Uncommon Descent Serving The Intelligent Design Community

The Original WEASEL(s)

Share
Facebook
Twitter
LinkedIn
Flipboard
Print
Email

On August 26th last month, Denyse O’Leary posted a contest here at UD asking for the original WEASEL program(s) that Richard Dawkins was using back in the late 1980s to show how Darwinian evolution works. Although Denyse’s post generated 377 comments (thus far), none of the entries could reasonably be thought to be Dawkins’s originals.

It seems that Dawkins used two programs, one in his book THE BLIND WATCHMAKER, and one for a video that he did for the BBC (here’s the video-run of the program; fast forward to 6:15). After much beating the bushes, we finally heard from someone named “Oxfordensis,” who provided the two PASCAL programs below, which we refer to as WEASEL1 (corresponding to Dawkins’s book) and WEASEL2 (corresponding to Dawkins’s BBC video). These are by far the best candidates we have received to date.

Unless Richard Dawkins and his associates can show conclusively that these are not the originals (either by providing originals in their possession that differ, or by demonstrating that these programs in some way fail to perform as required), we shall regard the contest as closed, offer Oxfordensis his/her prize, and henceforward treat the programs below as the originals.

For WEASEL1 and WEASEL2 click here:

WEASEL1:

Program Weasel;

Type

Text=String[28];

(* Define Parameters *)

Const

Alphabet:Text=’ABCDEFGHIJKLMNOPQRSTUVWXYZ ‘;

Target:Text=’METHINKS IT IS LIKE A WEASEL’;

Copies:Integer=100;

Function RandChar:Char;

(* Pick a character at random from the alphabet string *)

Begin

RandChar:=Alphabet[Random(27)+1];

End;

Function SameLetters(New:Text; Current:Text):Integer;

(* Count the number of letters that are the same *)

Var

I:Integer;

L:Integer;

Begin

L:=0;

I:=0;

While I< =Length(New) do Begin If New[I]=Current[I] Then L:=L+1; I:=I+1; End; SameLetters:=L; End; Var Parent:Text; Child:Text; Best_Child:Text; I:Integer; Best:Integer; Generation:Integer; Begin Randomize; (* Initialize the Random Number Generator *) (* Create a Random Text String *) Parent:=''; For I:=1 to Length(Target) do Begin Parent:=Concat(Parent, RandChar) End; Writeln(Parent); (* Do the Generations *) Generation:=1; While SameLetters(Target, Parent) <> Length(Target)+1 do

Begin

(* Make Copies *)

Best:=0;

For I:=1 to Copies do

Begin

(* Each Copy Gets a Mutation *)

Child:=Parent;

Child[Random(Length(Child))+1]:=RandChar;

(* Is This the Best We’ve Found So Far? *)

If SameLetters(Child, Target) > Best Then

Begin

Best_Child:=Child;

Best:=SameLetters(Child, Target);

End;

End;

Parent:=Best_Child;

(* Inform the User of any Progress *)

Writeln(Generation, ‘ ‘, Parent);

Generation:=Generation+1;

End;

End.

WEASEL2:

PROGRAM WEASEL;
USES
CRT;

(* RETURN A RANDOM LETTER *)
FUNCTION RANDOMLETTER : CHAR;
VAR
NUMBER : INTEGER;
BEGIN
NUMBER := RANDOM(27);
IF NUMBER = 0 THEN
RANDOMLETTER := ‘ ‘
ELSE
RANDOMLETTER := CHR( ORD(‘A’) + NUMBER – 1 );
END;

(* MEASURE HOW SIMILAR TWO STRINGS ARE *)
FUNCTION SIMILARITY(A : STRING; B : STRING) : INTEGER;
VAR
IDX : INTEGER;
SIMCOUNT : INTEGER;
BEGIN
SIMCOUNT := 0;

FOR IDX := 0 TO LENGTH(A) DO
BEGIN
IF A[IDX] = B[IDX] THEN
SIMCOUNT := SIMCOUNT + 1;
END;
SIMILARITY := SIMCOUNT;
END;

FUNCTION RANDOMSTRING(LEN : INTEGER) : STRING;
VAR
I : INTEGER;
RT : STRING;
BEGIN
RT := ”;
FOR I := 1 TO LEN DO
BEGIN
RT := RT + RANDOMLETTER;
END;
RANDOMSTRING := RT;
END;

VAR
X : INTEGER;
TARGET : STRING;
CURRENT : STRING;
OFFSPRING : STRING;
TRIES : LONGINT;
FOUND_AT : INTEGER;
BEGIN
RANDOMIZE;

CLRSCR;

WRITELN(‘Type target phrase in capital letters’);
READLN(TARGET);
(* PUT SOME STRING ON THE SCREEN *)
TEXTCOLOR(GREEN);
GOTOXY(1, 6);
WRITELN(‘Target’);

GOTOXY(10, 6);
WRITELN(TARGET);

TEXTCOLOR(BLUE);

GOTOXY(1,13);
WRITELN(‘Darwin’);

TEXTCOLOR(BLUE);
GOTOXY(1,19);
WRITELN(‘Random’);

TEXTCOLOR(WHITE);
GOTOXY(1, 25);

WRITE(‘Try number’);

(* PICK A RANDOM STRING TO START DARWIN SEARCH *)
CURRENT := RANDOMSTRING(LENGTH(TARGET));

(* RUN THROUGH MANY TRIES *)
FOUND_AT := 0;
FOR TRIES := 1 TO 100000 DO
BEGIN

(* Darwin *)
OFFSPRING := CURRENT;
OFFSPRING[ 1 + RANDOM(LENGTH(OFFSPRING)) ] := RANDOMLETTER;

GOTOXY(10,13);
WRITELN(OFFSPRING, ‘ ‘);

