I see that in the unending TSZ and Jerad Thread Joe has written in response to R0bb
Try to compress the works of Shakespear- CSI. Try to compress any encyclopedia- CSI. Even Stephen C. Meyer says CSI is not amendable to compression.
A protein sequence is not compressable- CSI.
So please reference Dembski and I will find Meyer’s quote
To save Robb the effort. Using Specification: The Pattern That Signifies Intelligence by William Dembski which is his most recent publication on specification; turn to page 15 where he discusses the difference between two bit strings (ψR) and (R). (ψR) is the bit stream corresponding to the integers in binary (clearly easily compressible). (R) to quote Dembksi “cannot, so far as we can tell, be described any more simply than by repeating the sequence”. He then goes onto explain that (ψR) is an example of a specified string whereas (R) is not.
This conflict between Dembski’s definition of “specified” which he quite explicitly links to low Kolmogorov complexity (see pp 9-12) and others which have the reverse view appears to be a problem which most of the ID community don’t know about and the rest choose to ignore. I discussed this with Gpuccio a couple of years ago. He at least recognised the conflict and his response was that he didn’t care much what Dembski’s view is – which at least is honest.
I think everyone accepts the possibility that there could be insurmountable gaps. At least in principle.
It’s never worked in any other branch of science, but there could be a first time.
kieths,
OK, what you seem to be saying is that a specification must PRECEDE the design. Otherwise, how could we possibly know whether the design met the specification. And to do this, we must know the INTENT of whoever produced the specification.
So I still don’t see the point. The ID folks SEEM to be doing the equivalent of looking at a bridge hand, ASSUMING this hand met the specification, deducing the specification from the hand, and calculating that it was highly unlikely.
Joe F.:
No, ‘specified complexity’ is the term Dembski uses, at least in Specification: The Pattern that Signifies Intelligence:
You continue:
I don’t think anyone will boo. I, for one, agree with what you wrote (though substituting ‘CSI’ for ‘SC’). It’s just another way of saying that a tornado in a junkyard won’t produce a 747. We all agree with that, but we also know that evolution is not a tornado in a junkyard. Selection makes all the difference.
No, Dembski explicitly states that P(T|H) represents the probability of producing the object in question via “Darwinian and other material mechanisms.” If that probability is low enough, then natural selection and other “material mechanisms” are ruled out.
I don’t see how the LCCSI, even if it were correct, would solve the circularity problem, since P(T|H) is right there in the CSI equation, and H includes “Darwinian and other material mechanisms.” You have to know that something couldn’t have evolved before you attribute CSI to it.
That’s right. It requires that incremental steps connect the various “islands of function”. That diverging descent with modification leads to a nested hierarchy is important evidence. We have all sorts of examples of intermediates in nature. We can not only find them in extant nature, but in fossils, as well.
I think both assume that what is was planned. I’m not sure how you think they escape that fallacy.
Flint,
It doesn’t necessarily have to precede the design temporally, but the specification needs to be independent of the event itself. For example, if I roll a 10-sided die nine times and come up with my Social Security number, then the outcome is specified even though I didn’t suspect ahead of time that I would get my SSN.
Yes, in one sense they are doing that, because evolution doesn’t aim for particular outcomes. They are drawing a bullseye around the arrow after it hits the target.
In another sense, they are saying that things like the eye are clearly specified, because sight is so obviously a useful thing for many organisms to have. The fact that eyes exist (particularly complicated ones) is a specified event that requires an explanation — it’s not just one random, unremarkable event among millions of other random, unremarkable events.
Where we differ from them, of course, is that we see how undirected evolution can produce things like eyes, while they don’t (or won’t).
Of course there are many kinds of eye. Presumably the result of many different sequences. All isolated islands, no doubt.
Mung picks a fight with himself — and Dembski:
I didn’t say it was. I said “If that probability is low enough, then natural selection and other “material mechanisms” are ruled out.”
I didn’t say that. I said “If that probability is low enough, then natural selection and other “material mechanisms” are ruled out.”
No one said otherwise. Who are you arguing against?
I didn’t say the probability was ruled out, whatever that means. I said “If that probability is low enough, then natural selection and other “material mechanisms” are ruled out.”
That’s Dembski’s argument. If you disagree with him, then fight it out with him, not me. He writes:
In other words, Dembski is saying that 10120·ϕS(T)·P(T|Hi) must be lower than 1/2 in order to rule out each chance hypothesis Hi. It’s exactly as I said: “If that probability is low enough, then natural selection and other “material mechanisms” are ruled out.”
Mung, how can you ever hope to defend ID if you don’t even understand what the ‘ID theorists’ say? Why are you even participating in this discussion if you don’t understand evolutionary theory and you don’t understand ID ‘theory’?
