Uncommon Descent Serving The Intelligent Design Community

We cannot live by scepticism alone

Share
Facebook
Twitter
LinkedIn
Flipboard
Print
Email

Scientists have been too dogmatic about scientific truth and sociologists have fostered too much scepticism — social scientists must now elect to put science back at the core of society, says Harry Collins.

Read more…

Comments
StephenB, "Whatever books you are reading, burn them and start reading G. K. Chesterton and C. S. Lewis." I'm glad to see another fan of these men, as I am.Clive Hayden
April 7, 2009
April
04
Apr
7
07
2009
02:19 PM
2
02
19
PM
PDT
DK: That the samples were chosen for "pedagogical" -- in fact intended RHETORICAL -- impact is irrelevant to their likely typicality of the outputs circa 1986. In fact they were plainly representative of "good" -- i.e fairly fast -- runs of Weasel at that time; it being evidently imagined that a steady march showed the power of the approach. Whoops! In this case, there is also -- as I stated in outline [adn which DK does not address ont eh specifics] -- excellent reason to infer that the Weasel output circa 1986, as published in sample form is the slice of the cake that has in it all the ingredients of the relevant whole. And, blood samples are not exactly random or stratified random samples. (Onlookers, for significant populations MOST beyond- law- of- large- numbers samples [~ 25 - 30+] will be typical, not atypical of the population. And, that is a statistical theory result. That is, in fact, the valid form of the layman's crude "law of averages." But, then, that is a core problem on the ID issue: evo mat advocates do not wish to acknowledge teh import of dominance of a population by sheer relative statistical weight. But, that is the premise of the well established Fisherian elimination approach to inference testing.) In this case we are looking at 300+ letters from the output, predominated by latching letters, i.e. some 2/3 of all mutated letters are o/p latched, with NO cases of a correct letter being caught in the act of reversion to incorrect status. As current discussionin the other thread shows, on charity we can take it that Dawkins' latching mechanism was implicit, i.e my T3 from 346 - 7 above. But all the above does not shift the bottomline: the whole Weasel project is distracftive from the issue of credibly getting to the shores of islands of functionality in large config spasces, large here being beyond 1,000 bits of info capacity as a good rule of thumb. The whole universe we observe acting as a search engine, is unable to seek out 1 in 10^150 of such spaces, i.e. islands of function are maximally unlikely to be found by random walks from arbitrary initial points, and until function is established, differential functionality cannot lead to hill-climbing. Targeted search that promotes non-functional configs on proximity is not a valid approach. And, by Dawkins' confession, that is what Weasel is. GEM of TKIkairosfocus
March 17, 2009
March
03
Mar
17
17
2009
04:46 AM
4
04
46
AM
PDT
kairosfocus [524], you need to distinguish between a statistical sample, which makes claims to randomness, and a pedagogical example, which highlights a tendency. That the few examples (the number of letters shown is miniscule compared to the total population) says nothing -- I repeat, nothing -- about whether latching occurred in the original program.David Kellogg
March 17, 2009
March
03
Mar
17
17
2009
04:08 AM
4
04
08
AM
PDT
PS: I forgot to give the "address" of the inadvertent case study on evidentialist form selective hypertskepticism provided by DK: 522 in this thread, just above my citation and comment.kairosfocus
March 17, 2009
March
03
Mar
17
17
2009
02:05 AM
2
02
05
AM
PDT
7 --> However, the 1986 Dawkins Weasel's OBSERVED steady lockstep march to the target (even in the absence of functionality) is in fact an obvious clue to the wider and crucial fact that can be (and has been) shown -- also from Mr Dawkins' words -- that the Weasel program is a question-begging, strawmannish distraction from the real challenge of bio-information creation: not to hill-climb to optimal forms of a functioning entity, but to get to the initial shores of islands of complex function. Invention is different, utterly different, from improvement; in short. 8 --> So, it is no surprise that within a year or so, in the 1987 videotaped version, the OBSERVED dominant behaviour of letter-latching, is replaced by flickbacks on correct letters, which occur fairly frequently. This contrasts glaringly with the lockstep advance we can see in the published 1986 samples. 9 --> That is, the 1987 outputs are not reasonable counter-evidence in the teeth of the published printoffs from 1986, where we see 200+ cases of latching in 300+ possible places for letters, and NO . . . repeat, NO, zip, zilch, nada . . . cases of flickback from a correct letter. 10 --> Therefore, back on the thread's topic: your captioned assertion of "no evidence" is a classic of evidentialist form selective hyperskepticism, in the teeth of something you cannot challenge on the merits:
(a) the printoffs are Mr Dawkins' own, which (b) can be presumed to present his "best foot forward" outputs circa 1986, (c) in these samples we see latching as the dominant feature of the output samples. (d) Then, the behaviour of both Mr Dawkins' Weasel circa 1987, and of many other implementations since, consistently move away from the letter latching, which (e) shows just how well-taken the point on that dominant behaviour was. (f) In all cases, "successful" Weasel type programs -- the ones that march forward to the target -- do not realistically address the key challenge of the quantum leap in functionality that is required for getting to the shores of islands of effective bio function, so that NS or other hill climbing algorithms could then -- at least in theory -- optimise off competition.
