Although I avoid opening any ID sites, from time to time I receive emails quoting certain comments posted to such sites. The other day I received one such email from Alan Fox. In his email Alan quoted three comments that appeared on one of Dembski’s sites and relate to my essay (critical of Dembski), which was printed in the Skeptic magazine, v. 11, No 4, 2005. (Its full text is available online - see here.) So I unwillingly found myself looking at three hostile comments regarding certain points in my Skeptic essay. I shall quote here two of these comments as they appeared in Alan’s email.
One of the hostile comments was by Salvador Cordova. In his frequent comments on Panda’s Thumb (PT) Salvador tries (not fully successfully) to restrain his apparent penchant for exaggerating his qualifications and denigrating the objects of his assaults. In his comment on Dembski’s site - where he is protected by the absence of counter-arguments - he indulges in wild attacks on Dembski’s critics, including me. His comment is full of repeated claims that I “mangled,” “fumbled”, and “misrepresented” Dembski’s great ideas and attacked straw men.
Assertions that his critics simply “do not understand” his concepts has been a device often used by Dembski – see for example his “replies” to the critique by Ellery Eels, Robert Pennock, Richard Wein, Erik Tellgren, Eli Chiprout, Wesley Elsberry, Jeffrey Shallit, and others. I seem to be in good company. Can it be that such a regularly employed accusation rather reflects Dembski’s (and even more so of Cordova’s) inability to offer more substantive counter-arguments?
Perhaps Salvador is sincere in his desperate attempts to find errors in the critique of Dembski. It is interesting, though, that, while Dembski lets Salvador jump high in the “defense” of his “Lord William,” (which is how Cordova referred to Dembski on PT) Dembski himself has so far never explicitly endorsed Salvador’s rants. Perhaps Dembski realizes the abysmal level of Salvador’s contentions and avoids being directly associated with them.
Salvador’s comment essentially repeats his earlier assertions on PT, which have been answered extensively in many other posts on PT. Therefore I will not take space here for one more demonstration of Salvador’s fallacies; they have already taken too much space on PT. Perhaps one brief comment may be in order. Replying to my earlier comment on PT, where I wrote that I’d not curtail Salvador’s freedom to post anything he wants in my threads, Salvador wrote that he respected me for that. If so, then, to be consistent, should he not disrespect Dembski, who deletes from his sites any comments he dislikes?
I’ll briefly discuss now the other two comments copied from Dembski’s site by Alan. Both relate to just one point in my Skeptic essay, namely one example of a false positive produced by Dembski’s explanatory filter (EF). This example refers to a rare form of snowflakes, which appears under certain weather conditions. Since the weather conditions in question are very rare, the appearance of such snowflakes has a low probability. These snowflakes also have a specific, easily recognizable form that is the simplest kind of snowflakes ever observed. Since in this case we have a combination of low probability with specification, the inference prescribed by Dembski’s EF is that the snowflakes in question are results of design – it is just one more case of a false positive.
Furthermore, according to Dembski, low probability is just another face of complexity: the more complex the object, asserts Dembski, the lower its probability (in my Skeptic essay there are a number of direct pertinent quotations from Dembski). In fact, however, the rare snowflakes in question have the simplest structure of all known snowflakes (see the relevant references in my Skeptic essay). This exemplifies the fallacy of Dembski’s thesis, which equates complexity with low probability.
Both hostile commenters hide their names, one using a pseudonym (“taciturnus”) and the other just a first name (“dave”). What are these critics of my essay afraid of? Do they know that their arguments are false? Or are they just not sure their comments make sense? Or do they hide their names so they can hurl insults with impunity?
Let us see if their specific critical remarks regarding the rare snowflakes have any merits.
To avoid accusations of distorting what my opponents say (if it is at all possible, given the predilection of some of Dembski’s supporters for slandering his opponents), I’ll reproduce here the full texts of the hostile comments as quoted by Alan.
Here is the first of these comments:
After addressing Alan, the comment continues as follows:
1. I’ve read your link to Mark Perakh (Dream_Dem), and I now see what the ID defenders mean when they imply that Mr. Perakh seems to go out of his way to misunderstand Intelligent Design. Consider some of his remarks about specified complexity:
I believe that the very concept of complexity as disguised improbability is contrary to facts and logic. For example, under certain (rare) weather conditions, an unusual triangular shape of snowflakes can be observed.26 Unlike more common forms of snowflakes with their intricately complex structure, these rare snowflakes have a simple structure. As Dembski asserted,(27 snow crystals’ shapes are due to necessity—the laws of physics predetermine their appearance. However, triangular snowflakes, while indeed predetermined by laws of physics, occur only under certain weather conditions, which are very rare and unpredictable. Therefore we have to conclude that the emergence of the triangular snowflakes is a random event. This is another example where at least two causal antecedents—chance and law—are in play simultaneously.
Since the appropriate weather conditions occur very rarely, the probability of the chance emergence of the triangular snowflakes is very small; also, they have a uniquely specific shape. Hence, according to the EF, these snowflakes were deliberately designed.
But complexity as improbability is obviously meant as conditional improbability. Given conditions A, the probability that B will occur is so low that we can infer design. Given whatever unusual whether conditions you prefer, the probability that wind and rain will carve the faces of Presidents on Mt. Rushmore is tiny. We can infer design. However, given the right weather conditions, the probability that triangular snowflakes will occur is high. We cannot infer design, especially since the only time we see these snowflakes is during the unusual weather conditions that make them highly probable. This does not seem a difficult point.
Comment by taciturnus — September 9, 2005 @ 7:31 am
To start with, the example of Mt. Rushmore is irrelevant. The Rushmore pattern has a human origin and in such cases design inference is a well established procedure based on our familiarity with human design and its results. This procedure has nothing to do with Dembski’s EF (which anyway is, in my view, as evinced in my Skeptic essay, a meaningless schema). In the case of snowflakes no background knowledge of the kind we have with a human design is available. This point has been thoroughly discussed in literature (see, for example, the collection Why Intelligent Design Fails, edited by Young and Edis, now in its third printing with a paperback edition forthcoming, where this point has been discussed in detail).
Look now at taciturnus’s argument which asserts that “complexity as improbability is obviously meant as conditional improbability.” Unfortunately for taciturnus, it is not only not “obvious” that Dembski’s schema indeed implies conditional probability, but in fact this schema does nothing of the sort, either obviously or implicitly. Taciturnus seems to mix up two different questions. One question is whether or not the snowflake in question was designed? The other question is what inference follows for Dembski’s schema? If we were searching for the answer to the first question, taciturnus’s notion would be reasonable: it is indeed obvious that in the case in point there is no reason to infer design; the appearance of the triangular snowflakes is predetermined by the combination of proper weather conditions and laws of physics. This correct inference is, though, done outside Dembski’s EF. The answer to the second question is that EF requires inferring design, which is a false positive. Indeed, nowhere does Dembski’s schema imply the use of conditional probability.
If we turn to Dembski’s actual writing, we find that he pays a lot of lip service to evaluating multiple “relevant chance hypotheses,” although he never himself bothers to go beyond evaluating a single chance hypothesis that uses the uniform distribution. Dembski’s schema prescribes evaluation of probability, period. In the case of snowflakes, the overall probability comprises two components, one random, and the other non-random. The random component is the (low) probability of proper weather condition. The non-random component is the (high) probability of “laws of physics producing the snowflakes in question under the proper weather condition.” Obviously the random component precedes the non-random one in the causal chain. “Taciturnus” suggest to ignore the random component and to base the inference only on the non-random one. Such an approach would be contrary to Dembski’s schema.
Indeed, why should we base our application of EF on the conditional probability of the appearance of this type of snowflakes under given weather condition (which is high) when it is obvious that in the causal chain the probability of the proper weather precedes the probability of “physical laws producing such snowflakes”? Following Dembski’s schema, we cannot ignore the probability that is “upstream” in the causal chain, as taciturnus suggests doing. The small value of the probability that is “upstream” overrides the larger probability that is “downstream.”
Taciturnus’s correct judgment (that snowflakes in question are not designed) is based on common sense and available background knowledge, but the question is not about that. It is whether or not Dembski’s approach yields the correct conclusion. It does not, in part because it does not prescribe using conditional probability – its use is just taciturnus’s common sense suggestion rather than a feature of Dembski’s thesis.
From another angle, the probability of the rare snowflakes being conditional on weather, again, does not negate the fact that these snowflakes have a low overall probability. Therefore Dembski’s formal thesis, if applied consistently, requires the snowflakes to be complex. But they are simple. Taciturnus’s argument can in fact be used to argue against EF and against Dembski’s thesis of “complexity being equivalent to low probability.”
