body of evidence clearly suggests. We cannot just keep endlessly labeling abundant evidence of formal prescription in nature “apparent.” The fact of purposeful programming at multiple layers gets more “apparent” with each new issue of virtually every molecular biology journal [179-181].[Abel, 2009]
______________ I think the above notes should suffice to show the onlooker where the balance of the argument really lies on its merits. And, if you doubt me, go get some graph paper, a pair of scissors, some backing, notebook, calculator with stats features etc -- don't forget a dart or two [esp those neat mini-darts] -- and some time to try out the darts and charts exercise. That will teach you more, more surely and confidently, than rivers of digital ink and pro-weasel rhetoric. GEM of TKIkairosfocus_{April 7, 2009
April
04
Apr
7
07
2009
05:28 AM
5
05
28
AM
PDT}

Onlookers: A few brief notes: 1] Zachriel's neo-Weasel irrelevant to the issue of Weasel 1986. Z's neo-weasel may like the 1987 version not latch, but that has little or nothing to do with the import of those 200 latched letters from 300 that could in principle change in the 1986 o/p. Cf above at 183 to see what would have to happen for the o/p to revert and correct while producing apparently latched o/p letters. It is overwhelmingly improbable that the published 1986 runs did not latch [probably via implicit mechanisms]. 2] Mr Elsberry on latching: Apart from insisting on violations of my privacy and on the ID = creationism slander [Onlookers: note again how the trend of uncivil conduct by evo mat advocates is ever so predictably seen . . . ], he is failing to observe the distinction between explicit and implicit latching so is tilting at a strawman misrepresentation; as Mr Kellogg is above. I repeat; if you want me to respond to the specific claims being made, you will have to summarise the case here at UD, without slanders and privacy violations. That is a basic insistence on civil discussion of matters on the merits, and it is significant that just that is what the evolutionary materialism advocates and fellow travellers seem unable or unwilling to do. I will thematically respond in outline to what seem to be the major points, per the above comments. 3] LOLN and darts and charts Again, I have given an illustration of how as sample points scattered at random mount in numbers from one to a few to a few dozen, they as a rule become more and more representative of the bulk patterns of a trend or distribution. That is the practical (and very useful) import of the law of large numbers. Then, as the sample points get large enough, the far skirts ["tails"] will tend to show up, as Np --> a reasonable fraction of 1, i.e as number of sample pints and relevant distribution probability give us reason to expect to see at least one point. That is not controversial, it is a simple enough exercise that you can do yourself, by making up a bell-chart [as a typical case in point] and throwing or dropping darts on it from a range where they will more or less scatter evenly across its face. Or, you could pour grains of sand out on it. If you do it by darts, you can record the points by coordinates [draw up the chart on graph paper with marked axes], and by number in sequence. That way you can record trend related data that can then be plotted on say a +/- 3-sigma sequence chart. So, we can relate apparently static distributions to trends; using statistical process control techniques. (In fact, the point of the little darts and charts thought exercise is the exact point of the hypothesis testing approach that looks at whether or not a point is in the far skirts of a null hyp of chance based on one or another bell shaped distribution. Up to a certain confidence level, on the effective sample size, you are sufficiently unlikely to get in the far-skirt region, that one infers to an alternative hyp, that usually is in context some species or other of intentional action. AS A BIOLOGIST, MR ELSBERRY MUST HAVE DONE THAT LEVEL OF STATISTICS, SO HE MUST KNOW THIS. interested parties can look at my linked discussion of the Caputo case, in appendix 6 the always linked. Here is a nice simple intro. Of course there are the usual debates and disputes over any academic issue where rules of thumb and conventions are used, but the fact is that hyp testing dominates the real world, for some very good reasons.) Relevance to our case is of course that the binomial distribution applicable to letter mutation per a certain probability p is going to be a skewed bell or more or less reverse-J, i.e. skirted, distribution. Implicitly latched cases -- relevant to Weasel circa 1986 -- happen because the balance of mutation rates and per generation pop size is such that few far skirt members will appear, so that the odds of a substitution and reversion between the 1986 samples, will be negligible, by the time the run of champions -- on mere proximity to target without reference to realistic function [hence Mr Dawkins' "nonsense phrases"] -- hits the weasel sentence. 4] How that works for Weasel 1986 (not 1987, and not any neo-Weasels and quasi-weasels out there that do not latch): By the data published in 1986, and by the considerations in 183 above, that is what Weasel Circa 1986 was, beyond reasonable doubt. (It seems I need to again link on the fallacy of selective hypersketicism, and its fellow traveller, endless objectionism.) Back on point: in Weasel circa 1986 as published the no-change members of the population were selected as champion about 1/2 the time, and the next most common case that dominated the output was a single-letter goes correct mutation. That is, no-change and single letter mutations were the bulk of the distribution. Double letter mutations that substituted for a reverting letter another that was correct are an order of magnitude or more less likely, and to then have the reverted letter correct itself was even less likely by another order of magnitude or so. And, to lock in the likelihood of winning the championship by having a third letter go correct at the same time pushes us out even further into the skirt. In short, implicit latching is not so hard to understand, if one is willing to take the issue up on the merits objectively. [ . . . ]kairosfocus_{April 7, 2009
April
04
Apr
7
07
2009
05:27 AM
5
05
27
AM
PDT}