Hmm. I think Dembski used “CSI” differently earlier. Anyway whatever the terminology …
I think he has used both LCCSI and his P(T|H)-from-Darwinism term. It’s possible that he has quietly abandoned the LCCSI though I think that as recently as 2007 when he reprinted No Free Lunch it was still there.
Joe,
I don’t doubt it. Dembski is notorious for being sloppy and inconsistent in his use of terminology.
If the equation he uses for CSI contains the P(T|H)-from-Darwinism term, then he’s got a circularity problem. I think his use of LCCSI is orthogonal to that.
I know he and Marks are still appealing to “conservation of information”, but they no longer seem to be saying anything about conservation of CSI.
P.S. Did you see that Mung quoted Hubert Yockey citing you?
Gpuccio blows a gasket:
And:
Gpuccio,
I have calmly explained to you, in three separate comments (link, link, link), why the dFSCI concept (as you’ve defined it) is circular. You obviously don’t agree, but instead of responding rationally, you’ve chosen to impugn my character, as if that were somehow a rebuttal.
The truth or falsehood of my argument doesn’t rest on whether I am an angel, a devil or something in between. It rests on the argument itself. If you disagree with my argument, show us precisely and explicitly where it fails. If you can’t rebut my argument, then you’ll have to come to grips with the fact — excruciatingly painful though it may be — that I’m right, and that your argument really is circular.
Yes, and you’ve made yourself look foolish and intemperate. I suspect you’ll regret it once you cool down.
Mung,
Why, did they have the answer? You’ve linked several times in the thread, what specifically are you talking about? If you have an answer, just state it. It’s not for me to read up on how I would do it, I’m not going to. It’s a challenge to you. Are you following KF’s route where every time someone disagrees with you it’s simply because they’ve not read his “always linked”? And if they’ve read it and still disagree? They have a cognitive lack.
Except you took no steps apart from to say
If I’ve missed something do tell. And that you could do more if you cared to, well frankly I don’t believe you. You are happy to spend many hours posting at UD but no time at all doing what you should trivially be able to do, i.e. attempt to detect design.
You seem to be mixing your tenses. You are saying you did look at the documents at the frequency level etc. but then say that you won’t bother because it will not prove diddly. So I think you never did any of that, you are just trying to spin what little you did do in a positive way and your writing revealed more then you intended to.
Pro-tip: If you’d have done what you said you did you would have found the answer. Close enough anyway.
This is how the UD crowd act when you challenge them to do what they say themselves can be done. All from the same thread
Joe,
Mung,
Joe,
Mung,
Joe,
Mung,
Joe,
Gpuccio,
Joe,
Kind of not. Pretty early in Bilaterian evolution, light-sensitive regions, presumably developmentally controlled by an ancestor of PAX6, arose. This is highly conserved – you can put eyes on Drosophila‘s elbows using mouse PAX6. The pigments we use are also highly conserved among sighted organisms. As these early photo-detecting tubes diverged, some founded the different dynasties – Arthropoda, Mollusca, Vertebrata etc, and each took a different approach to subsequent tuning – focussing, light-gathering, protection, integration with the nervous system and blood supply etc – all developmentally downstream of PAX6’s basic ‘put an eye here’ instruction.
This also raises an interesting angle on the ‘specified’ part of a protein string’s ‘function’. It is highly unlikely that there is a mere straw-in-a-universe-sized-haybale number of protein sequences that can say ‘put an eye here’. But once one does, the obvious usefulness of even ‘half-an-eye’, and the downstream dependence on the sequence that started the ball rolling, render it barely mutatable. ID will no doubt insist that the commonality must be Common Design – either there is only one way of saying ‘put an eye here’, or the Designer happened to already have a ‘put an eye here’ gene in his toolkit; no point coming up with several. I’d say those would be pretty obtuse interpretations.
So, eyes arose once, or 50+ times, depending on where you like to draw your arbitrary line demarcating the outputs of an essentially continuous process.
Concurrent anagenesis = divergence. Kill all the ancestors (they die anyway) and you have the discrete, sealed categories beloved of Creationists. If you try real hard, you can only see isolated twigs, suspended in mid-air, and deny the tree.
Gpuccio
I have not been following this debate in detail because it recaps discussions we have had before – but I was suprised to read:
When we discussed this before I made the same points about circularity that Keiths and Zachriel have been making. Of course neither of us changed our mind – noone ever does in these debates – but I thought we both had respect for each other’s views. There was nothing degrading, immoral or “cognitively infamous” (whatever that means) about the discussion. What changed? Am I also now a liar?