Unfortunately for “taciturnus” his (her) argument fails to properly address the question at hand – the validity of my example of the rare snowflakes.
Here is the second hostile comment, as quoted by Alan:
2.The snowflake example also fails because the triangular design isn’t specified beforehand. This is just another version of the arrow and the barn example. All points on the barn are equally unlikely to be hit. A particular point on the barn is only interesting if it has been specified before the event — for instance by a bullseye. (sic).
The triangular snowflake is no more interesting than a four-leaf clover, ball lightning, or the aurora borealis. All are rare, complex natural events, but none of them are specified before the event. Their patterns are reducible to being a function of the natural conditions that produced them, rare or otherwise. All of them are surprising and remarkable, but from them no reasonable person could ever infer design.
The idea of specificity is so fundamental to design inferences, it’s astonishing that Perakh considers this example applicable. Bill has asked if Perakh understands the relevant math. After reading this, I’m wondering if Perakh understands the relevant English.
If Bill or any other ID proponent had to correct every published essay that exhibited a basic misunderstanding of the argument, they’d spend all their time chasing down op-eds and blog blather.
Comment by dave — September 9, 2005 @ 1:33 pm
I will not respond to dave’s remarks about my misunderstanding “relevant English,” which parrots Dembski’s earlier infamous utterance – such derogatory remarks usually are offered when no arguments of substance are available. Let us instead look at his argument regarding the snowflake’s shape not being specified “beforehand,” which, according to “dave,” shows my lack of understanding of the concept of specification.
Before discussing dave’s specific notions, it is perhaps proper to point out that Dembski’s concept of specification has been severely critiqued by various reviewers. I have made some modest (although rather detailed) contribution to the discussion of Dembski’s specification in my book Unintelligent Design (Prometheus Books, 2004, pp. 47-53). In my Skeptic essay I also have analyzed that concept but dave chose not to notice that analysis. In an excellent article (which is available online - see here) Elsberry and Shallit made mincemeat of Dembski’s specification concept. (As could be expected, apparently incapable of providing a cogent response to Elsberry & Shallit’s article, Dembski’s camp responded with hysterical assaults like those by Salvador Cordova, who posted a number of meaningless pieces of “critique” baselessly accusing Elsberry and Shallit of [of course!] “misrepresenting” Dembski’s specification concept.)
Regarding dave’s specific argument (that specification must be made “beforehand”), dave may be surprised to learn that it is contrary to Dembski’s thesis. Dembski unequivocally asserts (see Dembski’s The Design Inference, page 14) that the pattern meeting the requirement of specification can be legitimately identified after the fact. Dembski’s criterion for distinguishing between “specification” and “fabrication” is not when the pattern was identified, but whether or not it meets what Dembski calls “detachability.” This term, explains Dembski, means that the pattern is “independent of an event.”
(It can be noted that if the requirement for the specification to be determined “beforehand” were adopted, it would make the entire “design inference” a la Dembski not applicable to biology. We never know “beforehand” which pattern will have, say, a hitherto unobserved chunk of DNA, or, say, how a hitherto unknown species of bacteria will look like. That is why Dembski prescribes testing for “detachability” rather than for “when the specification is made”.)
If dave’s comment, as he formulated it, were correct, it would first apply to Dembski himself.
Recall Dembski’s example illustrating his concept of specification. (See, for example, again Dembski’s The Design Inference, where the “detachability” is discussed in many words). A pattern may serve as a specification, says Dembski, only if it is “detachable.” Let us see if the rare snowflakes meet this condition.
In Dembski’s own example, he talks about a heap of stones which happens to reproduce the shape of a constellation (this example is on page 17 of Dembski’s The Design Inference). When a layman sees these stones he does not recognize the shape of a constellation so the observed shape is not “detachable” and does not serve as a specification. If, though, an astronomer sees the same heap of stones, he recognizes the image of a constellation (which he has previously stored in his mind independently of the particular heap of stones he came across) and in this case the observed pattern is “detachable” and serves as specification.
The astronomer infers that some intelligent agent has, by design, arranged the stones in the shape of a constellation. He came to such a conclusion because the shape of that constellation was antecedently familiar to him. The astronomer did not expect “beforehand” to find these particular stones arranged as this specific constellation. However, the pattern he observed was “detachable” as it matched an image he had, antecedently and independently from this particular heap of stones, stored in his mind. Recall that all this is Dembski’s own example illustrating his concepts of “detachability” and “specification.” This is the essence of the notion that specification is predicated on prior knowledge of the pattern – which is a point rather different from that made by dave. Dave avoided mentioning “detachability,” which would be a proper reference to Dembski’s thesis.
Exactly the same argument applies to the snowflakes in question. For dave the shape of the rare snowflakes is not familiar and therefore not “detachable.” Hence, for dave these snowflakes are not “specified.” However, to an expert on snowflakes the shape is known, so when such an expert sees those rare snowflakes, he recognizes them as conforming to the image he has antecedently kept in his mind. The pattern is, in this case, according to Dembski’s thesis, “detachable,” exactly as the pattern of the heap of stones in Dembski’s own example. Dembski’s “theory” requires inferring design equally in the case of stones and in the case of snowflakes. This inference may be true for the heap of stones but is false for the triangular snowflakes, and this shows the inadequacy of Dembski’s thesis.
When dave correctly concludes that the rare snowflake is not a product of design, he (like taciturnus) does so outside of Dembski’s EF, and in fact his conclusion is contrary to what EF yields. EF yields a false positive.
Dave’s comment shows his own misunderstanding of the subject he decided to argue about.
The fact that neither hostile commenter identified in my essay any more targets for their (fallacious) critique, besides the sole example of the rare snowflakes, is telltale. It points to their apparently being at a loss when confronted with the entirety of my arguments. Perhaps this is also the reason that, absent any more visible targets for assaults, both commenters resorted to general assertions regarding my “misunderstanding” of ID and of Dembski’s work.
If the comments by “taciturnus” and “dave” plus the rants of Cordova are the best the ID advocates can offer in response to my essay in the Skeptic, their case has to be relegated to the dustbin of history, to borrow Dembski’s favorite pompous expression.
I believe unbiased readers can themselves now infer who in this debate indeed poorly understands Dembski’s thesis, “relevant English,” and the super-sophisticated collections of math symbols so loved by Dembski but evidently beyond the comprehension of some of his supporters.
I thank Wesley Elsberry and Matt Young for taking time to read the initial draft of this piece and suggesting pithy comments.
141 Comments
bill · 14 September 2005
I believe the snowflakes are composed of frozen Waterloo.
Hyperion · 14 September 2005
Joseph O'Donnell · 14 September 2005
That isn't far away from what Behe said, after all he'll know design when he sees it. This concept is utterly worthless scientifically but that just reflects the ID movement in general doesn't it?
Cyrian · 14 September 2005
Re: Hyperion
I think that it's important to respond to these types of comments sometimes because it gives the rest of us (who don't have hundreds of published papers under our belts) information to use when discussing this with other people in daily life. Some of the ID arguments are so entangled that explaining to the average person why they're crap is very difficult. And posts like these make it much easier to do so.
Norman Doering · 14 September 2005
I like the concept of "detachability." A lot of patterns are in a sense "independent of an event." There's got to be a better way to express that though. The whole process of pattern recognition seems to involve detaching and abstracting patterns. We do make mistakes (remember the face on Mars? Seen a face in a cloud? And faces are more complex than constellations. Would seeing a face in the clouds represent a false positive to Dembski?)
I don't see how it applies to evolution though. What's detachable about DNA or life? If I saw a string of DNA I wouldn't think it looked like something else or know what it was for and a geneticist would only know what it does after the fact -- it doesn't really detach and match up with an "independent of an event" item. There's no DNA constellation or face to match the pattern against.
I don't recall Dembski ever explaining that.
Any Dembski defenders here -- I'll give you a fair shake at explaining that.
shiva · 14 September 2005
Mark,
Your experience in the sciences and wide world dwarfs the entire lifetimes of many contributors to this forum. Bill D can accuse you of many things (that's the only thing he is good at) but can never accuse you of not taking him seriously. Even after knowing that BillD deliberately misquotes, cooks up fake ideas, and smears people, you have chosen to spend time not only on his books but also on the smears and slime cast about by his factotums on anyone who shows up ID to be the crackpot idea it is. Only a true teacher would do that. I wonder if you still see in BillD a good student who somehow refuses to turn out what he is truly capable of.
germline · 14 September 2005
I know why Mark keeps on replying. It's so damn infuriating to wrestle with these slippery bastards, but it's bloody addictive ! I wasted the greater part of several days arguing with some guy named DaveScot who just started shouting at me and then I just got kicked off.