BTW Mr. M, I forgot to ask... Did you try Zachriel's version of "Weasel"? And I am beginning to wonder whether you think it is possible to win an argument by attrition. :)Alan Fox_{April 6, 2009
April
04
Apr
6
06
2009
10:23 AM
10
10
23
AM
PDT}

Mr. M, Others with more time and expertise have responded to you but it seems moderation has intervened. In particular, Dr. Elsberry has responded. I am afraid I didn't give much attention to your latest comments, relying on Hazel's assessment thazt there was no new content.Alan Fox_{April 6, 2009
April
04
Apr
6
06
2009
10:15 AM
10
10
15
AM
PDT}

KF writes, "I trust that sufficient has been said for the record." God, I hope so. To paraphrase a common saying, if you can't find anything new to say, don't say it.hazel_{April 6, 2009
April
04
Apr
6
06
2009
04:28 AM
4
04
28
AM
PDT}

kairosfocus,
If Mr Elsberry is prepared to state that Weasel did not IMPLICITLY latch, then he has a case to make, and one against stiff issues of evidence.
Oh come now. Dr. Elsberry has made such a case, you have been invited to respond, and you not done so. He provided a definition showing why latching runs counter to Dawkins's declared pedagogical goals, a precise statistical explanation for "weasel," and a shredding (his words, but they're true) of your faulty appeal to the law of large numbers. The math has been done, but you haven't dealt with it.David Kellogg_{April 6, 2009
April
04
Apr
6
06
2009
03:57 AM
3
03
57
AM
PDT}

PPS: Hot, just published, peer reviewed paper that hits hard on all the bases raised in this thread and previous ones, by Abel. Just one clutch of excerpts (note the tongue in cheek ironies of tone . . . ):
Attempts to relate complexity to self-organization are too numerous to cite [4, 21, 169-171]. Under careful scrutiny, however, these papers seem to universally incorporate investigator agency into their experimental designs. To stem the growing swell of Intelligent Design intrusions, it is imperative that we provide stand-alone natural process evidence of non trivial self-organization at the edge of chaos. We must demonstrate on sound scientific grounds the formal capabilities of naturally-occurring physicodynamic complexity. Evolutionary algorithms, for example, must be stripped of all artificial selection and the purposeful steering of iterations toward desired products. The latter intrusions into natural process clearly violate sound evolution theory [172, 173]. Evolution has no goal [174, 175]. Evolution provides no steering toward potential computational and cybernetic function [4, 6-11] . . . . At the same time, we have spent much of the last century arguing to the lay community that we have proved the current biological paradigm. Unfortunately, very few in the scientific community seem critical of this indiscretion. One would think that if all this evidence is so abundant, it would be quick and easy to falsify the null hypothesis put forward above. If, on the other hand, no falsification is forthcoming, a more positive thesis might become rather obvious by default. Any positive pronouncement would only be labeled metaphysical by true-believers in spontaneous self-organization. Those same critics would disingenuously fail to acknowledge the purely metaphysical nature of the current Kuhnian paradigm rut [178]. A better tact [SIC -- should be tack] is to thoroughly review the evidence. Let the reader provide the supposedly easy falsification of the null hypothesis. Inability to do so should cause pangs of conscience in any scientist who equates metaphysical materialism with science . . . . While proof may be evasive, science has an obligation to be honest about what the entire body of evidence clearly suggests. We cannot just keep endlessly labeling abundant evidence of formal prescription in nature “apparent.” The fact of purposeful programming at multiple layers gets more “apparent” with each new issue of virtually every molecular biology journal [179-181].
Muy interesante . . .kairosfocus_{April 6, 2009
April
04
Apr
6
06
2009
03:54 AM
3
03
54
AM
PDT}

KF:
If Mr Elsberry is prepared
Since we know your interest in making sure everything is stated accurately, KF, you probably should stop referring to Wesley Elsberry as Mr. and use his earned honorific of Dr.crater_{April 6, 2009
April
04
Apr
6
06
2009
03:43 AM
3
03
43
AM
PDT}

OOPS: 10's - 100"s of millions of bits worth of mutations to create novel body plans.kairosfocus_{April 6, 2009
April
04
Apr
6
06
2009
12:51 AM
12
12
51
AM
PDT}