Yes. He relies on dFSCI to ‘prove’ the Gap is insurmountable.
Common descent.
That’s right. And because of common descent, we can predict the existence of intermediates. That’s how paleontologists often work, for instance, by predicting the placement of fossils, then mounting expeditions to faraway lands to study the relevant strata.
Sure. But it’s important to keep in mind that the more ancient the transition, and the more delicate the structures, the less likelihood of finding fossils of those intermediate forms. Pointing to such a gap doesn’t demonstrate there is a problem with the overall theory of common descent. And the more we look, the more we fill in the gaps.
It’s also important to remember that organic populations are not simple. It’s not simple point mutation, fixation, then a new homogeneous population. Rather, populations are collections of highly diverse organisms with highly diverse traits. Furthermore, incremental change on the molecular level may not be simple point mutations, but can include homologous and non-homologous recombination, exon shuffling, duplication of part or entire genomes, and many other forms of variation.
As for superfamilies, many have roots near or even before the most recent common ancestor. As ‘incremental change’ may includes shuffling, that means there may not be a simple divergence pattern between superfamilies. While the true history is largely shrouded in time, the fact that random sequences can form protein folds indicates that the origin of the original proteins is not so unlikely.
Interestingly, gpuccio apparently accepts evolution within superfamilies, even though they have diverged wildly from their ancestral forms. Each IDer seems to have their own preferred flavor of gap.
No. Evolution can create functional complexity. You are assuming your conclusion.
Gpuccio is appealing to induction. But there are infinite ways that the future – and in the case of Darwinism, the distant past – does not resemble past experiences.
See this comic from XKCD, as an example.
By induction, from the Lenski experiment, we can expect between two and four function enhancing mutations per decade in a small population of bacteria.
I’m not sure what gpuccio is inducing, but it’s not based on observable rates of evolution.
Sure, but his induction is faulty.
For evolved sequences:
He says that everything with dFSCI for which the origin is known is designed, but that’s not the case with evolved sequences—unless you have apriori rejected the conclusions of evolutionary science. And that is the very thing he is attempting to show.
That’s somewhat different than simply saying “no known deterministic explanation”. Evolution really isn’t deterministic, but contingent. And you only mean what you consider having uncontrovertible causes.
That weakens your induction considerably. You only include the uncontroverible in the class, and exclude the plausible; and there are good reasons to think that biological structures are inherently different than the rest of the class. The induction just doesn’t work. You’re just back to arguing the plausibility of evolution, something strongly supported by the vast majority of biologists.
Yes, you do need to know the temperature and pressure, as well as any minerals dissolved in the water.
No. That you don’t have to know.
You seem to be arguing that we don’t have to know the origin of gravity to predict planetary orbits, or the origin of life to understand how it has evolved since its inception.
Mung quote mines me, and KF responds to the quote mine instead of looking at the original:
Mung:
KF:
Since we are communicating in the third person:
Mung, does KF understand that the position I was presenting was Dembski’s, not my own? Thus KF thinks that Dembski’s position is deserving of excoriation. I trust that he will soon be scolding Dembski in a new blog post.
Also, when will he be responding to my challenge?
Gpuccio writes:
Gpuccio,
Suppose you and I are arguing over a particular lengthy gene. I think it evolved, and you think it was designed. We agree to use dFSCI to try to resolve the dispute. To do this, we follow your guidelines:
a) We examine the gene and each of us independently determines that it contains 1.4 zillion bits of dFSI, well above the threshold for dFSCI, so we move on to criterion b.
b) We examine the known “necessity mechanisms”, including Darwinian evolution. You decide that none of them (including evolution) could have produced the gene, so you declare that it has dFSCI. I decide that the gene could have evolved, so I declare that it doesn’t have dFSCI. It’s controversial, as you said above.
We thus have two kinds of dFSCI: dFSCIgpuccio and dFSCIkeiths.
If something has dFSCIgpuccio, all it means is that
1) it is too complex to have come about by pure random variation without selection; and
2) gpuccio doesn’t think it could have been produced by Darwinian evolution or any other “necessity mechanism.”
If something has dFSCIkeiths, all it means is that
1) it is too complex to have come about by pure random variation without selection; and
2) keiths doesn’t think it could have been produced by Darwinian evolution or any other “necessity mechanism.”
Back to our gene. You say it has dFSCI, and I say it doesn’t. How do we break the impasse and decide whether it really has dFSCI? The only way we can resolve the dispute is to determine, once and for all, whether the gene could have evolved. And we have to do that before we can attribute dFSCI to it.