But I have to agree with Norman here, there is something interesting about the idea of detachability. Maybe it just boils down to cognitive science and the human mind. Either way, when someone really does come up with a way to work this question, there will be something to it.
How awesome would it be if someone blew the doors off this issue, and it got published in Nature and "design theory" really took of in a precise way within the field of cognitive science, but completely independent of creationism !
Norman Doering · 14 September 2005
germline wrote: "I have to agree with Norman here,..."
As should everyone.
"... there is something interesting about the idea of detachability. Maybe it just boils down to cognitive science and the human mind."
I'm not quite sure I can say this and be 100% correct yet, but it seems to me at the moment that everything a neural net does can be boiled down to the term "pattern recognition." And then all pattern recognition would involve some form of detachment and abstraction.
Another clue to our brains pattern recognition processes would be "apophenia."
http://skepdic.com/apophenia.html
According to Brugger, "The propensity to see connections between seemingly unrelated objects or ideas most closely links psychosis to creativity ... apophenia and creativity may even be seen as two sides of the same coin."
"How awesome would it be if someone blew the doors off this issue, and it got published in Nature and "design theory" really took off in a precise way within the field of cognitive science, but completely independent of creationism."
At the moment that looks like a real possibility to me. There really is an "illusion of design" in nature, but it probably says more about our perceptions than about any design in nature.
Hyperion · 14 September 2005
germline · 14 September 2005
Hey Norman,
Here is where all the real action is.
http://tinyurl.com/ajc8l
Too bad Dembski is about to miss the boat on it.
Norman Doering · 15 September 2005
germline wrote: "Here is where all the real action is.
http://tinyurl.com/ajc8l
Too bad Dembski is about to miss the boat on it."
Thanks, "Perception As Bayesian Inference" and pattern theory, some of those basic ideas were popping into my head, just not so clearly stated.
Jaime Headden · 15 September 2005
The snowflakes are messages from Him. They must be read and fitted like a super complex puzzle set that, because of global warming and the decline of piracy, will be lost to time. We must beseach His Noodly Appendage's reach!
Yamikage · 15 September 2005
So, anyone think Dembski will actually deal with the arguments that Mark has put forward, or you think he'll just put up another "You're out of your element Mark! OMG PWND!!!" post where he fails to repond to anything that has been written?
Timothy Chase · 15 September 2005
Norman Doering · 15 September 2005
Timothy Chase wrote: "... a great deal of the appearance of design ... getting passed through the sift of natural selection, and what arises from this process in the long-run are oftentimes very efficient solutions to engineering problems."
You have a point. A working machine can't really be called an "illusion." I over extended my reach.
Norman Doering · 15 September 2005
I said: "At the moment that looks like a real possibility to me."
I've changed my mind. It's not a real possibility. You can't blow the doors off this issue with pattern theory. The field of cognitive science can add depth to some Dembski concepts, like "detachability," but the fundamental question about designed or not designed in nature is really "intensionality" (if that's the right word). Did something intend for us to be here. Did something want us to be here.
The concept of intelligence is vague (can the genetic algorithm be called intelligent?) but at heart ID isn't about "intelligence" but about intension (intention?) and pre-planning and desire. That's what sets the concept of God in ID apart from the evolutionary algorithms and neural nets.
Dembski is barking up the wrong tree and so have I been. It's not a sign of intelligence you have to look for but a sign of intention.
snakefing · 15 September 2005
The big problem with using conditional probability with Dembski's EF is that you can't apply such an approach unless and until you understand the relevant conditions and processes. Of course, once you do understand those processes, the EF is rendered moot.
Consider how Dembski's EF would be applied by some design theorist who didn't know about the relevant weather conditions and physical processes. This person would not be able to apply the "right" conditional probabilities - he or she could only apply some set of a priori probabilities. The EF would then result in a (wrong) conclusion of design. Worse yet, having reached this erroneous conclusion, or hypothetical design theorist would then stop looking for other explanations. His error would only be discovered if some scientist ignored the EF and proceeded to discover the conditions under which the probability is appropriately high.
If you tried to use the EF with conditional probability, it is vacuous. You can't actually then calculate the right conditional probability until after you've determined whether it is in fact designed or not. If you try to use the EF with a priori probabilities, you'll get wrong answers.
Tracy Hall · 15 September 2005
Mt. Rushmore (or mountains, anyway) can be relevant... at least as a false positive! After all, the (now former) "Man of the Mountain" in New Hampshire was clearly of low probability ... so it must have been designed ... just like the "man in the moon" ... "the face on Mars" ... the constellations themselves, as such as they are ... etc ... etc ... etc ...
Moses · 15 September 2005
"taciturnus," if it's really the taciturnus that's been widely criticized on any score of subjects on which he's offered an opinion, is a shallow-thinking high school neocon that runs his own blog. I can't remember the name, but any ridicule directed his way would be lost in the competing voices (both left & right).
I don't know anything about this "Dave."
Steve · 15 September 2005
Actually Dembski has gone a great deal out of his way (and rather dishonestly, IMO) to reject the notion of conditioning on the data. There is an article (at ARN and other places) by Dembski where he tries to show how the Likelihood Principle and Bayesian methods is not valid with regards to the question of design. The Likelihood Principle is what motivates the idea of conditioning on data observed (and ignoring data that wasn't observed, but might have been). Hence this idea of conditioning suggests the commenter is completely clueless as to Dembski's actual argument.
Norman Doering · 15 September 2005
Steve wrote: "... Dembski ... tries to show how the Likelihood Principle and Bayesian methods is not valid with regards to the question of design."
I'm not a hundred percent clear on Dembski's argument, but the one thing he's right about is the concept of "detachability" in regards to an intended communication from an intelligent entity. I remember something from Dembski where he uses Carl Sagan's science fiction novel, "Contact," in which astronomers at SETI detected a radio signal of extraterrestrial origin. The signal was, in part, a sequence of beats and pauses representing prime numbers: 2, 3, 5, 7, ..., 101. Each prime number was represented by a sequence of beats equal to the number, with consecutive numbers separated by a pause. So you get: "110111011111011111110..." That would clue me in that it's not just noise. What natural process could do that? (It's possible that there is a natural process, but I have to agree - it's unlikely.) That's a good example of detachment. The prime numbers are detached and represented in the radio signal.
But I think you could apply Bayesian inferencing and pattern theory to SETI signals. Likelihood Principle and Bayesian methods would be the good thing for detecting detachment... I think.
Detachability would signal not just "intelligence" but also an intent to communicate something. (Though I suppose there could be unintentional, unconscious communication -- but unlikely by radio waves.)
I also do not see how detachability can be found in life, DNA, those functional machines that are supposedly specified and irreducibly complex.
Karl Lembke · 15 September 2005
RBH · 15 September 2005
germline · 15 September 2005
Is it me, or do the people here seem to understand Dembski more than the people at his blog ?
dave · 15 September 2005
Dr. Perahk,
My apologies for giving in to the general tendency to ad hominem that's been all too typical of this discussion.
You're correct that specificity as Dembski describes it is "detachable" not specified "before-hand," and you've charitably restated my argument according to its intent. Frankly I'm surprised and not just a little flattered that you singled out my little off-hand post for such rigorous treatment, out of dozens that were more thorough and articulate.
That said, I still believe you're wrong that the EF returns positive for the snowflake. But I was also wrong in arguing that specificity was applicable at this particular stage in the Explanatory Filter.
For any snowflake, no matter how rare the conditions that produced it, the explanatory filter would return negative in it's very first stage, because the snowflake's pattern is reducible to known natural algorithms.
Here's Dembski's description of the first stage of the EF: "At the first stage, the filter determines whether a law can explain the thing in question. Law thrives on replicability, yielding the same result whenever the same antecedent conditions are fulfilled. Clearly, if something can be explained by a law, it better not be attributed to design. Things explainable by a law are therefore eliminated at the first stage of the Explanatory Filter."
All snowflakes would be eliminated in this stage. The rarity of the conditions that produced them isn't relevant, because the first question that the EF asks is: can the pattern in question be replicated consistently in the same antecedent conditions? As you pointed out, an expert in snowflakes would immediately recognize the pattern as the product of natural conditions. That's it. It's that simple. The EF returns negative on all snowflakes, triangular or otherwise.
The question of specificity doesn't come up until you've eliminated both chance and natural conditions as a causal explanation. So whether specificity is or isn't simply a way of making the probabilities smaller is irrelevant, because the snowflake is eliminated before the specificity of the snowflake's pattern would even be examined under the EF.
dave · 15 September 2005
sorry -- "Perakh" not "Perahk"
dave · 15 September 2005
I should have said: "The question of specificity doesn't come up if the phenomena in question can be explained by chance or natural conditions."
dave · 15 September 2005
Steve · 15 September 2005
Steviepinhead · 15 September 2005
Dave, it's not clear to me that Dembski's "replicable by application of law" and your "explainable by natural conditions" are equivalent.