For the record: Onlookers, it is clear that this thread has had some positive effect, but also that Weasel has proved itself to be a telling example of a misleading icon of evolutionary materialism. For, it is precisely a capital illustration of refusing to acknowledge the force of the point that once we deal with complex, information-rich functionality, we first need to credibly get to the shores of islands of function within the available probabilistic/search resources, before we can get to hill-climbing algorithms. And so, the basic point that observed cell-based life seems to have a threshold at about 600 - 1,000 k bits of information, shows that there is no credible pathway form prebiotic environments to first life, on the gamut of the observed cosmos. [This, I discuss in my always linked, section B.] (And, it also illustrates that the application of unrealistically high rates of beneficial mutations will give the illusion that such hill climbing is more likely to succeed than it is. In the real world, double simultaneous mutations are hard to get, and triples are a practical barrier. To incrementally get 10's - 100's of mutations.) A few [hopefully] wrap-up observations on points raised: 1] fG, 185: Weasel shows that there are other ways to find complex solutions than creating entire phrases at random, and therefore such objections are without merit. I believe this is really all that Dawkins wanted to illustrate with the programme. As he himself says, setting a pre-defined target is ‘cheating’ and is one reason why Weasel does not actually model biological evolution, but just one element of the proposed Darwinian process . . . Weasel does so by ignoring the issue that before one can hill-climb up Mt Improbable, one has first got to get to the shores of said island. And, on the precise point that Weasel would have to be illustrative, Dawkins has had to admit that it bears no reasonable resemblance to natural selection, the proposed BLIND watchmaker. In short, we see here in a nutshell the point of just how rhetorically misleading Weasel is. 2] AF, 186: Whether “Weasel” is flawed has no bearing on the case for the theory of evolution. Weasel -- along with a long string of other misleading icons -- illustrates that all too often, the evidence used to persuade the public [and even students] of the apparent credibility of evolutionary materialism is highly misleading. When a theory -- across decades -- is sold to the public and students in large part based on a clutch of misleading icons, that does not speak well of its underlying degree of warrant. 3] The “Weasel” program did not latch. Wesley Elsberry has categorically confirmed this on Dawkins’ behalf. The o/p of Weasel circa 1986, as published, is in a latched condition beyond reasonable dispute. (Onlookers, cf 169 above for why I say that. You will see that of 300+ letters that could change -- well beyond the point where the law of large numbers lends credibility to the representativeness of a sample [not to mention the well known practice of showcasing typical "good" results in scientific or general audience publications; what most likely happened here is that the obvious o/p latching was a warning flag that was not spotted as a flaw in the rhetoric until after the fact . . . ] -- the only ones that do change are the ones that are not "on target." Once a letter is on target, for 200+ cases it stays there, in some cases all the way through a run. And, as discussed in 183, it is simply not credible probabilistically for this to be seen in the samples o/p's if there is sufficient of a high mutation rate that the sort of multiple letter changes required to make reversions and recoveries happen so rapidly, were at work. Especially if -- per 40+ and 60+ gens to target -- no change was winning the in-generation championship race about 1/2 the time.) For the o/p to show 200+ cases where there is evident latching while in fact all along between sample points, reversions and corrections are going on, strains the limits of reasonable probabilities, as I have summarised at 183. There is therefore a question of mechanisms to explain that o/p latching, and the two credible candidates are: explicit and implicit latching. Moreover, so far as the reported record above and in previous threads shows, Mr Elsberry has confirmed that Mr Dawkins has stated that Weasel did not EXPLICITLY latch. On that, I and others have accepted that per preponderance of evidence, Weasel circa 1986 was per best explanation of its o/p pattern, implicitly latched. This, relative to co-tuning of population size per generation and mutation rates in a context that rewards mere proximity of "nonsense" -- i.e. non-functional -- phrases. [Thus is begged the Hoylean challenge that underlies Weasel's context: getting TO shores of islands of complex functionality before hill-climbing can reasonably be applied. And dismissing this as "single step" change etc is an evasion of the point.] Therefore, to make the sort of blanket statement that Mr Fox has in that context, is an improper appeal to authority at minimum. If Mr Elsberry is prepared to state that Weasel did not IMPLICITLY latch, then he has a case to make, and one against stiff issues of evidence. No, one cannot simply say that Weasel circa 1986 -- on its o/p as sampled -- was not EXPLICITLY latched as to mechanism, and that it was therefore also not implicitly latched. (Mr Kellogg's unfounded assertions to the contrary notwithstanding.) Mr Fox, I would appreciate not being misrepresented on this; here or elsewhere. 4] The “Weasel” program did not need to latch. This underscores to me that Mr Fox evidently has not done a sufficient examination of the case he is objecting to before making adverse comments. FYI: No one asserts that implicit latching will hold for all sets of values of population and per letter mutation rates. (Indeed, that is precisely how the difference between the published behaviour circa 1986 and the videotaped run on BBC Horizon circa 1987 -- note the jump in time -- is explained. SO, MR FOX HERE HAS INADVERTENTLY CONFIRMED THAT HE IS FOLLOWING MR KELLOG'S ATTEMPT TO INSIST THAT LATCHING MECHANISMS ONLY INCLUDE EXPLICIT ONES. This is utterly without warrant, and misrepresents what has been argued by the undersigned and others, to the point of being a plain strawman fallacy.) In fact, under relevant circumstances, Weasel will latch as evidently happened in 1986. (Please note again that in the published cases runs to target were 40+ and 60+ gens long, i.e. no change won the generation proximity championship about 1/2 the time. Under those circumstances, the multiple mutation skirts plainly did not dominate the filter. So, with a high enough proportion of no-change cases, and probably also only 1-change cases coming up with sufficient probability to make a difference, the program will latch, implicitly. For, long before enough population members will come up to make the multiple correct mutation or substitution skirt cases show up, the target will be hit. Again, this is tied to unrealistic beneficial mutation rates and the use of a proximity filter, without reference to threshold of reasonable function.) But equally, from the statistics involved, as the population size grows enough relative to mutation rates, skirts will begin to tell, and the program will first quasi-latch, with occasional reversions,t hen eventually as the far skirt begins to dominate the filter on sufficient population per the per letter mutation rate, we will see multiple reversions and the like, i.e. latching vanishes. All of this has been explained in details above and in previous threads, repeatedly; and under the term IMPLICIT LATCHING (and QUASI-LATCHING). There is therefore no good reason why such an objection should be seen at this late stage. Mr Fox, your "summary" is highly misleading; and that without any reasonable justification. Please, do better than this, next time. __________________ I trust that sufficient has been said for the record. GEM of TKI PS: JayM 182: Cf 88 above. Reflect on the force of "cumulative selection" and "[t]he computer examines the mutant nonsense phrases, the ‘progeny’ of the original phrase, and chooses the one which, however slightly, most resembles the target phrase" in light of the context of the sampled o/p of the program. the most natural reading of this is that there was a letterwise partitioned search that once letters hit, they were locked as successful. A further telling point on this is that this was the "natural" understanding of the Monash University -- pro Darwinist -- team, who came to the issue fresh; as they acknowledge, they had to be "corrected" by Mr Elsberry to line up with the standard Darwinist line on Weasel. (NB: Even in the implicit case, generally speaking, lockup happens once the phrase is hit: the program in effect masks off the target in either case, the only question is whether per letter or per phrase.)kairosfocus_{April 6, 2009
April
04
Apr
6
06
2009
12:49 AM
12
12
49
AM
PDT}