11 --> So, what do we see in the teeth of such facts and implications: the classic dismissive claim that in fact decisive counter-evidence to the evo mart case is "no evidence." A very sadly familiar claim, and yet again -- as usual -- an ill-founded one. ________________ BOTTOMLINE: This thread has shown all too clearly and convincingly from inadvertent case studies provided by the evo mat advocates, the pernicious effects of mind- endarkening selective hyperskepticism. Oner hopes -- and prays -- that at least some such will wake up to their peril and will turn back before it is too late, too late for our civlisation. GEM of TKIkairosfocus
March 17, 2009
March
03
Mar
17
17
2009
01:59 AM
1
01
59
AM
PDT
Mr Kellogg: First, that threadjacking attempts trace to the early comments in this thread do not -- and cannot -- excuse such. For: one wrong does not justify another. However, such threadjacking has aptly served to illustrate the thesis of the original post by providing inadvertent case studies on selective hyperskepticism in action in some of its most virulent forms. Now, re the latest gem of a case study; the claim that letters are fixed once correct has been offered based on no evidence whatsoever. 1 --> Now, onlookers, let us observe carefully: the printoffs from 1986 are the published claims by Mr Dawkins to document his case. I have used his own evidence to highlight what is going on. (This is of course not irrelevant to GLF's wager rhetorical tactic.) 2 --> I am sure you, DK, are familiar with sampling theory: a sample that is credibly reasonably representative of an entity will be a useful analytical stand-in for it. A blood sample is a useful diagnostic tool indeed, and a slice of a cake has in it all the ingredients. 3 --> In this case sampling every tenth generation's "winner" as well as in some cases the startup and the finsishing off generations, gives a sampled space of over 300 letters. It is also reasonable that such a sampling will not correlate with any reasonable algorithm [unless you are implying a degree of manipulation of algorithm and/or output by Mr Dawkins that is beyond all reason]. 3 --> Of these 300+ possible instances, over 200 are latched correct letters. Indeed, in his 1986 printoffs, NEVER do we see a correct letter that reverts. Thus, we can rest assured that this is typical "good" behaviour of the program, and that as at 1986, Mr Dawkins (no mean rhetor) saw no serious problem with that; it probably seemed to be a further illustration of just how good such search was -- see, it even marches forth steadily. 4 --> Further to this, we see that the runs hit the target in 40+ and 60+ generations, with a consistent ratio between latched letters and sampled letters. Stable, consistent behaviour. We are thus credibly dealing with the same algor in both cases. 5 --> And, in both cases the time to target is consistent with good runs for a letter-by- letter latching partition algorithm (notice how we have significant early latching . . . ). 6 --> On preponderance of evidence, we have cases of the partition algor in action, and in the case of letter latching of outputs, that is morally certain, i.e. beyond REASONABLE doubt. [Note my use of longstanding terms describing reasonable degrees of warrant on matters of fact. MOST scientific theories are only provisionally warranted to preponderance of evidence. Indeed, Kuhn, Lakatos and Feyerabend etc point out that scientific theories usually face empirical data that does not fit. That is, they are "born refuted, live refuted and die refuted" in the Popperian sense -- one of the reason's why Popper's falsificationism is far less than the whole story. Science advances by relative success of research programmes, not by knockout falsificationist exercises. And right now ID is on the march, not evo mart, once we have begun to recognise the significance of functional complex information.] [ . . . ]kairosfocus
March 17, 2009
March
03
Mar
17
17
2009
01:58 AM
1
01
58
AM
PDT
crater [521], my point is that the claim that letters are fixed once correct has been offered based on no evidence whatsoever. KF cited a print run that contains a small minority of the total (In fact, if you include all mutations of all generations, it's very small indeed.)David Kellogg
March 16, 2009
March
03
Mar
16
16
2009
01:52 PM
1
01
52
PM
PDT
KF, do you ever cite a complete print run?