IF( SIMILARITY(OFFSPRING, TARGET) >= SIMILARITY(CURRENT, TARGET) ) THEN
CURRENT := OFFSPRING;

IF( (SIMILARITY(CURRENT, TARGET) = LENGTH(TARGET)) AND (FOUND_AT = 0) ) THEN
BEGIN
(* TELL THE USER WHAT WE FOUND *)
FOUND_AT := TRIES;
GOTOXY(1, 15);
TEXTCOLOR(BLUE);
WRITELN(‘Darwin’);
TEXTCOLOR(WHITE);
GOTOXY(9, 15);
WRITELN(‘reached target after’);
GOTOXY(37, 15);
TEXTCOLOR(BLUE);
WRITELN(FOUND_AT);
WRITE(‘tries’);
TEXTCOLOR(WHITE);

GOTOXY(1, 21);
TEXTCOLOR(BLUE);
WRITE(‘Random’);
TEXTCOLOR(WHITE);
WRITELN(‘ would need more than ‘);
TEXTCOLOR(BLUE);
WRITELN(‘1000000000000000000000000000000000000000’);
TEXTCOLOR(WHITE);
WRITE(‘tries’);
END;

(* Random *)
GOTOXY(10, 19);
WRITELN(RANDOMSTRING(LENGTH(TARGET)), ‘ ‘);

GOTOXY(27,25);
WRITE(TRIES, ‘ ‘);
END;

GOTOXY(1, 20);
End.