Well, kairosfocus, I don't see that you said anything new, and I've made and defended my points with specifics and clarity, I think, so there is no sense in repeating myself. At this point we'll have to let the onlookers decide who has made the best case for the various points we've discussed. And thanks to Alan for his summary.hazel_{April 5, 2009
April
04
Apr
5
05
2009
07:33 AM
7
07
33
AM
PDT}

Three points, Mr. M, (Onlookers, please note!) 1==> Dawkins wrote "Weasel" to illustrate a point in a popular book. Whether "Weasel" is flawed has no bearing on the case for the theory of evolution. 2==> The "Weasel" program did not latch. Wesley Elsberry has categorically confirmed this on Dawkins' behalf. 3==> The "Weasel" program did not need to latch. Anyone can download Zachriel's "Weasel" demo and see for themselves how there can be occasional reversions. There is even a handy counter included.Alan Fox_{April 5, 2009
April
04
Apr
5
05
2009
02:16 AM
2
02
16
AM
PDT}

I haven’t posted here before but I’m a regular lurker, so greetings to all contributors! I want to thank the participants in this Weasel discussion for their detailed analysis of the program. I read the Blind Watchmaker along time ago and I am aware of the ongoing discussions about the Weasel examples in the ID debate. I admit that I have never really thought in great detail about how the algorithm works, and in all fairness I more or less automatically assumed that it is so successful in finding the target because it explicitly latches correct letters. Because of the analysis here I now understand that explicit latching is not necessary at all, and that in fact it makes no sense in the context of what Weasel purports to illustrate: how the interplay between random mutations (of letters) and selection (of phrases) over many generations in a population finds a solution much, much quicker than single-step creation of entire phrases could ever do. So thanks for clarifying that! A long-standing objection to the Darwinian model of evolution is that the probability of it creating complex features is extremely small. These objections are often based on single-step assumptions (tornado-in-a-junkyard models). Weasel shows that there are other ways to find complex solutions than creating entire phrases at random, and therefore such objections are without merit. I believe this is really all that Dawkins wanted to illustrate with the programme. As he himself says, setting a pre-defined target is ‘cheating’ and is one reason why Weasel does not actually model biological evolution, but just one element of the proposed Darwinian process - random mutation of genes and selection of the fittest of the resulting progeny.faded_Glory_{April 5, 2009
April
04
Apr
5
05
2009
02:13 AM
2
02
13
AM
PDT}

3] H, 179 If one just naively looks at the published results in BWM and doesn’t think about the process that has been described to produce them, than one can say, “Hey, it looks like once a letter is set, it never changes.” But if one thinks about the process, and is aware that the published data is a highly selective subset of data that is a result of a process that moves towards a target, and is not a randomly distributed variable, then one realizes that the sample is entirely inadequate on its own to draw any conclusion about explicit vs implicit latching. Cf the above, on what is required on an implicit latching case for the 200+ samples to show apparent latching while all along there are reversions and corrections going on "offstage." (And of course the relevant population from the beginning is not eh generations as a whole but the march of published champions. It its for this population that o/p latching is credibly observed, and it is thus this population whose peculiarities need to be explained. This population is not to be presumed unrepresentative of itself. It is the evidence to be explained on inference to best explanation.) It is plainly far more plausible that the program latches, on considering the actual required processes to give the imagined offstage reversion and recovery, as discussed above. (That which is strictly logically possible is often so implausible probabilistically relative to a simpler explanation that for good reason we choose the simpler one.) 4] Explicit latching means latching at the mutation level. Implicit latching means non-latching at the mutation level. Not at all. In an explicitly latched case, members of a generation would compete on proximity, by which the champion is selected. Letters in the champion that are newly correct would then be added to the mask. Plainly, latching is based on population, but is letterwise. _________________ It should be plain that the idea that Weasel does not latch at all is riddled with implausibilities. The idea that per the 1986 data and statemets by Mr Dawkins, explicit latching is not a simple explanation and a very logical way to understand him, is equally implautible. Such is the import of the discussion on rewarding the least progress towards the target, the remarks on cumulative progress, and the published data that shows evident latching of letters leading to ratcheting progress to target. It is on the direct statement that he chose instead to not explicitly latch letters, that on the resulting preponderance of evidence that it is concluded that implicit latching based on co-tuning of population size and mutation rates -- both biologically very implausible -- with the all important selection on mere proximity without requiring reasonable functionality -- becomes the best overall explanation. GEM of TKIkairosfocus_{April 5, 2009
April
04
Apr
5
05
2009
01:55 AM
1
01
55
AM
PDT}