Good point, David. Our side (pro-ID)often dismisses work into understanding the bacterial flagellum by stating that the evolutionists must provide a step-by-step explanation of how it evolved. It is only fair that we should rise to the same level of evidence.crater
March 16, 2009
March
03
Mar
16
16
2009
06:22 AM
6
06
22
AM
PDT
KF, This thread was off-topic from the very first comment, so it's hardly fair to complain that later developments mark a particular offense.David Kellogg
March 16, 2009
March
03
Mar
16
16
2009
05:34 AM
5
05
34
AM
PDT
KF, do you ever cite a complete print run? The example from BW (as reprinted in the Creation article you link to) includes only about 10% of the selected character strings.David Kellogg
March 16, 2009
March
03
Mar
16
16
2009
05:32 AM
5
05
32
AM
PDT
Oramus: Thanks. Given the sad deterioration into selectively hyperskeptical threadjacking on the evo mat side of this thread, your comment is very necessary and most welcome. Thanks again. I trust others will pause and let the clouds of noxious smoke form ad hominem soaked burning strawmen clear enough to address the matter at length on the merits. Thanks, a third time. GEM of TKIkairosfocus
March 16, 2009
March
03
Mar
16
16
2009
04:25 AM
4
04
25
AM
PDT
DK and JayM: First, let me cite the key relevant statements on the material issue at stake: ______________ [Excerpt from 111 in the December thread, by GEM:] Weasel 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. [Citation from BW ch 3 by Dawkins, on Weasel:] We again use our computer monkey [NB: to select a specific target functional case form 28 27 state elements identifies to 1 in "only"~ 1.2 * 10^40, well below the credible challenge to get FSCI for OOL and body plan level biodiversity, which is what Hoyle had raised], but with a crucial difference in its program. It again begins by choosing a random sequence of 28 letters, just as before … 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. . . . . What matters is the difference between the time taken by cumulative selection, and the time which the same computer, working flat out at the same rate, would take to reach the target phrase if it were forced to use the other procedure of single-step selection [but just such single step to gain functionality is actually what is necessary for natural selection to compare on differential functionality]: about a million million million million million years. This is more than a million million million times as long as the universe has so far existed. ________________ Thus, at the outset, I have accurately characterised the Weasel program as a targetted search that rewards closeness to target without reference to fucntionality. This means that the bottomline I have so often stated is correct: it is not relevant to RV + NS. It is thus a rhetorical device not a reasonable educational aid. It is a misleading icon of evoloution, yet another one of a loooong list of such over the past 150 years. The material issue is therefore settled whether or not I have correctly observed that "once a letter is guessed it preserves it for future iterations of trials until the full target is met." But, on this SECONDARY ISSUE: by simply looking at and counting the published output of the 1986 version of Weasel [my objectors would have it that I can no longer safely use "printoffs" given hostile readings as above . . . ] -- the relevant case, onlookers -- we see from my just made comment on GLF, that:
By my count, the BW 60+ case has 123 repeats of letters published, without reversion, and the NS 40+ one has 85 [total of sample points if therefore 208; i.e. "dozens" was a conservative statement], and wee see a consistent average rate of finding and latching letters here per generation, i.e. about 2.
Now, gentle reader: if something happens with 208 instances in a narrow context of 308 letters in possible cases of repetition, total -- and, without an exception -- is this not a reasonable and quite obviously fair conclusion that the letters a la 1986, once found, are as a matter of fact latched in the output? Indeed, is not the observed letter latching a DOMINANT feature of the published output of the 1986 version? Whart best explains that? In 346 - 7, i have proposed: explicit latching and implicit quasi-latching. Either would work, pretty well, nbut the simplest explanation is the first, and wins by Occams Razor. In particular, to get quasi-latching so that in 308 sample points where it could be seen actual latching is seen 200 times and there are NO flick-backs, is telling. By contrast, the 1987 video plainly and to the casual observer even, shows a fairly high incidence of just the opposite, multiple letters flicking back and forth, having first been correct. So, it is further fair observation to conclude that the behaviours of the program in 1986 and in 1987 are VERY and even strikingly different. Thus, there are significant differences in the code in 1986 and in 1987. In that context of a case AGAIN shown beyond reasonable doubt, let us now take on a few sampled remarks: 1] DK, 509:Hilarious. When faced with video evidence over twenty years, Kairosfocus apparently believes Dawkins fudged the video Weasel demonstration in 1987 Appeal to mockery is of course a nasty form of ad hominem, especially int eh teerh of objective evidence as cited. FYI, DK: he video evidence as compared tot he printooffs from 1986 SUBSTANTIATES my point that there is something dramatically different in the behaviour of the Weasel program circal 1986 - 7. As to "fudging." the REAL fudging I have pointed out, is that Weasel, from get-go is NOT BLIND watchmaker at work. I have shown that from Mr Dawkins' own words. Just as objectively. On the secondary issue of latching: again, I have shown that as a matter of fact, latching dominates the OUTPUT as published in 1986. By the time of the 1987 video, the output does not latch on correct letters, per the videotaped run. The difference is obvious, to those willing to look, and truthful enough to stick to the evidence not the rhetoric. 