Comments
Onlookers: Observe -- yet again -- the ever increasing degree of tangentiality just above, coupled to studious non-addressing of that which is central about weasel per the description in 1986 and the showcased examples. A few remarks are in order: 1] Dembski vs divide and conquer etc My remarks are primarily on the Weasel output and description c 1986 and its context, not on parsing what Mr Dembski may or may not have said. (After all, ID theory is not God and Dembski is not its inerrant prophet. That is I find a subtle, unwarranted implication of blind adherence to holy writ in too much of the above remarks by darwinist objectors, that needs to be corrected. Mr Dembski is a scientist-philosopher-mathematician, one of a long and distinguished tradition. but he is finite, fallible and fallen just as the rest of us are.) Now, as one who has had to program systems and as one who has had to plan things, I am familiar with hierarchical breakdowns of problems into sub-problems whose solutions when put together appropriately will solve the overall problem. And, that is what, in essence a divide and conquer approach is. The key issue on ratcheting-latching and partitioning is different in emphasis, and we need to get back to it: incremental progress to target by preserving correct letters in current generation champions (explicitly or implicitly) and then getting new ones by a guess and test procedure. When applied, we see cumulative progress to target, and this has been demonstrated for both explicit and implicit latching as a component of ratcheting. Demonstrated since April 9. And, demonstrated to observably happen with varying per letter mutation rates and pop sizes so that selection to demonstrate and showcase cumulative progress to target easily accounts for the showcased runs and associated descriptions. The EFECT of such ratcheting progress is that the overall problem of getting to the target -- and note again it is targetting and warmer-colder search strategies that reward non-functional phrases that make all weasels fundamentally dis-analogous to claimed mechanisms for evolution, thus misleading on the claimed prowess of BLIND watchmakers -- is divided up on a letter-wise basis. So, the divide and conquer effect is achieved through the ratcheting, cumulatively progressive action. 2] W1 doesn’t fit this definition. And so, you just redefine “divide and conquer . . . W1 is credibly -- on balance of evidence -- the likely original Weasel 1986. It shows implicit latching, thus ratcheting and cumulative letter by letter progress to target. And that is my main concern. In showing this pattern of behaviour on an implicit mechanism of latching of generational champions, it exhibits letterwise progress splitting up the overall problem into an effectively letterwise one. Whether that fits or does not fit someone's particular definition out there is irrelevant to my concern. Especially when -- pardon my directness -- such tangential minutiae effrectively constitute straining at a gnat while swallowing a camel. 3] Given the W1 code (and our endless discussions around “implicit latching”), and Atom’s Proximity Reward Search, it is clear that an algorithm that lacks an explicit latching mechanism may show latching behavior. On this we are all agreed. Armed with this knowledge, what “evidence” is there that supports the hypothesis that TBW has an explicit latching mechanism? What evidence is there that argues against it’s being a Proximity Reward Search? The only evidence seems to be that you and Dembski once thought that TBW was explicitly latched This is a case of twisting words from one context to another, to fit a wider rhetorical pattern. I have spoken in the context of empirical evidence and weight on balance. On balance of evidence -- and matters of fact are not settle-able to mathematical certainty -- W1 is the probable original weasel, c. 1986. W1 shows implicit latching in a way that is related to Atom's adjustable weasel. Consequently the best current explanation -- note the opposite in logic to a proof -- of the data and explanation in BW etc is that W1 was the original weasel and that CRD showcased latched runs in the range 40 - 60 or so generations. Further to this, implicit latching is of course a mechanism, which accounts for cumulative progress to target. On the personalities side, it should be noted that a comparison of the record will show that I set the discussion in the context of the try 1, 2, 3 cluster of alternatives: T1 -- pure random chance, T2 -- explicit latching, T3 -- implicit latching. T1 can account for data on a logical possibility basis, as random chance can mimic anything else in a contingent situation, but it is not probable. T2, on the direct factual evidence of 1986 in BW etc, is the simplest most direct account for the behaviour of Weasel. It is on the injection of indirectly communicated testimony that T3 becomes the better explanation on balance. W1 and W2 above further underscore this. Onlookers, observe the sharp contrast between the actual pattern of discussion and the tellingly ad hominem-laced caricature I have excerpted above. 4] Given that all the evidence is on the side of “implicit latching”, how about we stipulate that TBW is a type of Proximity Reward Search that shows latching behavior. Then we can move on. Noticed, again, the difference between empirically anchored provisional inference to best explanation -- the essence of science as process -- and the caricature that is again presented, and capped off with an accusation of "whining." There is, of course, no "all the evidence" to point to just one possible explanation, given the points in 3 just above. On best explanation on balance of evidence from several directions -- and the only effectively certain evidence is the state of the text of BW c 1986 -- T3, implicit latching is the best estimate of the state of Weasel c 1986, and W1 is the best candidate to be that original weasel. 5] It is impossible for implicit latching to behave that way. Changing all of the incorrect letters would require a 100% mutation rate, which would entail the correct letters not being preserved at all. So it is a fact that Dembski was referring to explicit latching. The algorithm that he describes is a square peg that cannot be stuffed into the round hole of implicit latching. Again, tangentiality dominates, while the principal issue lies unaddressed. I am but little concerned as to what Mr Dembski's estimates of the state of Weasel were at any given time -- save to note that T2 on the direct factual evidence of BW c 1986 is a better explanation than T1 or T3. Indeed,t hat is whyt he Monash U Australia biology folks initially understood Weasel as an explictily latched cumulative ratcheting search, and had to be "corrected" by Mr Elsberry on that. And, for T2 type algorithms, since a 100% mut rate may be analytically convenient [especially for purposes of illustration], but is irrelevant to the achievement of cumulative progress to target. (To see this point in action simply imagine W1 above -- only and no more than a single letter mutates in each child -- with the augment of forcing the seed into the population of children with certainty, not just high probability. The backstop dog or pawl is now certainly present and the filter will at least preserve the seed's state of progress to target with certainty. Latching has now been built in explicitly. In the implicit case, as we can see in the original post, in effect the W1 algorithm relies on the odds of getting through such a case; which