All: One of the subtler aspects of the Weasel issue is that it reveals the force of the underlying challenge to get to complex functional information, and hos deeply that challenge has been underestimated for many years by the evolutionary materialist establishment. For instance, above there was the suggestion on a calculation of confidence levels, one could in effect dismiss the idea that latching on the o/p is the best explanation for the pattern in the o/ps published by Mr Dawkins in 1986. But, this overlooked some interesting little points:
a --> To have 200+ cases in which EVERY reversion is reversed by the time of the next sample, one of two things would more or less have to happen: b--> Option a: phase 1: a substitution where by another letter becomes correct to replace the "lost" letter, and an advance to the target would more or less have to happen [to win the champion contest], then phase 2: within the same frame, a reverse substitution to the reversed letter, and a further advance [to again win the contest]. this requires two favourable triple mutations of rather special sorts. c --> Option b: only the substitution occurs, but the want of advance is "covered" by either no no-change cases, and/or by fortuitously being picked as winner by a tie-breaker soub module. This, for both the phase 1 and the phase 2. [of course a blended mode of the options is possible.] d --> Such requires so high a rate of mutations that it would then be utterly unlikely that we would not see incomplete phase 1 only cases, with 200 letters in o/p latched condition. e --> It is much simpler explanation that the o/p appears latched for the very excellent reason that the algorithm has a mechanism that locks, either explicitly or implicitly. And of these, on the o/p and Mr Dawkins' discussion in 1986 only, explicit latching, i.e. letterwise partitioned search, is the simpler.
So, we may now see that here is a failure to address comparative difficulties before putting up objections to the explanatory power of the observations that the o/p credibly latches and that its best explanation is a mechanism that does that, explicitly or implicitly. look at the latest remarks, on points: 1] H, 179: the only necessary difference between the explicit and implicit cases, as we have agreed on, is the additional rule in the explicit case that if the letter is correct, p(mut) = 0 Of course, this omits that the population size and mutation rates then have to be co-tuned to achieve implicit latching. It also omits that there is already present the information to put in the line,and that the line of code or two to do so is not hard to code, indeed it is simpler than co-tuning. 2] in the implicit case mutation is random with respect to fitness, and in the explicit case it is not. Here,t he insistence on a misleading "standard" term -- here a case of improper appeal to modesty in the face of claimed authority that fails to reckon with the point that authorities are no better than their facts, assumptions and reasoning -- has led the argument astray. There is no fitness, only proximity to target without reference to function. So, the issue is whether the target was conceived as a phrase or on a letterwise basis. And if "function" is reduced to matching the target, rewarding the smallest increment in progress to target, it makes sense to latch individual letters that hit the target. On this, the random search process is happening on a letterwise basis, and once letters hit home, ta da . . . they have achieved optimal function. Only, the concept of function and the rates of "beneficial" mutation being implicitly used beg the question of the Hoylean challenge to get to credible body-plan based life functionality on the observed capacity of chance + necessity only. [ . . . ]kairosfocus_{April 5, 2009
April
04
Apr
5
05
2009
01:54 AM
1
01
54
AM
PDT}

kairosfocus @177
First, just from Mr Dawkins’ description, it is pretty plain that understanding Weasel circa 1986 as latching per the published runs is a legitimate understanding.
No, it isn't. The page you refuse to read demonstrates this very clearly. I certainly can't see how you can draw your conclusion from what Dawkins actually wrote. Here is the relevant text from The Blind Watchmaker pages 47-8 (excerpted under Fair Use):
So much for single-step selection of random variation. What about cumulative selection; how much more effective should this be? Very very much more effective, perhaps more so than we at first realize, although it is almost obvious when we reflect further. We again use our computer monkey, but with a crucial difference in its program. It again begins by choosing a random sequence of 28 letters, just as before: WDLMNLT DTJBKWIRZREZLMQCO P It now 'breeds from' this random phrase. 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. In this instance the winning phrase of the next 'generation' happened to be: WDLTMNLT DTJBSWIRZREZLMQCO P Not an obvious improvement! But the procedure is repeated, again mutant 'progeny' are 'bred from' the phrase, and a new 'winner' is chosen. This goes on, generation after generation. After 10 generations, the phrase chosen for 'breeding' was: MDLDMNLS ITpSWHRZREZ MECS P After 20 generations it was: MELDINLS IT ISWPRKE Z WECSEL By now, the eye of faith fancies that it can see a resemblance to the target phrase. By 30 generations there can be no doubt: METHINGS IT ISWLIKE B WECSEL Generation 40 takes us to within one letter of the target: METHINKS IT IS LIKE I WEASEL And the target was finally reached in generation 43. A second run of the computer began with the phrase: Y YVMQKZPFfXWVHGLAWFVCHQXYOPY, passed through (again reporting only every tenth generation): 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 and reached the target phrase in generation 64. m a third run the computer started with: GEWRGZRPBCTPGQMCKHFDBGW ZCCF and reached METHINKS IT IS LIKE A WEASEL in 41 generations of selective 'breeding'.
Can you please show how any possible reading of the textual description (not your inferences from the very limited sample output) could even suggest explicit latching? JJJayM_{April 4, 2009
April
04
Apr
4
04
2009
07:09 AM
7
07
09
AM
PDT}