2] JayM 511: you have shown a complete and utter inability to admit to even the most trivial error, despite it being clearly and unambiguously demonstrated. There is a point where I made a mistake above. I mistakenly said the 43 generation run was from BW ch 3. I immediately corrected it on seeing that I was in error. Jay, by contrast, if you look carefully at the above you will see that -- OBJECTIVELY -- in 1986, latching is a dominant feature of the output of Weasel. By 1987, this is not so. Something has therefore changed, and it is obviously the program modules. I am not saying that Dawkins in 1987 presented Weasel as being identical to his original version, I am saying that the 1987 version is sufficiently different that we cannot safely infer form its behaviour to that of the 1986 program; which has a very different output pattern. I have therefore looked, consistently, at the merits of the matter, and it is clear that I am nort wrong on the evidence. 3] Why should anyone even attempt to have a rational conversation with you in the future when you have demonstrated so clearly that you do not value the truth if it means admitting error? Jay, what I have demonstrated is that I insist that we face the evidence and draw our conclusions in its light. Why not specifically address the evidence I have pointed out? If there is no observed letter latching in the 1986 published outputs, please show me the evidence; it is acknowledged by one and all that the 1987 video does NOT latch. I may be off by a few letters on my counts, but hey are in the ballpark, I am sure. Similarly, it is obvious to anyone watching the 1987 run that my description of the run is accurate: the runs frequently flick back from correct letters. This sharply contrasts with the 1986 outputs, which are very stable once a letter is correct, tot he point of actually dominating the published output. _______________ Are you familiar with the old psychology experiements where a group is in a room looking at a display on a screen, and the majority says something that is actually obviously wrong? After a time of repetition, the only true subject of the experiment will as a rule adjust his view to agree with the others, as he has come to doubt his own eyes. THAT is how dangerous a selectively hyperskepticism- based pseudo-consensus is. GEM of TKIkairosfocus
March 16, 2009
March
03
Mar
16
16
2009
03:56 AM
3
03
56
AM
PDT
BTW, to whom it may concern, may I suggest the creation of a conversation tracker, where every poster can see the record of their comments on various threads. I have seen this tracker in action on another forum and it works great.Oramus
March 16, 2009
March
03
Mar
16
16
2009
03:42 AM
3
03
42
AM
PDT
KairosFocus, I am heartened to see you keep your cool and professional demeanor under the circumstances. Like StephenB said on another thread, too many folks are looking to derail threads (think some folks call it threadjacking)for various reasons instead of speaking to the topic at hand. P.S. I would like to suggest to the Moderators that when commentary is clearly vearing off the road, to politely steer it back on the road, or request everyone on the new sub-thread hop on a new bus and create a new thread. In this case a new thread should have been started with the title as "Debunking Evolutionary Weasels" or something like that.Oramus
March 16, 2009
March
03
Mar
16
16
2009
03:35 AM
3
03
35
AM
PDT
Onlookers, This is so sad to see . . . Re GLF, 507 - 8: 1 --> GLF imagines that a sample point could only mean a cited generation; but plainly, (i) I have been discussing two print runs as publihed in BW Ch 3 and in new Sicentist in 1986, and (ii) I am PLAINLY discussing on a letter by letter basis. Simply go here and check:, you will see that indeed the number of samples on a per letter basis (which is exactly what is relevant to letter-latching . . . ) is dozens [count up the BOLDED letters and don't forget the spaces . . . ], and that the two 1986 print runs show exactly the letter latching output that I spoke of in December last, beyond REASONABLE doubt. 2 --> This sort of fallacious agenda-serving reasoning is sadly typical of the selectively hyperskeprtical dismissive rhetoric that this thread is pointing out as a problem, and unfortunately GLF seems to be a capital example of what that intellectual enslavement does to even an educated, intelligent person; one who doubtless in personal life is civil and perhaps even charming. And, that is why it is so dangerous to our civlisation. 3 --> Now, too, GLF has had many occasions to know that I spoke of the 1986 published tables of results and have repeatedly LINKED them, as already above. But, sadly, he wants to suggest that I am failing to present a reasonable case by not typing them in here in this blog thread. (But, this is a capital example of dismissive seectivley hyperskeptical rhetoric with an ad hominem in it: is not a link more than adequate for a reasonable person? And, in fact, I have reproduced one of the runs above, discussing the letter latching effect on a point by point basis that already goes well beyond a dozen sampling points; even tracing the history of the T by bolding it all the way through. Add in those from the other one that is linked and you are into the dozens. By my count, the BW 60+ case has 123 repeats of letters published, without reversion, and the NS 40+ one has 85 [total of sample points if therefore 208; i.e. "dozens" was a conservative statement], and wee see a consistent average rate of finding and latching letters here per generation, i.e. about 2. So, GLF has plainly failed in duties of care to be fair and accurate before making adverse comment.) 