When what we're looking for are phenomena that are intentionally designed rather than naturally occurring, substituting your phraseology seems to come perilously close to "now that we've eliminated 'natural' causation, what's left must result from intentional design."
But that's precisely the issue under discussion--how to distinguish between the products of natural processes and intentional ones. Eliminating replicable processes resulting from the application of a well-understood law arguably winnows the field of candidates, without necessarily assuming that we can always tell what's natural or not. (And, please, understand that I'm very far from agreeing that Dembski successfully brings off the rest of his undertaking.)
Your equating of this initial and rather-limited winnowing process with the elimination of all naturally-caused phenomena goes too far, compressing into the first step what Dembski claims he can tell only after several additional steps--and in effect assumes your conclusion.
Norman Doering · 15 September 2005
Steve wrote: "To you (and me) such a signal would indicate something unusual. However, this is not true of everybody. Somebody who is clueless about the prime numbers might not consider it noteworthy."
Like an ant or a monkey. It takes intelligence to detect intelligence, so, for all we know we might be surrounded by intellects far in excess of our own which we don't recognise. But there would be more than primes in such a signal, there would be a "beat" and a sequence of incresing numbers even if you don't see primes.
"Again, this is 'detachability' is based on your own knowledge ..."
That is a point I concede. I agree with you.
"...and this is something that Dembski has rejected when it comes to using probability to evaluate various events as being designed or not."
And he rejected his best clue for the simple it didn't detect intelligence where he wanted to find it. Dembski is wearing religious blinders, but he is still a creative thinker.
Remember, I said: "But I think you could apply Bayesian inferencing and pattern theory to SETI signals. Likelihood Principle and Bayesian methods would be the good thing for detecting detachment... I think."
Likelihood Principle and Bayesian methods would incorporate much of our knowledge, such as about primes and more, eventually. You would train your neural and Bayesian networks with that sort of stuff.
The test would be -- can we see patterns and "detachability" and intelligence in the "language" of dolphins, the chromatophore signals of the octopus... We are already living among alien intelligences that we don't understand -- they're just like enough like us.
'Rev Dr' Lenny Flank · 15 September 2005
Norman Doering · 15 September 2005
Steviepinhead wrote: "When what we're looking for are phenomena that are intentionally designed rather than naturally occurring, substituting your phraseology seems to come perilously close to 'now that we've eliminated 'natural' causation, what's left must result from intentional design.'"
Another problem with that phrasing -- I think intelligence is naturally occuring. And not just natural, but extraordinarily varied.
If we consider Minsky's "The Society of Mind" and this quote: "What magical trick makes us intelligent? The trick is that there is no trick. The power of intelligence stems from our vast diversity, not from any single, perfect principle. - Marvin Minsky, The Society of Mind, page 308
Then we are likely to find enormous variation in the kinds of intelligence we encounter -- and the thing we have to avoid is anthropomorphising intelligence which is something theology seems to do.
dave · 15 September 2005
Henry J · 15 September 2005
Karl Lembke,
Re "(Is the number of possible curves in space aleph-one, or a higher order of infinity?)"
I'd answer that if I could remember what aleph-one is. But as I recall, the smallest transfinite number is the size of the set of integers (call it I), the power set of I is the size of the real number set, and the power set of that has the same size as the set of all functions on real numbers. I don't know if the number of curves in space would be larger than curves on a plane, or not. (I think not, but I'm not sure.)
Henry
Mark Perakh · 16 September 2005
Re: comment 48305 by dave.
The first part of your comment looks like an apology for ad hominem. If it indeed is, I appreciate it, and the apology is accepted. Moreover, you did something that is not easy to do --- you graciously admitted an error. I commend you for that and feel, therefore, that I owe you at least a brief response to your comment.
Regarding the reason why I have singled out your (and "taciturnus's") comments for a reply instead of replying to other comments, it is simple: Alan Fox emailed to me these two comments which otherwise I'd not have seen; neither he nor anybody else emailed to me any other comments, and I do not watch myself ID-promoting sites. Alan has in fact mildly prodded me to respond to you and taciturnus, and I did so.
As to your insistence that the rare snowflake would have been eliminated as a candidate for design, already in the first node of EF, I think you, like taciturnus, mix up two different approaches.
Of course every reasonable person would immediately recognize that the snowflakes are not designed. However, this conclusion is made outside EF.
To explain why I think so, I would like to point out that Dembski's trademark is inconsistency. This inconsistency is also on display in the quotation you provide. In this quotation Dembski says that events or objects that are caused by natural laws are excluded from a further consideration for a design inference immediately in the first node of his EF. It is easy, however, to provide other quotations from Dembski's output which offer a different story (I'll refer to them in a moment). Obviously, if we encounter in the writing of the same author mutually incompatible notions, and if we are nevertheless prepared to continue a discussion of this person's ideas, we have to choose for discussion those of his statements which are definitive for this author's views. Regarding the procedure for deciding whether an event is designed or is due to chance or a natural law, there is little doubt that the idea that is definitive of Dembski's approach is his schema of a three-step EF, represented by a graphic schematics of EF. Indeed, Dembski has published a graphic description of EF several times and has never given any indication that he is not adhering any longer to that schema. Again, this schema is represented by Dembski graphically in a schematic image of EF, published by Dembski several times. It is sufficient to look up any of these graphic images to immediately see that the quotation you provide in your comment is at odds with Dembski's definitive description of EF. Therefore in my discussion I referred to Dembski's EF as it has been described in those graphic representations published more than once.
In my view the first and the second nodes of EF make no sense (why it is so, I have explained at length in my book and in the essay in Skeptic). Now you quote Dembski asserting that already in the first node all events that are due to natural law are excluded from possible design inference, and therefore, you say, the snowflakes in question, if subjected to EF, would not be attributed to design. Unfortunately the quotation in question is contrary to Dembski's definitive description of EF. According to that definitive description, in the first node the probability of the event under investigation has to be first determined. If it turns out to be "large," the event is attributed to law (necessity, regularity) and the investigation stops, eliminating a possible design inference. However, probability of an event cannot be "read off the event," so the realistic procedure would be first determining that a law (necessity, regularity) was at work, and only from that we can conclude that therefore probability must be "large." Since the first node cannot be used as prescribed in Dembski's definitive description of EF, Dembski, in the quotation you provided, is forced to ignore his own (non-working) schema, and applies common sense, not noticing the discrepancy between this realistic approach and his own definition of the 1st node. The common sense approach meets no objection, but if we consistently applied Dembski's EF procedure, we'd have to abandon common sense and instead be looking for an impossible task of somehow estimating the probability of an event without any knowledge of its causal history. If we confine ourselves to EF, then, encountering the triangular snowflake, we are stymied in our investigation because we possess no clue as to its probability and hence, if following the EF schema, cannot decide whether it is due to a natural law, or perhaps is designed, or is due to chance, or to a combination of more than one cause.
In the real world, we do not apply EF (and neither does Dembski himself, as your quotation shows; indeed, contrary to his definition of 1st node, in the quoted passage he does not try estimating probability as prescribed by EF, but resorts to a common sense notions, which are not part of EF).
In fact what I and you really do, is ignoring EF and reasonably concluding that the triangular snowflake is due to a law of physics and is not designed --- and this conclusion is based on our ken rather than on an impossible estimate of probability prescribed by Dembski's definitive description of EF. We do so not by applying the (non-working) 1st node of EF, but without resorting to EF at all. We don't need EF to come to our conclusion. The quotation you provide just illustrates that Dembski himself could not apply his EF when discussing the events caused by regularities and used instead a common sense approach, not noticing that it was contrary to his EF description.
I think you have missed that discrepancy and, on the one hand, apply common sense, and on the other hand, try to fit it in with EF, while these two approaches are incompatible.
Btw, not only the 1st and the 2nd modes of EF make no sense, also the most crucial 3rd node in Dembski's schema is based on a fallacious separation of probability from specification as if they are two qualitatively distinctive categories. As I argued in my book and in the Skeptic essay, in fact specification is not qualitatively distinctive from probability but is rather just a component of the overall probability, so the 3rd node boils down to argument from improbability, which is another incarnation of the argument from ignorance, which also is a version of the God-of-the-gaps argument.