kairosfocus writes,
Second, the samples of 200+ letters that on observation latch beyond reasonable doubt on the champions is reason per LOLN, to conclude that the most reasonable mechanisms at work are explicit or implicit latching ones. Of these, and per the published evidence circa 1986, explicit latching is the simplest explanation. It is on the reported testimony of Mr Dawkins, that I have accepted that implicit latching per preponderance of evidence is the best explanation of Weasel 1986.
No, the simplest explanation is NOT explicit latching, for two reasons, one practical and one philosophical: a) practical reason: the only necessary difference between the explicit and implicit cases, as we have agreed on, is the additional rule in the explicit case that if the letter is correct, p(mut) = 0. This rule is not necessary in the implicit case. It is surely simpler to not have the rule than to have the rule. b) philosophical reason: in the implicit case mutation is random with respect to fitness, and in the explicit case it is not. Also, you invoke the “samples of 200+ letters” and state that, of the possible mechanism of explicit or implicit latching, “per the published evidence circa 1986” that explicit latching is the simplest explanation. Ironically enough, in 170 you write,
2 –> In that light, the mere presentation of statistical calculations that ignore that context is therefore worthless, save as a means to distract attention from the matter on the merits:
And yet your arguments about sampling fall prey to exactly what you warn against: they are out of context and therefore worthless. If one just naively looks at the published results in BWM and doesn’t think about the process that has been described to produce them, than one can say, “Hey, it looks like once a letter is set, it never changes.” But if one thinks about the process, and is aware that the published data is a highly selective subset of data that is a result of a process that moves towards a target, and is not a randomly distributed variable, then one realizes that the sample is entirely inadequate on its own to draw any conclusion about explicit vs implicit latching. And last, for the record: Explicit latching means latching at the mutation level. Implicit latching means non-latching at the mutation level.hazel_{April 4, 2009
April
04
Apr
4
04
2009
06:17 AM
6
06
17
AM
PDT}

kairosfocus, One more relay from Zachriel at anti-evo (because I know you won't go there). He is direct and clear.
kairosfocus: It is on the reported testimony of Mr Dawkins, that I have accepted that implicit latching per preponderance of evidence is the best explanation of Weasel 1986.
Zachriel: It took many threads, hundreds of comments, tens of thousands of words (and years in the case of Dembski) to reach an understanding of how Dawkins' Weasel works—but you still miss the essential point. The children in each generation may very well exhibit reversions in letters. There is *no* latching.
kairosfocus: Kindly note: implicit latching as a mechanism to explain o/p latching in the run of champions is NOT non-latching, regardless of what you want to assert.
Zachriel: That's the whole point of Weasel. Letters that match the target tend to become fixed across succeeding generations—even though there is no direct selection by letter.
kairosfocus: And of course I have never said that Weasel “has” to latch or has to latch explicitly to work in the sense of getting to target ...
Zachriel: The problem is that you don't understand evolution, so you still have no idea what Dawkins was trying to show. This is fundamental. Contrary to your strawman, in Dawkins' Weasel, mutation is random with respect to the fitness function. As Dembski et al. have had trouble absorbing this simple point, even when explained repeatedly, why should we expect them to have any insight into the limitations of such a simulation?
kairosfocus: In that light, the mere presentation of statistical calculations that ignore that context is therefore worthless ...
Zachriel: Yes, that was my point concerning your faulty appeal to statistical sampling.David Kellogg_{April 4, 2009
April
04
Apr
4
04
2009
05:57 AM
5
05
57
AM
PDT}

kairosfocus [178], I have not said that ID is "creationism in a cheap tuxedo." I'm just saying that the confusion between the two is made possible by ID. That's why I suggested the term "creationish." If ID wants to avoid the confusion, it should reject creationism outright and quit working hand-in-hand with creationists in publications. The insistence of (some) ID proponents that they are entirely separate viewpoints is regularly belied by their practice. For a start, you might quit citing creationists in your own work. As for the sampling issue, I give up. People who have a far better understanding of statistics and of evolutionary computing than either of us (for example, Zachriel) have tried to explain where you're wrong. Both Zachriel and hazel have been clear and patient. For a moment in looked like hazel and you had reached some sort of agreement. Alas, no. So I give up.David Kellogg_{April 4, 2009
April
04
Apr
4
04
2009
05:48 AM
5
05
48
AM
PDT}