4 --> Similarly, he imagines that "printoffs" does not refer to the specific context in which I have repeatedly used this term: the 1986 runs published by Dawkins in BW and NS. 5 --> He then appeals to the similar skepticism of others, instead of addressing the fact that at 497 - 8, I took time to go back to the December remarks he quote-mined at 236 above, and showed just how these remarks substantiate themselves in light of what Mr Dawkins has said. [Observe this in light of the foolish wager argument that he has made.] 6 --> On the difference between mr Dawkins a la 1986 and a la 1987. Let us just note here that I already pointed out that mr Dawkins' statement is that his BASIC run ran up to 1/2 hr, and that he published runs of 40+ and 60+. His 1986 PASCAL version, at 11 seconds, and about 100 -- more or less the avg for partitioned search -- will thus reasonably run at about 10 generations/second. (We can safely take the 11 seconds report as a typical value.) 7 --> That sort of rate [think digital stopwatch here, for a simple illustration of how you can see the rough and ready estimate] is about what we see in the 1987 video, and it runs on and on, with correct letters in multiple instances flicking back and forth fairly frequently. That is there for anyone to see, as is the linked set of printoffs from 1986. Remember, onlookers: by count, the number of sample points for failure to flick-back in 1986 published data is northwards of 200. 8 --> Had that been the algorithm that was at work in the 1986 printoffs, we would NOT be seeing the 1986 pattern of output letters latching, across several dozen sample points of correct letters; at least, beyond reasonable doubt. (Onlookers, note how in 1986, Mr Dawkins also explicitly states that the slightest advance towards the target is rewarded, regardless of functionality. This fits very well with a partitioned search, and partitioned search fits very well with the published 1986 number of runs to hit the target. You need to ask yourself, why is it GLF does not address these little, material observations even as he hastens to make dismissive claims and to make ad honminem laced insinuations.] 9 --> So, prima facie, the 1987 version of Weasel is NOT the same as the 1986 one. THAT is the context in which I used "flickback module" as a shorthand for this point. (Pardon me, onlookers, for naively assuming that I would be read with a reasonable attitude. I will have to be more careful of the adverse reader next time.) As to Mr Dawkins, I have long since -- indeed, FROMT HE OUTSET, IN MY DECEMBER REMARKS THAT GLF QUOTEMINED -- pointed out that the WHOLE Weasel program project is a grand begging of the real question on the merits: getting TO shores of islands of functionality where that function depends on complex information. 10 --> So, while the output in 1986 and that in 1987 differ on observed letter latching [and strongly suggest that in 1987 we have an "improved version" at work], implicit or explicit letter latching are besides the material point, apart from that such behaviour is symptomatic of the underlying basic flaw and fallacy at work. 11 --> But if you reject the objective evidence and its reasonable best explanation, you can only resort to selective hyperskepticism, as we see ever so sadly and intensely on display above. 12 --> And, from the evidence in this thread inadvertently presented by others yesterday, it achieved just what such selectively hyperskeptical rhetoric is maeant to: poisoning the atmosphere against reasonable discussion. ________________ (BTW, other commenters supra, yesterday was Sunday. Can you think of a reasonable reason why after spending a little while very early in the morning, I was not available for the rest of the day? Why then the insinuations and personally cutting remarks above?] PLEASE, GLF ET AL, DO BETTER THAN THIS. GEM of TKIkairosfocus
March 16, 2009
March
03
Mar
16
16
2009
02:49 AM
2
02
49
AM
PDT
Apollos 496:
Just to note, certainly if the problem can be made smaller, such as finding words instead of entire phrases, it’s easier to solve. In that case, we only need to find combinations of 26^2 to 26^12 or so in most cases (still not trivial).
If you're thinking that to find any word of length n requires 26^n tries or something directly proportional to it by a constant value, that's not correct (That's the monkey with typewriters scenario). First of all, even with a random fitness landscape, if there are say 20,000 words of length N, your odds are 26^(n-4), (so 10000 times greater than 26^N.). But the whole point of Zachriel's program, is to show that the fitness landscape of english words is not random. If you run his program, starting with the word "be" for example, in about 2 seconds its generated hundreds of words completely at random with length of 8 and greater (just based on the few times I've used it). It starts by finding the shortest legal word closest to "be". If the word it find at random is "bet" it preserves that because its a legal word. Its also narrowing the fitness landscape, because english words are hierarchical. So you then it start mutating "bet" to find the next longest word after it. So you can very readily generate very long words and the process is not constrained by an exponential factor of the word length like you described.
I suspect however that assembling phrases from entire words (using words as informational bits instead of characters) would still present a significant problem if we don’t: a) dramatically limit the size and/or scope of our library;
There are millions upon millions of legal sentences. There are less than 100,000 legal words in English. Its much more likely to hit a valid twelve word sentence choosing from random words than it is a twelve character word choosing randomly from letters. With "Methinks it is a weasle", there are thousands of valid words you could plug in for weasle. The target in the Zachriel program is "Any phrase in hamlet" and you can specify the length of target phrase or sentence. And furthermore, the program works by maintaining viable forms at every intermediary step, where viability is also defined as being a phrase from Hamlet.
b) use a directed search, such as one that employs a heuristic.
If you're saying that "Any more viable biological organism" is a directed search, then I agree with you.
Moving from one 10 word phrase to another while retaining meaning at each step isn’t a problem from a teleological standpoint, but from a blind search it’s just as difficult of not more so.