I am sorry to have briefly repeated something I have addressed in more detail in my Skeptic essay and in even more detail in my book. Cheers, MP
Mark Perakh · 16 September 2005
Addition: in my response to dave's post on ID site, I referred to small probability of triangular snowflakes because the weather conditions ensuring their appearance are rare. On the other hand, in my comment 48399 I pointed out that, encountering a triangular snowflake, we have no clue as to its probability. This may be misconstrued as a discrepancy between two of my statements. In fact there is none. When discussing the low probability of triangular snowflakes I did it from the position of Dembski's thesis which requires an estimation of probability. This estimation cannot be done as prescribed by EF, i.e. prior to determining the causal history of an event, but it can be done based on common sense and our ken, which enables us to reasonably assume such a causal history.
David Tye · 16 September 2005
Dr. Perakh,
I am Taciturnus on the Uncommon Descent blog. Thank you for taking the time to respond to our comments on Uncommon Descent.
With respect to your point about conditional probability, any probability measure implies a context and is therefore in some sense conditional. Common sense with respect to context is often assumed and not spelled out. Consider the following assertion:
The probability of a coin flip turning up heads is approximately 0.5
Is this an accurate statement? Apparently not, since the occurence of coin flips is itself a rare event (how often do people do it?)so the probability of a coin coming up heads is actually very small. Any reasonable person reading the sentence, however, understands that the probability measure implies the conditional context of a coin flip.
Now we have the case of a triangular snowflake. It is only seen under certain peculiar weather conditions. Dembski's EF asks: What is the probability of this snowflake's occurrence? Dembski does not spell out the context, but does this really present us with a dilemma? Shall we measure the probability in the context of the peculiar weather conditions in which it is seen, or shall we also include the context of the Mojave Desert in midsummer? I submit that it is justifiable common sense to measure the probability in the context of the peculiar weather conditions, and Dembski doesn't need to spell the point out.
Note that I don't need to have a causal knowledge of snowflake formation to make this determination. In fact, I have no idea how triangular snowflakes are formed. I only know that they occur under certain weather conditions and not under others. This is enough for me to define the conditions of the probability measure as including the peculiar weather conditions and not others.
As for actually determining the probability, isn't this the sort of thing weathermen do all the time? Given certain atmospheric conditions, the probability that a tornado will occur is X. If X is high enough, then an alert is issued. I would be surprised if a study could not be done that gave a reasonable estimate as to the occurrence of triangular snowflakes under certain weather conditions.
Flint · 16 September 2005
I gotta admit, Taciturnus' explanation makes little sense to me. If we define our context as that in which no coins exist, then the probability of a flip of heads is zero. If we narrow our context to all coin flips that come up heads, then the probability becomes one. If we get to define our context AFTER we make our observation, we can make our probability anything we please. How does this help us?
To me, this sounds like the typical creationist approach of starting with conclusions, and constraining the "relevant" data to only those which can be construed as supporting them. Drawing the bullseye around the arrow.
DrFrank · 16 September 2005
The EF argument sounds particularly paradoxical to me: the Explanatory Filter is used to detect design (and thus a lack of natural cause) but you aren't allowed to put anything with a natural cause into it. But how do you know something has a natural cause unless you've put it through the EF?
Surely the formation of triangular snowflakes in those weather conditions is only a theory (like evolution), in that it matches current empirical evidence, and consequently doesn't it make sense to run them through the EF to make sure the Designer isn't intervening elsewhere?
Although I think I've just seen where ID is going wrong - they should never have applied the EF to biological systems since they obviously have a natural cause ;)
I'm guessing, though, that I just have a `terrible misunderstanding of Dembski's theories' lol
Flint · 16 September 2005
DrFrank:
I think we've been around this block more than once. To the best of my knowledge, even Dembski admits that the "specification" in CSI is inherently subjective. Which means that you FIRST decide whether something was or was not designed. THEN you apply the EF to it, and out the other end comes the predetermined result, every time. And this is why the EF is simply never used by anyone for anything, even by Dembski. It can't tell you anything you didn't already "know"; it is incapable of producing surprising, nonobvious, or suggestive results. It is nothing more than a means of obfuscating preconceptions with mathematistical-looking incantations.
Can you imagine Dembski agreeing to be handed something he personally has no clue as to whether it was designed, and applying the EF to decide? You know; a blind experiment like actual scientists are obliged to do every day? He could save time and effort by flipping a coin -- and might even improve his accuracy.
Mark Perakh · 16 September 2005
Re: comment48414 by David Tye.
I appreciate your revelation of your real name.
Regarding the substance of your comment, it looks like you explain what Dembski thinks.
We can't read his mind, we only can read what he writes. Nowhere does he indicate anything along the lines of your argument.
Since some other posters have responded to your comment 48414, I will be very brief.
To properly analyze the problem of accounting for "context," (which anyway is beyond Dembski's thesis) we have to approach it from the likelihood standpoint rather than staying within pure probabilities. Your argument requires a re-statement in terms of likelihood.
I have written on that point before and in detail, so instead of repeating all of it here, I am taking the liberty of pointing to my post titled "A Free Lunch in a Mousetrap" - see here. In particular, look up there the example of soldiers traveling through a mountain gorge.
Cheers, MP
Mark Perakh · 16 September 2005
PS. I share the view of Flint as evinced in comment 48418. His is a good relevant point imo, although there also are other points against David Tye's argument.
Mythos · 16 September 2005
I have not read The Design Inference, so I'm relying on Mark Perakh's explanation of Dembski's notion of 'specification', viz. a 'detached pattern'.
If this is correct, it is a rather unsophisticated approach to the notion of a 'detached pattern'. A 'detached pattern', in this sense, is nothing more than a representation, a picture ('mental' or otherwise) of the structure of some fact.
But anyone who has thought seriously about representation or picturing (and part of 'thinking seriously' about picturing must include working through the development of that notion by Wittgenstein), will quickly realize that any pattern whatsoever can be used as a representation of any other pattern.
As an intuitive example of this consider one of Picasso's paintings, or ink-blots, or clouds, etc. As a less intuitive example of this (a la Wittgenstein, see the Blue Book) consider language itself. A picture is such according to a method of projection. Given an adequate method of projection, my living room can be a representation of a football game, a bacterial flagellum, or a recipe for pea soup.
And once this is admitted, as it must be, the idea of 'detachability' becomes pointless.
Jim Harrison · 16 September 2005
The humorist Roger Price invented the Droodle, little drawings with a comic titles. One of his droodles was a random collection of lines with the title "A Very Bad Drawing of Anita Eckberg."
Which pretty much exemplifies the problem with detachability.
Dave Sims -- "dave" · 16 September 2005
Mythos · 16 September 2005
Aureola Nominee, FCD · 16 September 2005
Professor Perakh,
apparently your whole point of showing that, without prior knowledge of the context, we cannot say anything at all about the probability of a given event, and so Dembski's EF cannot really tell us anything we don't already know, is lost on your detractors.
Your Afghanistan patrol example illustrates this very clearly. Thank you for your clear explanation.
David Tye · 16 September 2005
Dr. Perakh,
We seem to be talking past each other.
My understanding of probability is that it always implies a sample space over which the probability is defined (this is what I meant by context.) A probability detached from any sample space does not mean anything. Often, the sample space of a probability is not explicitly defined if it is obvious to the reasonable reader. This was the point of my coin-flipping example. When I say that the probability of a coin flip turning up heads is approximately 0.5, it is implied that the sample space is coin-flipping events, not the totality of events in the universe, in which case the probability would be vanishingly small.
The probability in question is the probability of triangular snowflakes. Over what sample space shall we choose to calculate this probability? Shall it be all possible events in the universe? Or perhaps all weather events on Earth? How about all weather events on Earth meeting certain rare conditions?
You are correct that Dembski gives us little guidance on how to choose the sample space, but that is because he doesn't need to (at least in this case.) I don't need to read Dembski's mind, or have knowledge of the causal formation of snowflakes, or smuggle in the assumption that snowflakes are caused by physical law, to know that the latter sample space is the appropriate one. I just need the information you have provided, which is that triangular snowflakes are observed only under certain rare weather conditions.
I'm not necessarily sold on the EF myself, by the way. I'm thinking about some of your other objections, in particular the objection that the EF artificially categorizes events as being exclusively due to either law, chance or design, when events may in fact be due to combinations of the three.
But I don't think the triangular snowflake counterexample works because it depends on positing an obviously unreasonable sample space, when the only reasonable sample space is clear from the definition of your example. The fact that it is obvious has nothing to do with Dembski and everything to do with how you defined the problem. There may be other examples that show that the EF doesn't work without question-begging assumptions about probability, but this isn't one of them.
Thanks for your patience and cheers...
DMT
Aureola Nominee, FCD · 16 September 2005
Let me get this straight:
if the non-obviously-unreasonable sample space for the formation of triangular snowflakes must include the existence of the appropriate conditions as a given...