PPS: Mr Kellogg: I repeat, I have refused to deal with slanderous incivility from the outset. For excellent reason; and YOU know or should know that design theory is quite distinct from Creationism -- i.e. your remarks just above, sadly, indicate you are joining in the "creationism in a cheap tuxedo" slander. Worse, up to date -- after several threads and dozens of explicit statements on the point -- you seemingly cannot even get what I have pointed out, repeatedly, straight:
[DK, 173:] No latching there, certainly not explicit latching: yet it will tend to show the same results that kairosfocus insists must be due to explicit latching.
Mr Kellogg, this is truly sad. How many times do I have to point out that:
a --> we OBSERVE evident latching of the o/p on the run of champions? b --> That this calls out for explanations based on mechanisms? c --> That there are two credible ones: T2, explicit latching and T3, implicit latching? d --> That on the evidence of the showcased o/p of 1986 and the statements made by Mr Dawkins then, that the simplest explanation is T2, explicit latching? e --> Or, that on the reported testimony that he did not explicitly latch even in 1986, the best explanation on preponderance of evidence is IMPLICIT latching due to co-tuning of [unrealistically high] beneficial mutation rates, population sizes and the [question-begging] mere proximity without reference to complex information based functionality filter? f --> Indeed, I have more than once explicitly corrected you on this very point, e.g. above when you tried to assert that "implicit latching" means "non-latching." g --> Zachriel's neo-Weasel, per your remarks, exhibits implicit latching behaviour, which is one of the two mechanisms I have taken pains to point out and explain. Kindly note: implicit latching as a mechanism to explain o/p latching in the run of champions is NOT non-latching, regardless of what you want to assert. h --> Do you not understand that the persistent, agenda-serving distortion of what another person says, twisting it into the opposite of what he said is uncivil misrepresentation? i --> FYI: I have NEVER said that latching on o/p's always requires explicit latching of letters. j --> FYFI: Just the opposite, I have put forth two mechanisms, pointing out that implicit latching with sufficient detuning becomes quasi-latching then complete breakdown of latching behaviour as the increased presence of far skirt multiple letter correct cases [due to unrealistic mutation rates -- cf on what it takes to see beneficial double mutations in real world genomes] and letter substitution cases with the unrealistic proximity filter causes a different behaviour to emerge.
Now, also, you have for some time now had an open invitation to present the argument here without uncivil behaviour and have not. That tells me that the argument you and otehrs have linked is not particularly strong in the context of the excerpts you may re-read at 88 above; which constitute telling admissions against interest by both Mr Dawkisn and Wikipedia. _________________ PPPS: As to Zachriel's statistical calculations, my comment is simple -- barking up the wrong tree: 1 --> Mr Dawkins made some pretty strong contextual remarks, per 88 above, that STRONGLY indicate that the published runs circa 1986 were very typical not atypical o/p; as is a commonplace in reporting scientific results -- showcasing "good" data. (Kindly note that socio- institutional context of what typical praxis is. that creates a very strong presumption on what was done in TBW ch 3 and New Scientist circa 1986. In short, it it those who argue against the import of the natural sense of what was said and showcased who have a stiff burden of proof to meet. I refuse rto go along with selective hyperskepticism that pretends otherwise.) 2 --> In that light, the mere presentation of statistical calculations that ignore that context is therefore worthless, save as a means to distract attention from the matter on the merits: Weasel per the 1986 o/p obviously latches on the run of champions. 3 --> And of course I have never said that Weasel "has" to latch or has to latch explicitly to work in the sense of getting to target, only that the OBSERVED latching per 1986 points to the REAL problem with it: targetted search without a realistic functionality threshold that rewards mere proximity. 4 --> So, it is just the opposite of a BLIND watchmaker at work. It is a misleading icon of evolutionary materialism, and, sadly, what you have just presented illustrates just how effective it is at being misleading. 5 --> Strawman mischaracterisations of what I have had to say and rebuttals to what I have NOT said, simply underscore the point that Weasel and its modern derivatives and kin, are in the end fundamentally a rhetorical exercise in ducking the challenge of credibly getting to complex bio-information on the gamut of our observed cosmos without intelligent direction. Sadly. GEM of TKIkairosfocus_{April 4, 2009
April
04
Apr
4
04
2009
04:04 AM
4
04
04
AM
PDT}