What about moving from the phrase "Methinks it is a beagle" to "Methinks it is a weasel". Those are completely different sentences. Well, it depends on what you mean by completely different sentences.
I’m still stuck on the blind search, so forgive me if I’m essentially just talking to myself here. Thinking again about using a character based approach on a simple phrase, a modest goal might be to start with this: methinks it is like a weasel and morph it into another 28 character phrase: my dog pepper has three legs Would it be possible to do this by substituting 2, 3, or 4 characters at a time, with each substitution resulting in a grammatically correct, meaningful English phrase?
I wonder why your focussed in on morphing one sentence to a completely different sentence of the exact same length. Its like someone saying, " I want you to make an entire set of unrelated changes to your program, But I want the resulting program to be the exact length of the previous one, and furthermore, the new program must have all functional code, doing what I ask and nothing more." The Zachriel program works by starting with a very short phrase or word and gradually lengthening it, with each intermediary being a valid english phrase. No one thinks you could turn an octopus into a wolverine, with a further requirment that there be no actual change in the quantity of functional information in any intermediary form. What natural scenario is being modelled by your requirements. --------------------- KF [498]:
6 -> But, yet something else is quickly evident: the generations are piling up rather quickly, and just the time that the run is on screen strongly suggests that the runs to the target phrase here are now well beyond 100 generations. (And unless you knew that the 1986 published, sampled runs did the deed in 40+ and 60+ generations, you might not be alert to see that . . . )
Just wanted to point this out - With a population of 500 and a 5% mutation rate, the total number of generations to the target varied drastically with different runs - everywhere from 36 to over 500.JT
March 15, 2009
March
03
Mar
15
15
2009
06:56 PM
6
06
56
PM
PDT
Will KF respond? Inquiring minds want to know.David Kellogg
March 15, 2009
March
03
Mar
15
15
2009
05:19 PM
5
05
19
PM
PDT
Another pattern is visible in this failure to change views in the face of compelling evidence. That is the standard ID position that no evidence is ever good enough. I'm talking about evidence for actual biological evolution, rather than a toy program that does what it says, so what has happened here is merely a typical pattern. If a plausible mechanism is proposed for the natural evolution of, say, the flagellum, the ID response is to say that since the mechanism does not account for every minute step, the case means nothing. Since every step can be divided into further sub-steps, no evidence is ever enough. Just so: if a video from 1987 shows correct letters going wrong, KF responds something must have happened between 1986 and 1987. What if we found the original code, dated 1986? Odds are that KF would say the code was backdated and really from 1987. The only suitable evidence would be an ID advocate witnessing every step of Weasel's creation and initial run. It's also interesting that while KF gets offended that others (he claims) are calling him a liar, he is calling Dawkins a liar based on nothing but evidence from 1987 that KF is wrong. Shorter version: Because KF cannot be wrong, Dawkins must have fudged the 1987 run.David Kellogg
March 15, 2009
March
03
Mar
15
15
2009
10:37 AM
10
10
37
AM
PDT
kairosfocus @497 KF, prior to this exchange I had thought you to be one of the most gentlemanly and decent participants here. Your recent behavior has caused me to re-examine that view. You have been presented with overwhelming evidence that your claim is wrong. It was a relatively minor claim, and no shame would have accrued to owning up to your mistake. Instead, you have shown a complete and utter inability to admit to even the most trivial error, despite it being clearly and unambiguously demonstrated. Rather than recognize the simplest explanation, namely that you were incorrect and Richard Dawkins' statements and video show that, you accuse him of lying. As David Kellogg notes, this reflects incredibly badly on you. Why should anyone even attempt to have a rational conversation with you in the future when you have demonstrated so clearly that you do not value the truth if it means admitting error? How can we expect you to honestly consider evidence that might challenge your most cherished beliefs when you can't even own up to a minor mistake? I expected better of you. JJJayM
March 15, 2009
March
03
Mar
15
15
2009
09:58 AM
9
09
58
AM
PDT
Hilarious. When faced with video evidence over twenty years, Kairosfocus apparently believes Dawkins fudged the video Weasel demonstration in 1987. Which raises the question: Who are you going to believe, Kairosfocus or your lyin' eyes? On balance, we have learned something: namely, that something is irrevocably fixed. That something is not the letters in the Weasel program but the beliefs of Kairosfocus, which are "latched" and will not be moved no matter what the evidence.David Kellogg
March 15, 2009
March
03
Mar
15
15
2009
09:41 AM
9
09
41
AM
PDT
In either case, you have directly implied that I am a liar [at least as bad as GLF's insinuations above], which is a serious — and as I just demonstrated, slanderously false — accusation.
I don't see how you can say that with a straight face after accusing Dawkins of secretly implemeting a "flick back module". How is it that in the years since this all happened you are the only person to notice this module? Perhaps "liar" is too harsh a charge. Maybe it's just that you are congenitally unable to admit error and change direction.George L Farquhar
March 15, 2009
March
03
Mar
15
15
2009
04:46 AM
4
04
46
AM
PDT
Kindly read 497 - 8, to see why the 1987 Weasel is per observable characteristics, materially different from the 1986 one.