...shouldn't the non-obviously-unreasonable sample space for abiogenesis, or the evolution of the blood clotting cascade, also include the appropriate conditions as a given?
steve · 16 September 2005
Arden Chatfield · 16 September 2005
Mark Perakh · 16 September 2005
David Tye · 16 September 2005
Let's see if we can help the unfortunate squad of American soldiers in Afghanistan with probability theory but without assuming a knowledge of the history of rock falling in Afghanistan.
The lieutenant in charge of the squad knows it took them 2 hours to move through the gorge. During that time, one rock fell, and it happened to fall right into the path where the soldiers were standing. The lieutenant estimates that his squad was in the path of the rock for about 1 minute.
The lieutenant does a quick calculation and estimates that the probability a rock would fall when his squad was in the rock's path to be 1 min / 2 hours = 1 min / 120 min = 0.0083.
This probability is small enough that he doubts physical law and/or chance can explain the fact that his squad was nearly killed by a rock. He radios his captain and requests to investigate the heights for insurgents.
The capt., unfortunately, dismisses the lt.'s probability calculation as bunk because they have no knowledge of the causal history of Afghan rock falling. He insists the lt. proceed on to the next gorge. In the next gorge, the lt. and squad are wiped out by an even larger boulder that "happens" to roll down on them right when they are in its path.
Aureola Nominee, FCD · 16 September 2005
In other words, Mr. Tye, you are estimating probabilities based on the context (the two hours' worth of observations by the lieutenant).
Which is exactly what Prof. Perakh was saying.
David Tye · 16 September 2005
Aureola,
I don't know what the sample space would be for abiogenesis or bloodclotting. That doesn't mean it might not be obvious in other cases.
You have moved to EF as applied to biological systems. I have little to say about this, as I don't know how it would be applied either, or how one would define sample spaces with respect to biological origins. I would agree that this is a major lacuna in ID theory, perhaps a defect serious enough to prevent ID from ever proving anything significant about biological systems.
Aureola Nominee, FCD · 16 September 2005
Mr. Tye,
whether context is obvious or not has no relevance to the validity of the EF. What has relevance is whether context must be considered or not.
Prof. Perakh argues very clearly that context must be considered for Dembski's probability calculation to have any meaning at all; this would sink EF faster than the Titanic.
If, on the other hand, context can be safely ignored, then Prof. Perakh shows that triangular snowflakes would give a false positive when fed into the EF.
That's all. Biological systems are nothing special from this standpoint. EF doesn't work with them because it doesn't work - period.
Flint · 16 September 2005
There seems to be a lot more context here than just the length of observation and the selection of the gorge as the scope of that observation. There is also the implication that only a single rock fell, and this is important. After all, if the lieutenant notices that hundreds of rocks of all sizes fall frequently, then the gorge is a dangerous place to walk even if none of the rocks came close to the soldiers. The frequency of rockfalls is also surely estimated from examining the terrain inside the gorge. It's also appropriate to note that walking through the gorge places the soldiers in a vulnerable position in a land where insurgents populate the territory at the top of the walls, something any military person would be intensely aware of. Finally, even if the falling rock were pure chance, it would be expedient to assume otherwise given the overall context.
This is a good example of why the leader on the ground is granted considerable autonomy in military situations: he has the most immediate knowledge. Tye's captain doomed the soldiers by ignoring the context the lieutenant had fleshed out. And this is what happens to those who pretend context doesn't matter. The challenge is in identifying those aspects of the context which are meaningful with respect to some stated purpose.
1st Lt. Aureola Nominee, FCD · 16 September 2005
Flint:
exactly right. In addition, field commanders couldn't care less about false positives; what they are worried about is false negatives, i.e. assuming a little noise is insignificant when in reality it was generated by an enemy readying his weapon for an ambush.
David Tye · 16 September 2005
But the lt. doesn't need to know any history of rock falls to estimate if the rock's plunge was due to design. He only needs the information gained while in the gorge, and that involves only one rock falling.
David Tye · 16 September 2005
Gentlemen,
I appreciate the civil exchange and wish you well, but I have to coach a soccer game...
Cheers,
DMT
dave · 16 September 2005
Flint · 16 September 2005
Aureola Nominee, FCD · 16 September 2005
Mr. Tye,
that's still contextual information about the frequency of rocks falling.
What cannot be done is watching one rock fall and decide how likely that event is without considering the context.
Alan (Fox) · 16 September 2005
Dave Sims · 16 September 2005
Aureola Nominee, FCD · 16 September 2005
Main Entry: 2ken
Function: noun
1 a : the range of vision b : SIGHT, VIEW <'tis double death to drown in ken of shore -- Shakespeare>
2 : the range of perception, understanding, or knowledge
David Tye · 16 September 2005
Mr. Fox,
Am searching for where I wrote the words you quoted of me: "failure to understand". Haven't found them yet. Not sure I will, but I will keep hunting. I appreciate the lesson in manners nonetheless, and wish you well.
Cheers,
DT
Dave Sims · 16 September 2005
Aureola Nominee, FCD · 16 September 2005
Mr. Tye,
I think that the similarity of your first name with Mr. Sims' may have induced Mr. Fox in error.
Indeed, it was Mr. Sims who wrote "After reading this, I'm wondering if Perakh understands the relevant English."
You merely wrote "...I now see what the ID defenders mean when they imply that Mr. Perakh seems to go out of his way to misunderstand Intelligent Design."
Dave Sims · 16 September 2005
Aureola Nominee, FCD · 16 September 2005
Mr. Sims:
the point is exactly that. Since the context is the event, the EF fails miserably, because it cannot establish probabilities for an event without considering its context... which includes why its probability is as high or as low as computed.
In other words: we have two ways to compute the chance of obtaining "heads" while flipping a coin; one is analysis of the "event" (i.e., finding out all theoretically possible outcomes of the event and thus determining the theoretical probability of each outcome), the other is observation of repeated instances of that event (i.e., flipping large numbers of coins and counting the outcomes).
Clearly, if we can follow the first route, the EF is useless, as we already know enough about the event to determine whether it was due to regularity, chance, design, or a combination of the three.
So, let's assume that we are forced to follow the other route. In this case, we only know what happens but not how. We must therefore determine the probability of our event without the benefit of knowing its mechanism. How do we accomplish that for, say, the coming into being of the blood clotting cascade, or abiogenesis, or rocks falling in a gorge?
Mark Perakh · 16 September 2005
Aureola Nominee, FCD · 16 September 2005
Mr. Sims:
the probability of an event can be computed in one of two ways:
a) by analyzing its mechanism and determining all its possible outcomes. Here, the EF if worthless, as we already now enough of the mechanism behind the event to know whether it is due to regularity, chance or design.
b) by observing as many occurrences as possible, and counting the various outcomes. Now, how do you propose we do this, in the absence of a mechanism?
Norman Doering · 16 September 2005
'Rev Dr' Lenny Flank on the question of specificity, respnding to: [it]
doesn't come up until you've eliminated both chance and natural conditions as a causal explanation.
"How, exactly, does one rule out not only all currently existing possible explanations that invoke either chance or 'natural conditions,' but *all possible future explanations that have not even been thought of yet*?"
If it's part of something functionless and "seems" to be meant as a communication. Dembski is wearing religious blinders. He doesn't realize how many of his examples are communications attempts. From SETI, to Rushmore, to rocks that look like a constellation -- they are all trying to communicate something - not make something. And, as has been pointed out already, there would be a lot of false positives -- I've seen highly specified faces in clouds -- pretty close to Rushmore is specificity.
Specificity and detachment would signal to me that an intelligence is trying to communicate something to something else. It takes knowledge of what's communicated to see what is detached and specified. You can't just look at the message, you have to know the world the message is about. Specificity alone doesn't work because we know evolutionary algorithms produce irreducible complexity and specified complexity.
If specificity is claimed to be a measure of some ambigous term called "intelligence" than the evolutionary algorithm is intelligent and ID and Darwinian evolution are one and the same thing.
"Dembski's complicated-sounding math boils down to nothing more than 'if we can't explain it now, then goddidit.' How does that differ from the thousand-year-old 'god of the gaps' reasoning?"
It's no different at all -- but turn it around, isn't there also a "naturalism of the gaps"? The difference is, God is moving out and naturalism is moving into the gaps.
I don't see any use for Dembski's filter -- but some of his concepts do look useful. I think one day someone might do something with ideas about specificity and detachment, like use them to decode the language of dolphins and octopi. -- That's just an intuition, not an argument.
Dave Sims · 16 September 2005
Dave Sims · 16 September 2005
'Rev Dr' Lenny Flank · 16 September 2005
'Rev Dr' Lenny Flank · 16 September 2005
Steviepinhead · 16 September 2005
Let's see, then, DS (and let's hope those initials don't mask the one they make us think so non-fondly of...):
Instead of taking WAD at his word, as Dr. Perakh does, you are instead suggesting that a "charitable interpretation" would be to credit WAD with what you wish he had said.