are fairly high. Consequently on Dieb's calculation, 199 or so times out of 200, we will see implicit latching with a population of 100 per generation, and those who have run it say the pattern is to run to target in about 40 - 60 gens.) Implicitly latched mechanism can achieve the same effect of cumulative progress, and on adding the indirectly reported testimony of CRD c 2000 (which we were made aware of sometime in March this year) and applying the ancient documents rule to the W1 source code provided by Oxfordensis, T3 is the best current explanation. In short, the issue is best explanation on a cumulative body of evidence of various kinds. Evidence with quite different degrees of weight and type, but evidence relevant to the conclusion that on balance of evidence, W1 -- an implicitly latching, ratcheitng mechanism program -- is the most likely source of the showcased runs of 1986 in BW etc. _____________ And all of the aboved points back tot he real issue, the elephant int he middle of the room that so many so often avoid discussing: Weasel and kin are fundamentally dis-analogous tot he claimed power of CRD's BLIND watchmaker [chance variation plus natural selection, illustrative of undirected chance + necessity], but in fact inadvertently illustrate that intelligences can use constrained random variation and artificial selection on explicit or implicit knowledge of a target or the landscape of an objective function, to achieve a design. That is, these are inadvertent examples of intelligent design and its power to account for even evolutionary patterns of development. (That should be no surprise to one who knows that technologies evolve through intelligent direction in a competitive environment.) GEM of TKIkairosfocus
September 27, 2009
September
09
Sep
27
27
2009
04:06 AM
4
04
06
AM
PDT
Here is my take on these weasels: for short, w1 would fit Dawkins's algorithms, while w2 is something different. I can't exclude the possibility that w2 was used in the video of the BBC, but I'd prefer if it wasn't: it's just a randomized version of the hangman's game (Optimization by Mutation With Elitism). And it is (implicitly) latching as hell in the sense that a correct letter will not be changed in future generations....DiEb
September 27, 2009
September
09
Sep
27
27
2009
02:52 AM
2
02
52
AM
PDT
DNA_Jock is more charitable than I. I think that kairosfocus also needs to acknowledge the incontrovertible fact that Dembski was not talking about implicit latching in his descriptions of WEASEL. The algorithm described in the Nature of Nature conference, No Free Lunch, the EIL Weasel Math page, and the IEEE paper is as follows: For each iteration, the correct letters are held fixed, and all of the incorrect letters are randomly altered. This is a fact. Each of the above sources either makes the above bolded statement explicitly or presents math that entails it. It is also a fact that the above bolded sentence cannot refer to implicit latching. It is impossible for implicit latching to behave that way. Changing all of the incorrect letters would require a 100% mutation rate, which would entail the correct letters not being preserved at all. So it is a fact that Dembski was referring to explicit latching. The algorithm that he describes is a square peg that cannot be stuffed into the round hole of implicit latching.R0b
September 26, 2009
September
09
Sep
26
26
2009
09:57 AM
9
09
57
AM
PDT
DNA Jock, Using Dawkins description of cumulative selection and the illstration of CS using "weasel", the only inference one gets is that once a matching letter is found the search for it is over. IOW there isn't anything in what he says nor illustrates (in TBW) that would demonstrate found letters can change.Joseph
September 26, 2009
September
09
Sep
26
26
2009
06:22 AM
6
06
22
AM
PDT
kairos - My offer from post84 stands:
Here’s an idea: you stipulate that TBW Weasel does not contain an explicit latching mechanism, and we’ll all agree that it can, given certain parameters, show latching behavior. Then we can all move on to discussing the ‘distant target’ issue.
You keep repeating that TBW is probably an algorithm with 'implicit latching', based on the "preponderance of evidence". I am puzzled. Given the W1 code (and our endless discussions around "implicit latching"), and Atom's Proximity Reward Search, it is clear that an algorithm that lacks an explicit latching mechanism may show latching behavior. On this we are all agreed. Armed with this knowledge, what "evidence" is there that supports the hypothesis that TBW has an explicit latching mechanism? What evidence is there that argues against it's being a Proximity Reward Search? The only evidence seems to be that you and Dembski once thought that TBW was explicitly latched (Joseph's tortured interpretation of the word 'cumulative' is not evidence of explicit latching, after all). Given that all the evidence is on the side of "implicit latching", how about we stipulate that TBW is a type of Proximity Reward Search that shows latching behavior. Then we can move on. It is quite frankly hilarious for you to whine about distractions while you refuse to acknowledge the obvious.DNA_Jock
September 26, 2009
September
09
Sep
26
26
2009
05:34 AM
5
05
34
AM
PDT
--kf if we want to go back, we should revisit this statement which started the current outbreak of weasels:
P.S. Our critics will immediately say that this really isn’t a pro-ID article but that it’s about something else (I’ve seen this line now for over a decade once work on ID started encroaching into peer-review territory). Before you believe this, have a look at the article. In it we critique, for instance, Richard Dawkins METHINKS*IT*IS*LIKE*A*WEASEL (p. 1055). Question: When Dawkins introduced this example, was he arguing pro-Darwinism? Yes he was. In critiquing his example and arguing that information is not created by unguided evolutionary processes, we are indeed making an argument that supports ID.
My point - and that of many other on these threads - is: Dembski and Marks weren't talking about Dawkins's example. So, whether they were indeed making an argument that supports ID is entirely unsettled (ex falsum quodlibet). Mainly, I'm saying that the math developed by Dembski and Marks isn't applicable to the weasel algorithms as they are commonly understood, and - explicitly - that the program W1, doesn't even fit the description in the paper of a “divide and conquer” procedure.DiEb
September 26, 2009
September
09
Sep
26
26
2009
01:57 AM
1
01
57
AM
PDT
--kf I gave you a link to a commonly accepted definition of "divide and conquer" in the field of algorithms. I showed how this definition is applicable to the algorithm/example of Marks and Dembski in the paper under discussion. I explained that W1 doesn't fit this definition. And so, you just redefine "divide and conquer"... Or is their any source which backs up your version In particular, in this context, the “divide and conquer” approach takes in partitioning in the sense of a catch and keep bin or net if you will.? Could you provide a short hint?DiEb
September 26, 2009
September
09
Sep
26
26
2009
01:28 AM
1
01
28
AM
PDT