Just so everybody is clear on this. You are using as your evidence the printed run in new scientist #1 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 That is the example from New Scientist, 34, Sept. 25, 1986; p. 34. Yet you say
8 –> That is, when several dozen sample points that should not be correlated with the algorithm’s progress circa 1986 show NO observed flickbacks, and circa 1987 there is frequent flickback every few moments in multiple sites, something is cleasrly very different at work here.
Several dozen sample points? Where are you seeing these "several dozen sample points" exactly? There are exactly 5 sample points (excluding the first line) in the new scientist article and 6 sample points in the two runs given in Blind Watchmaker. The run above is the table that you refered to as
Table follows, will not reproduce cleanly here
Yet it only took me 30 seconds to type in the table that "would not reproduce" myself. For somebody who appears happy to type tens of thousands of words it's odd that you would not take 30 seconds to type in the table that proves your case. I think you unconsciously knew the table did not support your argument and that's why you did not bother to manually reproduce it. Where do you see "several dozen"?
And, the latching I have described is very evident in the 1986 outputs (a sample of several dozen points that should be uncorrelated to the program’s algorithm is well within the law of large numbers) but conspicuously absent in the 1987 one. (In the latter, Dawkins has tuned up his flick-back module just a little too much.)
Do you have any evidence other then then printed runs for the inclusion of a "flick back module"? Again, you say "several dozen points". Where are these several dozen points? I leave it to the onlookers to decide who has made their case.
2 –> Once a letter is guessed [a la 1986] it preserves it for future generations, cf printoffs, check
Where are these "printoffs" that show several dozen sample points, exactly? If this is your proof then please produce them.
And evo mat advocates love to complain that we quotemine them . . . ?
You are quotemining yourself. As I have said over and over I am not interested in discussing if Weasel is a targetted search, if it is a proper RV + NS model or not. The issue is if you are capable of representing your opponents arguments and examples correctly. On the balance of evidence, you are not capable of such. Tell me Kariosfocus, if you knew of this "flickback module" all along why are you only letting us know of it's existence now? If you knew all along that the "original" in 1986 was different from the version shown in 1987 then why have you waited until 2009 to mention it? Onlookers, judge for yourselves who is right. Kariosfocus with his accusations of dishonestly on the part of Richard Dawkins and attempts to move the goalposts to the blindness or otherwise of Weasel when that is not and has never been the issue or Richard Dawkins who, whatever else you may think of him, has now been accused of implementing a "flickback module" on the evidence of 6 lines of text that statistically would not have likely shown such behaviour in any case.
For, Dawins’ alleged BLIND watchmaker, especially for Weasel, is anything burt blind. That was my main pint ever since last December,and it was long since directly shown from Dawkins’ own mouth.
No, the main point is if your original claim, that correct letters are latched, is a misrepresentation or not. You would like nothing more then to talk about the blindness of the watchmaker, targetted search etc, as that distracts from the core issue, one which you are provably wrong on.
Enough has already been said on the GLF case, above. There seem to be no substantial new developments, sadly: prognosis is not good.
We'll see if the onlookers believe you or me, won't we? And no "substantial new developments"? I'd call finding a video showing you are wrong quite a significant developement. It's telling that even when all the evidence is against you you refuse to back down. Why? Do tell us more about this "flick back module". If word of this reaches Richard Dawkins and he denys such will you continue to claim it and so call him a liar? On the balance of evidence here there is only one liar, and it's not Richard Dawkins.George L Farquhar
March 15, 2009
March
03
Mar
15
15
2009
04:39 AM
4
04
39
AM
PDT
Kariosfocus, I certainly did not mean to imply that you were lying. If anyone gained that impression, I unreservedly withdraw any implied accusation. BTW wouldn't that be libel rather than slander? Anyway, my apologies for any unintended offence. Right, so you did watch the video. Dawkins states his weasel program is "homing in on a distant target" and "in another way it is a bit of a cheat". He indicates it is intended to demonstrate just "the power of cumulative selection".Arthur Smith
March 15, 2009
March
03
Mar
15
15
2009
04:22 AM
4
04
22
AM
PDT
Arthur: This is the third time in a few minutes. I watched a 40-something looking Dawkins on moths, CDs, safes being opened and the 1987 version of Weasel flitting across the screen, compete with a very long very fast run and correct letters flicking back and forth quite frequently for multiple letters, with what looks like the current mutant "spinning" away madly below a la old fashioned 1980's style spell-checks. (Which is what Weasel, in the end, is . . . ) Having watched, spending a good 10 minutes to do so, I wrote on my observations above. Which you have either not read nor else have chosen to dismiss. In either case, you have directly implied that I am a liar [at least as bad as GLF's insinuations above], which is a serious -- and as I just demonstrated, slanderously false -- accusation. GEM of TKIkairosfocus