You guys slay me (and, no, employing this rhetorical device is not intended to promote the commission of any illegal act).
Dave Sims · 16 September 2005
'Rev Dr' Lenny Flank · 16 September 2005
Stuart Weinstein · 16 September 2005
'Rev Dr' Lenny Flank wrote:
How, exactly, does one rule out not only all currently existing possible explanations that invoke either chance or 'natural conditions,' but *all possible future explanations that have not even been thought of yet*?
To which Dave Sims replies:
You don't. This sounds like a version of the problem of induction. It's a stickler, no doubt.
But can you apply the same demands to all scientific hypotheses? Why stop at any hypothesis, ever? There's a measure of induction involved in all scientific reasoning.
BINGO!
We don't. Science never stops looking for a better, more concice, more precise theory to explain an ever wider variety of phenomena.
We don't stop observing and/or measuring our universe or formulating new explanations. THere may come a time, when sceince fails to make progress, but that time is certainly not now.
"If you set the bar that high for ID and not for all the sciences, it begins to look like special pleading."
But we're not. We want to hold ID to the same standards, and when we do, we find it wanting.
Dave Sims · 16 September 2005
RBH · 16 September 2005
Dave Sims neglects to note that in The Design Inference and other 'canonical' versions of the Explanatory Filter, one determines whether an event is explained by natural law by examining its probability. That is, in Dembski's reasoning we do not conclude that some event E is a high probability event because we have side information that E is a consequence of natural causes. Rather, we infer that E is a consequence of natural causes (what Dembski calls "regularity") because it is a high probability event, given specified antecedent conditions. We must therefore estimate the probability before entering the Explanatory Filter's decision tree. The first two decision nodes of Dembski's Explanatory Filter are traversed solely by examining probabilities.
RBH
Dave Sims · 16 September 2005
Ok, RBH, I'll bite. In the particular case of the snowflake, what constitutes the event itself and what constitutes "side information"?
'Rev Dr' Lenny Flank · 16 September 2005
'Rev Dr' Lenny Flank · 17 September 2005
Wow, all the IDers suddenly got very quiet . . . . .
Moses · 17 September 2005
Dave Sims · 17 September 2005
'Rev Dr' Lenny Flank · 17 September 2005
'Rev Dr' Lenny Flank · 17 September 2005
AR · 17 September 2005
Dave Sims · 17 September 2005
Aureola Nominee, FCD · 17 September 2005
Mr. Sims:
If we know that an event was the result of natural processes, of course we do not feed it into Dembski's EF.
So, remind us: what do we do feed it?
And how do you propose to compute probability for an event of unknown mechanism, which is the ONLY instance where Dembski's EF would be applicable?
Let me give you a specific example, in case you miss the significance of my question: how do you propose we compute the probability of abiogenesis?
If you find that a loaded question, what about the coming into being of the blood clotting cascade? How do we know whether it was out of necessity (i.e., given the appropriate conditions - you know, like the "right" weather for the formation of triangular snowflakes -it simply had to evolve), out of chance (i.e., those organisms where haemorrhages managed to stop on their own survived to produce more offspring than those where bleeding did not stop as easily), or out of design (i.e., goddidit)?
I have more than a passing familiarity with math, so I would be really, really interested in hearing about your computational method.
'Rev Dr' Lenny Flank · 17 September 2005
'Rev Dr' Lenny Flank · 17 September 2005
'Rev Dr' Lenny Flank · 17 September 2005
'Rev Dr' Lenny Flank · 17 September 2005
Dave Sims · 17 September 2005
Aureola Nominee, FCD · 17 September 2005
Mr. Sims,
don't go away yet, please. We're still waiting for your revolutionary method for computing probabilities for an event of unknown causal history.
Also, if the formation of triangular snowflakes is, given the appropriate weather, highly likely, why shouldn't the evolution of eyes, given the approppriate evolutionary pressures, be just as highly likely?
If, as you said, the context is the event, why should we consider the context of the snowflakes but not the context of abiogenesis?
Come on, Mr. Sims. Don't be afraid. We don't bite. We may dismantle the occasional piece of IDBS, but simply out of sincere compassion for beheddled thinkers.
Dave Sims · 17 September 2005
Dave Sims · 17 September 2005
Out for now. Real life calls. Back later.
Aureola Nominee, FCD · 17 September 2005
It's OK, Mr. Sims. I understand.
It's not as if Panda's Thumb hasn't already seen many ID "supporters" turn tail and disappear, only to return, one week or six months later, and start afresh, without answering any questions.
Don't worry, people here tend to remember such less-than-stellar performances, especialy thanks to PT's policy of storing all comments, no matter how asinine they might be.
'Rev Dr' Lenny Flank · 17 September 2005
'Rev Dr' Lenny Flank · 17 September 2005
'Rev Dr' Lenny Flank · 17 September 2005
Henry J · 17 September 2005
How many posts does it take to say that the identification of deliberate engineering requires noticing a similarity to something that is already known to have been deliberately engineered?
Imo, using "design" in the context of "I.D." is an attempt to distract the audience of the "I.D." arguments from realizing that a "design" has to then be engineered, or manufactured.
A machine is built by engineers after being designed, but an organism is in effect built by its parents - which makes the analogy between them unreliable for determining origins of species. At least that's how it seems to me - unless I'm missing something? (And at the risk of stating what's obvious to most of the people here.)
Henry
AR · 17 September 2005
Stuart Weinstein · 17 September 2005
AR writes"Given the above description, what is (if any) connection between it and Dave Sims's convoluted and obscure statement which he declares to be a "standard definition of Bayesian analysis"? I still view it as mumbo-jumbo. Furhermore, what is the connection (if any) between Bayesian reasoning (in whatever rendition) and the first node of EF? I see none, but perhaps Dave Sims is indeed in possession of some advanced knowledge which he would be kind to explain in "plain English," using again Dembski's expression?"
Well you know the old saying; If you can't dazzle them with brilliance, baffle them with bullshit.
'Rev Dr' Lenny Flank · 17 September 2005
Alan · 18 September 2005
Sorry if this is a stupid question but...
Before analysis with the EF, one has to select an event and assign a probability? How can a biological structure or process be considered an event and how can (presumably) the structure or process be assigned any meaningful probability? The EF doesn' seem to work for analogies, but how can you get off first base when applying it to the real world, if the selection of an "event is so problematic?
Dave Sims · 18 September 2005
Lenny, I've answered your questions. The problem of induction cannot be solved, (well, Popper claims to have "solved" it -- actually he does an end-run). No one can ever specifically eliminate all possible explanations for a given event. If this were a show-stopper for ID, it'd be a show-stopper for all scientific hypotheses. If you can explain why this is specifically a problem for ID and not all of science (which of course it IS, which is one reason we talk about falsifiability and not verifiability), I'd be happy to engage the conversation. Otherwise I'm done talking to you.
Dave Sims · 18 September 2005
AR -- ok, fine, simpler language: Bayesian analysis, as I understand it, is a way of measuring epistemic confidence -- degrees of belief -- that a given event will happen, as a function of the probabilities surrounding that event.
Reading some of Dembski's responses to Sober et. al., I get the impression that the Bayesian/likelihood approach doesn't play much of a role in locating design. So, probabilities only play a role at the first node of the EF -- if the probabilities for the event are high, then we say the event was a result of necessity. If it's small, then we move on and decide whether or not the event was caused by chance or design.
So, Bayesian analysis is one proposed way of rescuing inductive reasoning from Hume -- and Lenny (down boy) -- but maybe not in conjunction with Dembski's EF.
Aureola Nominee, FCD · 18 September 2005
Mr. Sims,
would you mind answer some of my questions as well, please?
1) How do you compute the probability of abiogenesis, or of the coming into being of the human eye?
2) Do you think that, just like the existence of the appropriate preconditions must be factored in the probability for the appearance of triangular snowflakes, so should be the appropriate preconditions for abiogenesis? If not, why not? If so, how do you compute the probability of those preconditions?
Thank you for not evading them once again.
Henry J · 18 September 2005
Re "If this were a show-stopper for ID, it'd be a show-stopper for all scientific hypotheses. If you can explain why this is specifically a problem for ID and not all of science"
Oh for Pete's sake. It's a show stopper for the "explanatory filter" thing, not for I.D. per se.
Henry
AR · 19 September 2005
David Tye · 19 September 2005
David Tye · 19 September 2005
Mr. Fox,
If you would like to continue the discussion, drop me a line at dmtphilosophy at hotmail.com. I won't be posting here anymore because the conversation is degenerating into exchanges of profanity.