Hazel: First, just from Mr Dawkins' description, it is pretty plain that understanding Weasel circa 1986 as latching per the published runs is a legitimate understanding. As mot only Marks-Dembski and Roula Truman et al show, but Monash University's pro-Darwin site. All, as long since discussed in adequate details. Second, the samples of 200+ letters that on observation latch beyond reasonable doubt on the champions is reason per LOLN, to conclude that the most reasonable mechanisms at work are explicit or implicit latching ones. Of these, and per the published evidence circa 1986, explicit latching is the simplest explanation. It is on the reported testimony of Mr Dawkins, that I have accepted that implicit latching per preponderance of evidence is the best explanation of Weasel 1986. In that context, Weasel as videotaped for BBC circa 1987 is best seen as a de-tuned for video version, which shows fairly regular reversions; by sharp contrast with what one has the perfect right to see as the representative o/p for "good" runs, circa 1986 as published by Mr Dawkins. In both cases -- and As Mr Fox below you shows -- I again need to underscore, the material point on Weasel and the like since 1986, is that they were put up in the confext of the Hoylean challenge to get TO shores of functional complexity in the configuration space of organic molecules starting with pre-biotic environments; and also the need for novel body plans. But, since Weasel and close kin ate targetted search that reward mere proximity without reference to achieving complex function before allowing incremental hill-climbing, they are fundametnally flawed and misleading. In short, as Mr Fox et al should understand, you have to credibly get TO the shores of Island Improbable befoer you can climb to the mountaintops by your favourite hill climbing algorithm. (And for that even 1000 bits of information is such that our observed cosmos acting as search engine will be unable to get through more than 1 in 10^150 part of configs before using up its search resources. First life credibly starts at 600,000 bits [> 10^180,000 configs] and body plan innovation credibly requires increments of 10's - 100's of mega bits of information. Thus, origin of functionally specific, complex bio-information is a major roadblock to the materialist origins stories we are being told. And Weasel ducks rather than answers the question.) And, that is what I pointed out from December last, highlighting the o/p latching effect as illustrative of the fundamental flaw in Weasel and kin. the second of the below is the paragraph rthat was taken out of context to try to suppoert what has tunred into a camapign to give the rhetorical impressiont hat the main issue we have had is latching, and that this reflects a misunderstansding of the issue. On the contrary it is only a signpost pointing to the real issue -- Weasel et al are yet another misleading icon of evolutionary materialist philosophy imposed in the guise of origins science. Re-excerpting, as in 64 point 4 above [Please note, Mr Fox]:
[107:] the problem with the fitness landscape is that it is flooded by a vast sea of non-function, and the islands of function are far separated one from the other. So far in fact — as I discuss in the linked in enough details to show why I say that — that searches on the order of the quantum state capacity of our observed universe are hopelessly inadequate. Once you get to the shores of an island, you can climb away all you want using RV + NS as a hill climber or whatever model suits your fancy. But you have to get TO the shores first. THAT is the real, and too often utterly unaddressed or brushed aside, challenge. [111, excerpted paragraph used by GLF in his threadjack several threads back now:] Weasel [of course, in context, circa 1986] sets a target sentence then once a letter is guessed it preserves it for future iterations of trials until the full target is met. That means it rewards partial but non-functional success, and is foresighted. Targetted search, not a proper RV + NS model.
GEM of TKI _____________ PS: Mr Fox: On fair comment, you need to read the context of the discussion before making adverse comments, as already noted above. In particular, cf. 88 above, and the remarks on the REAL root problem with Weasel and kin, as reiterated above in this comment.kairosfocus_{April 4, 2009
April
04
Apr
4
04
2009
03:42 AM
3
03
42
AM
PDT}

For a confidence interval of 2% (letters from 0 to 26 ± 0.5), and a confidence level of 95%, we need to sample 70% of a population of 1000. (Interestingly, we only need to sample about 2400 in a population of a million or a billion for the same level of confidence. This is why a drop of blood containing trillions of particles can represent the composition of all the blood.) Anyone can see that Weasel doesn't require latching to work. With reasonably large populations, the best of the brood will only occasionally show a letter reversion. A typical sample of ten Mother Weasels will show the same results that kairosfocus insists must be due to latching. The appeal to sampling is obviously faulty as it is contrary to simple observation.
ZachrielIt is a shame that Zachriel and Kairosfocus could not communicate directly, if that is possible, (having observed Hazel's efforts. Zachriel's WeaselAlan Fox_{April 3, 2009
April
04
Apr
3
03
2009
07:32 AM
7
07
32
AM
PDT}

I don't think you get the sampling issue, and aren't going to, so I'll drop it.hazel_{April 3, 2009
April
04
Apr
3
03
2009
07:17 AM
7
07
17
AM
PDT}

Hazel; Follow up points: 1] 0.1 % I of course cited the blood sampling paradigm to highlight that relative scale of sample once the absolute scale is adequate, is irrelevant. And I gave a reason for that, too. One that as I pointed out earlier, you agreed with in the end. It is one that is backed up by the force of sampling theory: once samples are reasonably randomly or evenly scattered [and not pathologically correlated to the dynamics], and once the numbers become reasonable,t hey will pick out a picture of what is going on. So, the blood sample at 0.1% analogy is apt. For, we have a couple of good reasons to see that a 300+ sample of the letters will be representative of the Weasel runs circa 1986. 2] the 8 phrases Dawkins’ printed What is relevant is as just shown, the 300 sampled letters, with 200 showing latching. And, he did not print 8 phrases. Just go count: 15 x 28 = 420 letters; and letters is what is relevant. of these 200+ show the latching of the o/p. That from 300+ that count as variable. 3] We know, however, that the 9600 phrases (assuming N = 150 for illustration’s sake) are absolutely NOT the same. In the relevant aspect, they suffice to show a strong trend. There are two readily identifiable classes of letters, those that match and those that do not. As the run progresses, those that do not vary art random sharply, and once they find the correct value, they stop varying, per the evidence. There are 200+ instances of the latter in action, without counter-instance of reversion. And, that is a context that -- cf 88 onlookers -- explicitly speaks of cumulative progress to target. So, we have every reason to infer tha the progress of champions as illustrated is typical of "good" runs circa 1986. 4] you can object each time I use the word “fitness” if you wish, but it is the standard term and I’m not going to abandon it When a "standard" is misleading, it should be changed. Here, there is no fitness, just proximity without reference to function. thus, Weasel fails to illustrate NS or any other credible blind watchmaker. Worse, it dodges the issue that the Hoylean challenge was to ge TO shores of functionality in vastly beyond astronomical configuration seas of non-function. Hill-climbing to optimise function is by comparison a mere afterthought. GEM of TKIkairosfocus_{April 3, 2009
April
04
Apr
3
03
2009
07:13 AM
7
07
13
AM
PDT}

Prev 1 … 5 6 7 8 9 … 13 Next

You must be logged in to post a comment.

Leave a Reply