March 15, 2009
March
03
Mar
15
15
2009
04:15 AM
4
04
15
AM
PDT
OK so you didn't. Fair enough.Arthur Smith
March 15, 2009
March
03
Mar
15
15
2009
03:57 AM
3
03
57
AM
PDT
Arthur: Kindly read 497 - 8, to see why the 1987 Weasel is per observable characteristics, materially different from the 1986 one. You will see that to make those remarks I had to waste 10 minutes of my time watching what is in the end a grand question-begging strawman argument. For, Dawins' alleged BLIND watchmaker, especially for Weasel, is anything burt blind. That was my main pint ever since last December,and it was long since directly shown from Dawkins' own mouth. And, the latching I have described is very evident in the 1986 outputs (a sample of several dozen points that should be uncorrelated to the program's algorithm is well within the law of large numbers) but conspicuously absent in the 1987 one. (In the latter, Dawkins has tuned up his flick-back module just a little too much.) QED. Please pay closer attention before commenting next time. GEM of TKIkairosfocus
March 15, 2009
March
03
Mar
15
15
2009
03:52 AM
3
03
52
AM
PDT
Apollos, 496: Very good insights and analysis. Once, even in a toy example, we insist on a reasonably serious degree of funcitonality indeendent of any partiular target, the search space challenge explodes. So, as Dawkins explicitly acknowledged in BW, from the outset it was known that Hoyle's challenge to get To the shores of islands of functionality was effectively insuperable. Weasel is a grand question-begging distraction from the real issue, for 23 years now: getting TO the shores of an island of functionality, in a world where genomes start at 600 - 1,000 kilo bits of info storage capacity for "simplest independent life forms." Let us therefore never forget that with 26 letters plus a space the odds of getting any one letter right in a fixed length target phrase is 1 in 27. a 27 state config space is utterly unlike one of order 10^180,000+. And, Mr Dawkins knew that or the like right formt he outset. THAT, too, is why test runs that use 5% odds of mutation show such fast progress. And, 1/20 of 28 is 1 - 2, so we see how this form of T3, let's call it T3a, will at least quasi-latch implicitly on thee output [which is the direct observation circa 1986; and the import of that point in my quotemined note]. It is actually worth the while to excerpt the quiotemined and twisted paragraph, to see how it sounds in the light of this context:
Weasel 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.
1 --> Weasel is targetted search, check. 2 --> Once a letter is guessed [a la 1986] it preserves it for future generations, cf printoffs, check 3 --> W rewards partial, non-functional success, check in light of Dawkins on "nonsense phrases" 4 --> Targetted search, not a proper RV + NS model, check. Indeed, here is an additional note from 107 in the original thread -- this is December 2008, NOT March 2009:
Nor did I ever claim that Weasel was the state of the art [i.e. I clearly am speaking of Dawkins and 1986]; that is putting words in my mouth to make up a convenient strawman. What I did say was that “Weasel is an apt example of a search algorithm that undertakes DIRECTED search, in an environment that is designed; and based on active, foresighted information fed in at the beginning.”
And evo mat advocates love to complain that we quotemine them . . . ? For shame! GEM of TKI PS: Cf. my original challenge that GLF quotemined, as excerpted at 404 above. notice how no-one has taken up the Hoyle-style tornado in Round Rock challenge or the Zener noise on a hard drive to rewrite the operating system by lucky noise challenges. No prizes -- see my anti-wager argument! -- for guessing why.kairosfocus
March 15, 2009
March
03
Mar
15
15
2009
03:45 AM
3
03
45
AM
PDT
Crater, 471: Please (especially given your track record . . . sadly) take time to read 346 - 7 before making comments on this issue again. No to mention, the original quote-mined remarks, as cited and linked at 404 and 407.
KF, I have read them. And, even to a sympathetic onlooker, it is obvious that they do not support your contention that the Weasel program was latching letters. I have tried to give you every benefit of the doubt and I can only conclude that your refusal to admit your (minor) error is due to pride. And such an exhibit of pride can only remind me (sadly) of Job 15:2-6.crater
March 15, 2009
March
03
Mar
15
15
2009
02:09 AM
2
02
09
AM
PDT
Kariosfocus: Did you notice David Kellog's comment upthread linking to the horses teeth or rather from the horses mouth: video of Dawkins demonstrating his weasel program It clearly shows letters are not fixed.Arthur Smith
March 15, 2009
March
03
Mar
15
15
2009
02:03 AM
2
02
03
AM
PDT
Crater, 471: Please (especially given your track record . . . sadly) take time to read 346 - 7 before making comments on this issue again. No to mention, the original quote-mined remarks, as cited and linked at 404 and 407. GEM of TKIkairosfocus
March 15, 2009
March
03
Mar
15
15
2009
01:41 AM
1
01
41
AM
PDT
1 2 3 18

Leave a Reply