PS: Observe, too, that we see a repeated pattern of increasing degrees of tangentiality relative to the primary matter, and that as we move to further and further tangents, personalities are increasingly either explicit or lurk just below the surface discussion. Instead, let's get back to the point: W1 is the probable original Weasel 1986, and it clearly shows targetting, warmer-colder signals to nonfunctional stage by stage phrases and implicit latching, so it -- or whatever actual Weasel 1986 program there is -- is fundamentally dis-analogous to the claimed mechanisms of evolution, and is certainly not a good example of a claimed BLIND watchmaker in action. Even poor Mr Paley had a better point 200+ years ago than is commonly allowed today: a mechanism with self-replication is manifesting a high degree of complex function which needs to be explained in light of empirically credible mechanisms for origin of complex integrated function.kairosfocus
September 26, 2009
September
09
Sep
26
26
2009
01:22 AM
1
01
22
AM
PDT
Onlookers: Let's go right back to the beginning, first of all: that which is primarily and fundamentally "misleading" -- I here cite Mr Dawkins himself -- in the whole weasel discussion is weasel itself. For, weasel (and we now have a credible original program code that underscores it) is utterly dis-analogous to the claimed "blind watchmaker" -- I don't have time here to do more than say that poor Mr Paley has been even more strawmannised: do you know he discussed for instance the implications of discovering a self-replicating watch . . . in Chs 1 - 2 of his key book? That is, he understood and seriously addressed the issue of origin of complex integrated function -- of chance variation and natural selection: ________________ >> It [Weasel c. 1986] . . . begins by choosing a random sequence of 28 letters ... it duplicates it repeatedly, but with a certain chance of random error – 'mutation' – in the copying. The computer examines the mutant nonsense [= non-functional, i.e utterly dis-analogous to organisms that must function to a high degree to reproduce themselves] phrases, the 'progeny' [nonsense phrases are precisely the opposite of "progeny," so the scare quotes are a subtle admission of how misleading the whole exercise is] of the original phrase, and chooses [i.e. programmed, artificial selection, utterly dis-analogous to natural selection] the one which, however slightly, most resembles the target phrase [i.e. there is a built-in target, so weasel does not create complex info out of mere noise but has it inserted from the beginning, as W1 and W2 show], 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 [so, CRD knew in 1986 that the injection of targetted search rewarding non-functional phrases on being slightly "warmer" on distance to target made all the difference, and managed to dismiss the crucial issue of functionality with a rhetorical flourish: "single-step selection]: 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 . . . . Although the monkey/Shakespeare model is useful for explaining the distinction between single-step selection and cumulative selection, it is misleading in important ways. One of these is that, in each generation of selective 'breeding', the mutant 'progeny' phrases were judged according to the criterion of resemblance to a distant ideal target, the phrase METHINKS IT IS LIKE A WEASEL. Life isn't like that. Evolution has no long-term goal. There is no long-distance target, no final perfection to serve as a criterion for selection . . . In real life, the criterion for selection is always short-term, either simple survival or, more generally, reproductive success. [TBW, Ch 3, as cited by Wikipedia, various emphases and colours added.] >> ___________________ You will go back over the past months or the thread above in vain for a serious and frank facing of the implications of the above by the many darwinist objectors here, at Evo info [which is the darwinist site that began the now coming on year-long push to debate on the issue of latching, through one of its representatives, as I have noted here], or elsewhere. So, before we do anything else, we must first see that any attempt on the part of those who refuse to face the above frankly to tag Mr Dembski or others with label-words like "misleading" or "distorting" becomes little more than a turnabout accusation rhetorical stratagem designed to try to drag us down to the level of immoral equivalency where Weasel has been since 1986. Worse, observe that the accusers above again -- sadly, predictably -- fail to address the fact that EIL, sponsored by Marks and Dembski, for many months, has hosted a cluster of algorithms addressing the different aspects of and approaches to the Weasel issue, ranging from an algorithm that shows the challenge to get to the shores of complex function without artificial assistance [comparable to CRD's "single step selection"], to explicitly latched algorithms, to an adjustable proximity reward algor that demonstrated the reality of implicit latching, and onward others that seek to create reward functions that do not reward mere proximity. (And note, EIL makes its code explicitly available; to date we do not have an acknowledged original Weasel 1986 program, though we now have reasonably good reason to infer that W1 is the version that created the showcased runs of 1986 and W2 the BBC Horizon video.) In that context, that on p. 1055 of the IEEE paper, M & D discuss partitioning in terms of what I have called "catch and keep" and happen to use (for purposes discussed above on reading the reversed text of the phrases used illustratively) a case with 100% mutation rate is being used to set up an ad hominem laced strawman. For, it is long since demonstrated that explicit latching does not require 100% mutation rates, and it is also demonstrated that implicit latching -- as has been ever so often described and explained above and in my discussion in App 7 the always linked -- is possible and actually demonstrated here at UD since April 9th. In short, since April 9th, more than enough information has been available to answer the real secondary issue on the merits, namely that an implicit latching mechanism -- again note AmHD [when your quarrel is with a respectable dictionary that should be a warning]: An instrument or a process, physical or mental, by which something is done or comes into being -- is real and would account pretty well for the showcased Weasel runs of 1986. That should have settled the matter on the merits, apart from the tertiary issue of getting credible original Weasel Code. That last point seems to have now been covered, in this thread. (Cf onward discussion, here.) Now, with the above in hand, let us pick up a few illustrative points from commentary overnight: 1] Dieb, 100: M&D talk about a divide-and-conquer procedure. W1 doesn’t fit this description. EIL, of course, hosts a cluster of algorithms that cover the bases. And, in describing partitioned search the key point the make is that once a letter in a generational champion goes correct, the search for it is over. That can be achieved explicitly or implicitly, under a fairly wide array of circumstances. Circumstances that EIL cover in the various algors at the already linked GUI page. We must bear this in mind when we consider the argumets and points made below. In particular, in this context, the "divide and conquer" approach takes in partitioning in the sense of a catch and keep bin or net if you will. 2] Rob, 101: regardless of how broadly one defines “partitioned”, the fact remains that Dembski has been mischaracterizing Dawkins’ WEASEL algorithm for a long time, even after being corrected. But in fact, partitoning means what it means, and ratcheting means what it means -- and latching though an antireverse pawl or dog is a part of that -- and cumulative progress to target means what it means. All as long since repeatedly and carefully discussed in more than adequate details. The EIL cluster of algors, which antedates the actual publication of the IEEE paper by several months, shows a considerable range of algors that give explicitly and implicitly latched, ratcheting behaviour, quasi-latched [ratchets that sometimes slip] behaviour, and clearly non-latched behaviour. In that context the above, sadly, becomes little more than an ad hominem laced strawman used in a turnabout attack. 3] The EIL website presents the same math as the IEEE paper, specifically saying that it refers to the “partitioned search used by Dr. Dawkins”. As discussed repeatedly, it’s impossible to reconcile that math with the description and output of the algorithm in TBW, as well as the Pascal code that Dembski now considers to be Dawkins’. The reference --note it is not given in context [and that the highly relevant further context of the cluster of algors at the EIL GUI is not mentioned] -- is:
First, let's look at partitioned search used by Dr. Dawkins. Assuming uniformity, the probability of successfully identifying a specified letter with sample replacement at least once in Q queries is . . .
The first statement is clearly an empirically based inference that the showcased Weasel runs of 1986 show partitioning, in the sense that once a letter in a generational champion in these runs goes correct, it does not revert. This holds true without exception for a sample of 200+ letters that potentially could revert, from 300+ letters total. That is, it is a dominant and striking behaviour pattern of the run o/ps. Next, EIL proceeds to a particular model: "Assuming uniform probability . . . " What follows is indeed the same model given in the IEEE paper. However, we may also see that -- through links on the very same page that take you to live versions of the GUI -- so soon as the EIL proceeds to computer modelling, we see that they present a cluster of various algorithms (and models as a direct consequence), inviting participative comparison and exploration of various flavours of weasel ware. So, while I can sympathise with desiring editing of the EIL math discussion page or the IEEE paper, I cannot agree that M & D are in any reasonable sense of the term, willfully misleading. 4] This is how Dembski described it at the “Nature of Nature” conference . . . . This matches the IEEE paper and the EIL website, but not TBW The excerpt (and, remember the context above):
But consider next Dawkins’ reframing of the [origin of complex, functionally specific bio-information] problem. In place of pure chance [and until you attain to complex, specific info based function, a BLIND watchmaker cannot bring to bear differential success at function to climb the hill to optimal performance so this is a fundamental dis-analogy to the generally proposed evolutionary mechanisms . . . i.e. the primary issue; note how this is again not addressed by Darwinist objectors], he considers the following evolutionary algorithm: (1) Start with a randomly selected sequence of 28 capital Roman letters and spaces (that’s the length of METHINKS IT IS LIKE A WEASEL); (2) randomly alter all the letters and spaces in the current sequence [poorly phrased, and the point where D does invite a rhetorical wedge; but the point from the contextual reference and showcased runs is plainly that letters are subject to random alteration, and applicable mutation rates from the EIL models and elsewhere can plainly vary] that do not agree with the target sequence; [that is, in context of what was to be explained -- the showcased 1986 runs -- once a letter in a generational champion goes correct it is credibly preserved thereafter; and there is in the context of the showcased runs plainly no irrevocable commitment to impose a 100% mut rate per incorrect letter, though he does mathematically model such a case -- so again cf the cluster of EIL algors] (3) whenever an alteration happens to match a corresponding letter in the target sequence, leave it and randomly alter only those remaining letters that still differ from the target sequence. [that is, partition the search so that when a letter becomes correct in a generational champion, it is preserved in succeeding champions until the target is hit. ]
When I compare the above with the showcased runs of Weasel 1986, in the contexts as noted in parenthesies, I find that Mr Dembski's overall description of the behaviour of the Weasel 1986 algortihm is generally correct, noting that Weasel 1986, as described and showcased in BW, does not necessarily show 100% mutation per incorrect letter in each child in each generation. [Note again, the significance of the cluster of algors at EIL on this.] Moreover, the W1 program above, the best candidate to be the 1986 algorithm, ratchets through an implicit latching mechanism, as has also been described repeatedly in this thread and similar to the explanation and examples over the past months. Observe also the rhetorical pattern of shifting from one topic to the next, ending up in personalities: [1] Weasel 1986's problem of targetted search with built in target, to [2] the long debate over latching and incidental issues, to now [3] Dembski the claimed mischaracteriser. Rob, a note to you: I would take your points over Mr Dembski and his phrasing or 100% mut rate model far more seriously if I saw a balancing serious and frank facing of the primary issues with Weasel. But, over many months now, I do not. Think about the message that is sending me and others in the context of the all too commonly seen darwinist debater tactic of distraction, distortion, demonisation and dismissal. 5] CIT, 102: If retention of any letter is simply a result of the initial conditions, then how can the program itself be considered latching? This conflates two distinct issues: ratcheting-latching action, and using explicit mechanisms to do this. A program, generally speaking, takes in inputs, processes them per algorithms, data structures and coding, and produces outputs. The main mechanism of the program thus lies in its processing. And, where that processing is in material part driven by the values of key parameters, then setting such parameters is necessarily a major part of the mechanism: how 'tweredun. In the case of weasels that show latching-ratcheting action, it has been shown that his can happen two distinct ways: explicit setting up of a masking mechanism that blocks reversion of correct letters, and implicit interactions that through the presence of no-change cases [or the like] in mutant populations provide a backstop that blocks reversion when the filter is applied to choose a new gnerational champion. The showcased 1986 runs show that in fact the no-change case [or the comparable side-step case where an incorrect letter shifts to a different incorrect value . . . ] won the generational championship race roughly half the time. In short, this is a major behaviour of the program as implemented c. 1986. So, "simply" clearly misses the key point. GEM of TKIkairosfocus
September 26, 2009
September
09
Sep
26
26
2009
01:00 AM
1
01
00
AM
PDT
kairosfocus, #99
Above, we should not lose focus on the actual matter to be explained: what is actually to be explained on the secondary issue of latching-ratceting is that for generation champions, once a correct letter has been identified — due to the action of the explicit or implicit latching mechanism as a component of the ratcheting action — the search for it is effectively over. On the case of particular interest, implicit latching and ratcheting, once the population dynamics on the seed are such that there is a high enough probability of no-change cases being present in the child population for a generation [and the odds of double mutations etc are low also], the filter will reject anything that does not at least preserve advances to date.