DMT
'Rev Dr' Lenny Flank · 19 September 2005
Flint · 19 September 2005
Dave Sims · 19 September 2005
The original subject of this thread was the question of whether a triangular snowflake would produce a false positive under the EF. I've asked several pointed questions in this regard and recieved no direct explanations, but I have recieved a number of unrelated queries that I've done my best to engage in good faith, the responses to all of which seems to be: "I have no idea what you're talking about, therefore you're wrong. Now answer my question."
I'm afraid I can neither conquer the problem of induction, nor invincible ignorance. So, I'm retiring the thread. And no, you can't email me offline.
But for the record, guys, I've analyzed the response pattern that Lenny, Flint and AR seem to have mastered. So for future reference, in case you forget, here's a 4-step process that I like to call the Panda's Thumb Dialectical Design Dodge:
1. Ask a question of respondent. The less relevant to the current thread, the better, particularly if respondent him/herself has unanswered questions on the table.
2. No matter what answer is given, NEVER address it on its own merits; rather, yawn or curse at the respondent and demand that they answer the question.
3. When the inevitable good-faith clarification is posted by the sucker, er, respondent, proclaim loudly that their answer makes no sense whatsoever. Some forehead-slapping and groans of disbelief will help you get into the spirit of things.
4. Repeat steps 2 and 3 until respondent(s) give up in exasperation. Crow loudly. Buy beer. Look fondly at your autographed picture of Daniel Dennett.
Aureola Nominee, FCD · 19 September 2005
Mr. Sims:
Thank you for evading my questions. Once again, you handwave and project your behaviour onto your interlocutors.
Please, if your precommitment to a magical worldview hasn't completely obliterated your critical thinking skills, I would really like you to explain the difference, if any, between the formation of triangular snowflakes and abiogenesis.
Do we consider the appropriate preconditions of both? Do we not? In this case, why? If yes, how do we compute the probability of abiogenesis occurring? Can we? Can Dembski's EF tell us anything at all about events where the causal mechanism is unknown?
Mark Perakh · 19 September 2005
Dave Sims · 19 September 2005
Aureola, it's a good question and probably fruitful for discussion, but I've grown bored of navigating the trolls.
Please post at uncommondescent and we may be able to continue.
Mark Perakh · 19 September 2005
I believe comment 48804 by Dave Sims is just at the border between reasonable debate and an irrelevant display of arrogance. However, it has not been moved to bathroom wall - and this is just one more illustration of PT's overall tolerance. Does Dave Tye approve such comments as Sims's latest exercise in sarcasm?
Dave Sims's recent comments wherein he accuses his opponents of "invincible ignorance," while his own contortions trying to redefine Bayesian approach make one suspect his not quite adequate familiarity with the matter in point, speak for themselves.
Aureola Nominee, FCD · 19 September 2005
No, thank you, Mr. Sims.
I think I'll stay where both sides can articulate their point of view without fear of censorship.
Flint · 19 September 2005
After some reconsideration, I admit I can't find any trace of the "response pattern" Dave Sims allegedly finds in my remarks. I do, however, find clear and present indication of what Aureola Nominee calls "projecting your behaviour onto your interlocutors." Perhaps Dave Sims should understand that explanations based on false premises cannot be honestly answered by anyone presuming those premises. When the false premises are corrected, answers based on them are quite clearly shown to make no sense.
Sims: Have you stopped beating your wife?
PT crowd: Since I do not beat my wife, this question makes no sense.
Sims: You are evading the question, which calls for a simple yes or no. How can we continue this discussion if you refuse to answer?
PT crowd: But I DID respond. Either a yes or a no answer would tacitly ratify an assumption contrary to fact. Do you wish me to lie?
Sims: You people NEVER address a question on its own merits. I give up.
Hopefully, we can all understand that essentially this exchange describes all debates here between PT regulars and creationists. The creationists insist that their assumptions be accepted as fact, PT people refuse to do so on the grounds that the assumptions are false, the creationists accuse the PT people of evading the questions and responding in bad faith, and eventually most of them give up in disgust. And I suppose it makes equally little sense from the creationist point of view for the PT crowd to be saying "Assume there are no gods. NOW, in light of this absurdity, how might we explain what God has done?" Equally senseless, right?
JS · 19 September 2005
Oh, crap. D. Sims left. Well, in the hope that he's still lurking, and in the interest of fairness, I'll try to explain to the best of my abilities the answer to the one specific, unanswered question of his that I've seen. The question reads:
"If you can explain why this [the induction problem] is specifically a problem for ID and not all of science [...]"
The specific problem for the EF in particular and most of ID in general is not so much what you describe as the induction problem (is this a well-known philosophical problem? I'm rather lousy at philosophy).
The problem with the EF is that it tries to infer design by ruling out laws of nature (and there really is only this single step, since the distinction between 'regularity', 'chance', and 'luck' that the filter makes is merely a question of sample size - which is another reason to be sceptical: Redundancy in a model is a Bad Thing in science).
Now there are two ways (in principle) to rule out natural law (please note that I am not necessarily prepared to grant that the only alternative to natural law is design - but that is rather irrelevant to this explaination):
Either the object, event or whatever else you put through the filter is impossible (or sufficiently highly improbable - as compaired with the time available) through solely natural law - in which case we don't need the filter and Ockam's Razor is slashed across its throat, to put it a little poetically. Note that in this case probability/possibility is an intrinsic quality of the event or object in question - and as such should be easily determinable. The fact that none have, to my knowledge, been found yet suggest either that there are none, that there is no such intrinsic probability - or at least no way to reliably determine it, or that Intelligent Design Creationists are putting in too few lab hours.
Alternatively one might rule out natural law by examining the probability of an event or object based on its causal history. This, however leaves us with the problem that this probability is, in general, not single-valued (or, in layman's terms: The probability computed by looking at two different causal histories is in general not the same). Since the path-dependent probability function is in general not single-valued, one has to take into account *every possible causal history,* which is of course an impossible task.
Now, your question was why this problem does not apply to scientific theories. The answer is simply that scientific theories do not try to prove negatives. "But," I hear you cry, "Newton's 2nd Law tries to prove that there is no maximum attainable speed!" (This is actually a bad example, but I'll get back to that). In fact N2 doesn't do anything remotely like that. N2 tries to prove that F = ma. This is a *positive, testable prediction.* In fact, if N2 did postulate that there was no ultimately unattainable speed, then that postulate would also be impossible to verify. That postulate would, however, not be science.
(The reason that N2 was a bad example is that N2 actually says that the sum of external forces equals the change of total momentum of a given system - and momentum and speed aren't trivially related.)
There is actually also a semantic problem with the EF when applied to biology: When applied to biology it seeks to distinguish between natural and supernatural phenomena (for want of a better term - in fact 'supernatural phenomenon' is a contradiction in terms). This, of course, means that it is not science, since by way of definition stuff described by science is *natural*.
Please note the philosophical difference between this and the notion that science excludes the supernatural. If something is what we would today call supernatural and is then described by science, then it would *cease to be* supernatural. Which BTW is why trying to describe God(s) scientifically is a pointless exercise, since she/they are *by definition* supernatural - so describing her/them by science would also be a contradiction in terms.
- JS
'Rev Dr' Lenny Flank · 19 September 2005
'Rev Dr' Lenny Flank · 19 September 2005
'Rev Dr' Lenny Flank · 19 September 2005
'Rev Dr' Lenny Flank · 19 September 2005
Flint · 19 September 2005
But a special sort of god of the gaps. They frame all their arguments in such a way that their faith is the default unless someone can prove it wrong to their satisfaction. In this way, they are no different than Hovind's bogus reward for anyone who can "prove evolution" with Hovind as the judge.
The notion of continuing a "discussion" on a forum where the opponent's posts are reflexively deleted and the opponents are banned for posting them is so comically characteristic of Creationist "debate" that you have to laugh. Yet one further (though entirely superfluous) illustration that it's simply impossible to be honest and a creationist at the same time.
'Rev Dr' Lenny Flank · 19 September 2005
'Rev Dr' Lenny Flank · 19 September 2005
Stuart Weinstein · 19 September 2005
Mark writes "If you, Dave Tye, can point to a specific comment using profanities, I may consider moving it to the bathroom wall."
I did use the profanity "bullshit" when writing "If you can't dazzle them with brilliance, baffle them with bullshit"
I apologize to all of the, until just recently, virgin ears.
Mark Perakh · 20 September 2005
I see, Stuart. My not so virgin (given my age of 81) ears failed to register the profanity you cite. Since yours was a generalized statement, not adressing any particular person, I'll let it stay. Cheers! MP
Alan · 20 September 2005
Alan · 20 September 2005
David
My email has been returned "mailbox unavailable"
Joseph O'Donnell · 20 September 2005