If retention of any letter is simply a result of the initial conditions, then how can the program itself be considered latching?camanintx
September 25, 2009
September
09
Sep
25
25
2009
05:24 PM
5
05
24
PM
PDT
kairosfocus, regardless of how broadly one defines "partitioned", the fact remains that Dembski has been mischaracterizing Dawkins' WEASEL algorithm for a long time, even after being corrected. This fact can't be denied simply by interpreting the IEEE paper in certain way. Consider: - The EIL website presents the same math as the IEEE paper, specifically saying that it refers to the "partitioned search used by Dr. Dawkins". As discussed repeatedly, it's impossible to reconcile that math with the description and output of the algorithm in TBW, as well as the Pascal code that Dembski now considers to be Dawkins'. - This is how Dembski described it at the "Nature of Nature" conference:
But consider next Dawkins' reframing of the problem. In place of pure chance, he considers the following evolutionary algorithm: (1) Start with a randomly selected sequence of 28 capital Roman letters and spaces (that's the length of METHINKS IT IS LIKE A WEASEL); (2) randomly alter all the letters and spaces in the current sequence that do not agree with the target sequence; (3) whenever an alteration happens to match a corresponding letter in the target sequence, leave it and randomly alter only those remaining letters that still differ from the target sequence.
This matches the IEEE paper and the EIL website, but not TBW. - The same description is repeated almost verbatim in his book No Free Lunch.R0b
September 25, 2009
September
09
Sep
25
25
2009
03:43 PM
3
03
43
PM
PDT
--kf have a look at #86 and #90: M&D talk about a divide-and-conquer procedure. W1 doesn't fit this description.DiEb
September 25, 2009
September
09
Sep
25
25
2009
02:43 PM
2
02
43
PM
PDT
Onlookers: Above, we should not lose focus on the actual matter to be explained: what is actually to be explained on the secondary issue of latching-ratceting is that for generation champions, once a correct letter has been identified -- due to the action of the explicit or implicit latching mechanism as a component of the ratcheting action -- the search for it is effectively over. On the case of particular interest, implicit latching and ratcheting, once the population dynamics on the seed are such that there is a high enough probability of no-change cases being present in the child population for a generation [and the odds of double mutations etc are low also], the filter will reject anything that does not at least preserve advances to date. Indeed, in the showcased 40+ and 60+ gen runs of 1986, no advance evidently won the generation champion race about half the time. Plainly, this is a significant, though somewhat subtle effect. (If we started with no correct letters and stepped just one letter closer per generation, we would hit target in 28 steps.) So, while individual members of the population can vary all over the place, the presence of the no-change cases acting with the dominance otherwise of single step advances, will create implicit latching and ratcheting to target. All of this corrective effort on a side issue is probably best justified by highlighting how it teaches us to think dynamically and to recognise the reality of probabilistic barriers as a real-world mechanism. Which is not without general significance for the wider design issue. So, let us study how the proponents of the different points of view handle the question of mechanisms, where mechanisms are important and may not necessarily be obvious or explicit. GEM of TKIkairosfocus
September 25, 2009
September
09
Sep
25
25
2009
02:22 PM
2
02
22
PM
PDT
Excellent Joseph, we are finally getting somewhere. Let X denote the statement "once a letter is found, the search for that letter is over". With your daughter's help, we have determined that X may be false for TBW Weasel. Can X be false for the search described on p1055? Only after you have answered that easy question, we may explore the issue of X appearing to be true, when in actuality it is false...DNA_Jock
September 25, 2009
September
09
Sep
25
25
2009
08:30 AM
8
08
30
AM
PDT
PS: Similarly, on partitioning. To partion is not a mystery in language, or in context: once letters in the showcased 1986 Weasel runs go correct, they stay that way; and that can happen explicitly or implicitly, as has been demonstrated. Further to this, EIL has put up a cluster of Weasel algors [and that is a highly relevant fact], showing the two patterns, so there is no one "the" weasel algorithm a la M & D that can be strawmannised into a cartoonish contrast to Dawkins' work. Resorting to the rhetorical device of speaking as though the facts are not so, is revealing that, again, the real facts are not in your favour.kairosfocus
September 25, 2009
September
09
Sep
25
25
2009
08:29 AM
8
08
29
AM
PDT
DNA-J: I have cited, for weeks now, the reason why I have used "mechanism" as I have -- and this is a fairly standard and entirely appropriate usage. I again did so just now, as a simple scroll-up will show. You may choose to act as though those unwelcome- to- you facts are not do, but that evidently willful refusal is quite revealing that the balance on the merits is not in your favour. Please, think again. GEM of TKIkairosfocus
September 25, 2009
September
09
Sep
25
25
2009
08:23 AM
8
08
23
AM
PDT
DNA Jock, The answer is "most likely"- that is according to my daughter's magic 8 ball.Joseph
September 25, 2009
September
09
Sep
25
25
2009
08:11 AM
8
08
11
AM
PDT
Joseph So, in TBW Weasel, the search for a found letter is "for all intents and purposes" over.
If you omit this qualifier, is your statement still true?
Please answer Alice's questionDNA_Jock
September 25, 2009
September
09
Sep
25
25
2009
07:52 AM
7
07
52
AM
PDT
kf - Still with the "mechanism is a synonym for behavior", H.D.! (see kf's 233 on prev thread) Talk about recycling cogently answered objections, indeed. You completely failed to respond to my point re the meaning of "partitioned search". And no, 'partitioned' is not a synonym for 'latched', H.D. A search algorithm may be latched but not partitioned (a conversation we have had previously, you will recall), or it may be partitioned but not latched (useful if the target is moving). p1055 describes a partitioned search. TBW weasel is not partitioned. The PASCAL under discussion will show impressive latching behavior, almost every time, but it is not a partitioned search.DNA_Jock
September 25, 2009
September
09
Sep
25
25
2009
07:47 AM
7
07
47
AM
PDT
DNA Jock:
I am troubled by the word “essentially”.
And you want me to comfort you? OK then- DNA Jock, do not be troubled by the word "essentially". It cannot do you any harm and its meaning can be easily ascertained. But anyway in context "essentially" means "for all intents and purposes".Joseph
September 25, 2009
September
09
Sep
25
25
2009
07:20 AM
7
07
20
AM
PDT
DNA-J: You know or should know that we have been discussing the 1986 showcased runs, and a mechanism that when pop sizes, mut rate per letter and filter are mutually adapted, will produce a good enough fraction of runs that will show the pattern as showcased in 1986. For isntance, int eh as printged version of W1 abovge, runs will latch implicitly 199 out of 200 times. tha tis good enough to account for what was needed. GEM of TKI PS: onlookers, I find it interesting that back in March-April, the attack was that the runs as showcased did not show good reason to believe they latched. Notice how that talking point has now disappeared. Now that we have credible code for Weasel 1986 [and the video runs of 1987] it should be clear that the inference to latching was well supported, and that the implicit latching is real enough.kairosfocus
September 25, 2009
September
09
Sep
25
25
2009
06:59 AM
6
06
59
AM
PDT
--kf D&M write
Partitioned search is a divide and conquer procedure best introduced by example.
They don't write
Partitioned search is a latching procedure best introduced by example.
You may call W1 implicitly latching, but you can't call it a divide and conquer procedure.DiEb
September 25, 2009
September
09
Sep
25
25
2009
06:50 AM
6
06
50
AM
PDT
Joseph wrote:
The way Dawkins describes cumulative selection and illustrates it with “weasel” it is a divide and conquer stategy- that is once a correct letter is found/ matched the search for it is essentially over.
I am troubled by the word "essentially". Does it mean "if the target string is sufficiently short and the generation size is sufficiently large and the mutation rate is not too high." If you omit this qualifier, is your statement still true? AliceDNA_Jock
September 25, 2009
September
09
Sep
25
25
2009
06:49 AM
6
06
49
AM
PDT
PS: I think I need to underscore: the latching in view and ratcheting in view is that of generational champions. So, it is that behaviour that needs to be explained, including the description that the effect of the process is to "catch and keep" on-target letters as they are successively guessed; creating cumulative progress to target.kairosfocus
September 25, 2009
September
09
Sep
25
25
2009
06:36 AM
6
06
36
AM
PDT
DNA-J: First, latching behaviour is produced by a latching mechanism, which may be explicit or implicit, as has already been pointed out to you. And, a mechanism -- again, onlookers! -- is: An instrument or a process, physical or mental, by which something is done or comes into being. [AmHD] If you took time to actually follow up on the roots of the matter, you would soon see that the heated and prolonged exchanges over latching, etc over several months are secondary to the primary issue, and actually have been distractive -- as in red herrings and ad hominem soaked strawmen -- as posed by objectors to the primary point: Weasel, by CRD's confession, is fundamentally dis-analogous to any reasonable mechanism for chance variation and natural selection. Generational champions are the focus of attention for the discussion for a very simple reason: it is these that were showcased in the CRD runs of 1986. So, it is their evidently latched behaviour that needs to be explained. And, explicit and implicit mechanisms are both able to do that. We now know on preponderance of evidence, that the likely original Weasel -- W1 above -- exhibits implicit latching. And since old, long since cogently answered objections are now being recycled, it seems that the matter is pretty well settled on the merits. Just, the answer on the merits does not sit well with champions of the blind watchmaker thesis. GEM of TKIkairosfocus
September 25, 2009
September
09
Sep
25
25
2009
06:32 AM
6
06
32
AM
PDT
For instance: A divide and conquer algorithm works by recursively breaking down a problem into two or more sub-problems of the same (or related) type, until these become simple enough to be solved directly. I'll use the example of D&M to identify the sub-problems: 1. Find the correct string out of the 1.19 * 10^40 strings matching ............................ 2. sub-problem 1: Find the correct string, using the 1.64 * 10^37 strings matching …………………..E.S.. sub-problem 2: forget all strings not matching …………………..E.S.. 3. sub-problem 1: Find the correct string, using 1.14 * 10^30 strings matching ..T……….S….E._..E.S.L sub-problem 2: forget all strings not matching ..T……….S….E._..E.S.L At each step, the space of feasible solutions is reduced considerably: we're in fact looking for a sub-string on a smaller space. The complement of this smaller space is discarded. So, now try to apply this definition to weasel1: What are the sub-problems? Each sub-problem should correspond to a sub-space of strings. Which are the sub-spaces?DiEb
September 25, 2009
September
09
Sep
25
25
2009
05:45 AM
5
05
45
AM
PDT
DNA Jock, The "weasel" program doesn't need to contain explicit programming for latcching. It does it as a matter of course. The way Dawkins describes cumulative selection and illustrates it with "weasel" it is a divide and conquer stategy- that is once a correct letter is found/ matched the search for it is essentially over.Joseph
September 25, 2009
September
09
Sep
25
25
2009
05:44 AM
5
05
44
AM
PDT
kairosfocus My irony meter broke when you started complaining about the effort expended to dismiss the concept of implicit latching. Latching, smatching. Here's an idea: you stipulate that TBW Weasel does not contain an explicit latching mechanism, and we'll all agree that it can, given certain parameters, show latching behavior. Then we can all move on to discussing the 'distant target' issue. In the meantime, your defense of D&M's use of TBW as a citation for a "Partitioned Search" goes ever further off the rails. You are now claiming that the word 'partitioned' somehow refers to the separation of letters in generational champions into those that are an improvement vs those that are not.
Partitioning — AmHD: “The act or process of dividing something into parts . . . To divide into parts, pieces, or sections” –in effect finds a way to put the letters of generational champions — what was to be explained, per the showcased examples of 1986! — into two bins: the ones already on target [that will not slip back] and the ones not yet on target [which will vary until they can be put into the on-target bin]. [emphasis in original]
But TBW Weasel retains phrases, not letters. And this is not what "partitioning" means. D&M describe, quite unambiguously (albeit with a sophomoric example), a 'divide and conquer' search, and give the math (eqn22) for a 'divide and conquer' search without generational champions, where the search for each element is independent of the search for the other elements. And they call this search a "partitioned search" in keeping with the accepted use of this term by the rest of the world. Obviously, TBW Weasel is not a 'divide and conquer' search, so you, in your Quixotic defense of the uncharacteristically silent authors, are reduced to Humpty Dumpty land.
8 –> It is in this context that I have objected to trying to turn partitioning to a synonym for the sort of algor that may be interpreted from the didactic example and associated calculation on p 1055 of the IEEE paper.
There's glory for you" H. DumptyDNA_Jock
September 25, 2009
September
09
Sep
25
25
2009
05:18 AM
5
05
18
AM
PDT
DiEB:
“divide and conquer” implies that at each step we can exclude a set of strings from being possible solutions:
What do you base that on? Or is it just another strawman? But anyway as I said if you Read TBW the way Dawkins describes cumulative selection and illustrates it with "Weasel" it is a partitioned search- that is once you find what you are looking for the search for it is over.Joseph
September 25, 2009
September
09
Sep
25
25
2009
04:39 AM
4
04
39
AM
PDT
OOPS: 100% mutation rates, not 10%.kairosfocus
September 25, 2009
September
09
Sep
25
25
2009
01:19 AM
1
01
19
AM
PDT
1 2 3 4 5

Leave a Reply