If we have to figure out whether it could have evolved before we ascribe dFSCI to it, then what good is dFSCI? It doesn’t tell us anything new, and the only question it answers is this: Could this lengthy gene have arisen through pure random variation without selection? No one on either side thinks it could. It’s a question no one is asking.
So dFSCI answers a question that no one is asking, and it contributes nothing to answering the question we actually are asking, which is: Could this gene have evolved?
dFSCI is a useless concept.
Sorry, gpuccio. I know you won’t like hearing that, but it’s the truth.
If it were not useless they would have done something, anything practical with it by now.
Joe says knowing something was designed enables you to “look at it in a different way”. The sad thing is that is the most detailed use yet given as to the practical application of dFSCI.
That’s fine. It doesn’t change what is happening. You have a bin full of sequences with functional complexity. You keep those that pass your test for incontrovertibly designed sequences, and toss those that don’t pass your test. Of course, those you keep are designed. It’s incontrovertible!
Now you extrapolate your design inference to the sequences you just tossed, including those that have plausible non-design explanations, including those that the vast majority of scientists say have non-design explanations. It’s a faulty extrapolation.
Which puts us right back where we started. Your dFSCI business did nothing to advance your argument. Words, just words.
We have strong evidence of evolution, in vivo, in vitro, in silico.
Your dFSCI is an attempt to sidestep the normal process of science. It doesn’t. Your back to nested hierarchies, fossils, observations of evolution, etc.
Thought it was a notorious marker of evolution.
Did he really say that? We know it’s dFSCI because there’s lots of it?
gpuccio,
It seems that you are no longer following the other thread dedicated to your ideas. I have two comments there to which I’m very interested in your response. The primary question from the first one is:
A more interesting test would be the strings used as solutions to Lizzie’s problem. Let’s say you are given two strings, both of which represent a solution to the problem. One was generated by a human who thought about the problem for a bit and wrote down his best solution. The other is the output of one of the GAs mentioned in that thread. Since both are solutions to the problem, as previously agreed both have high functional complexity.
Do you consider the string created by human thought to have dFSCI and the one generated by the GA to not have dFSCI? Please explain your reasoning.
gpuccio,
I am also still very interested in your answers to my questions about Tierra. It seems to me that Tierra meets your requirements for modeling natural selection.
gpuccio,
From your 559 on the UD thread:
and
You are continuing to use the term “empirical” idiosyncratically. Unless you are claiming to have exhaustively tested all possible evolutionary mechanisms, you are merely asserting your conclusion. In actual fact, you don’t know the provenance of the biological artifacts you claim exhibit dFSCI. That’s why you can make the claim — your definition of dFSCI requires that a mechanism not be known. You haven’t done any empirical testing.
Until you actually do some tests, you should stop using the word “empirical”. It doesn’t describe what you are claiming. You certainly can’t claim 100% accuracy simply by assuming your conclusion.
gpuccio,
From your 583 on the UD thread:
ID is not an explanatory theory. Asserting that an unevidenced, undefined designer with unknown capabilities and unknown motivations performed some unknown actions at some unknown time or times is not an explanation of anything.
Comparing that hand waving to the research represented by hundreds of thousands of peer reviewed papers over the past 150 years is beyond ridiculous.
gpuccio,
From your 584 on the UD thread:
I don’t see it as a shift, simply as keiths pointing out yet another problem with your position. Don’t sell yourself short — you’re actually wrong in multiple different ways!
Wrong and useless are orthagonal. As ID advocates love to point out, biologists have over the decades, proposed many specific explanations that turned out to be wrong. But they served as useful conjectures because they suggested lines of productive research.
Gpuccio’s proposal of a non-material, continuously intervening designer is useless because it is entirely ad hoc.
His denial that his reasoning is circular is simply wrong. Denials repeated a million times do not, by repetition, become correct.
Gpuccio, My argument hasn’t shifted at all. There are (at least) two reasons that dFSCI is useless. One involves circularity and the other doesn’t.
I wrote:
The first part — answering a question that no one is asking — makes dFSCI useless, but it’s not circular.
The second part is the circular part. Gpuccio isn’t convinced that this gene could have evolved. Why? Because it has dFSCI. How does he know it has dFSCI? Because he isn’t convinced that it could have evolved.
dFSCI is nothing more than window dressing. It sounds much more scientific to say “object X is designed because it exhibits dFSCI”, versus saying “object X is designed because a) it couldn’t have been formed by pure random variation without selection, and b) gpuccio isn’t convinced that it could have evolved,” but the meaning is exactly the same.
Mung writes:
Seriously, Mung? All this time you’ve been thinking that T|H is a fraction, with H as the denominator?
P(T|H) is a conditional probability — the probability of T given H.
Too funny.
keiths wrote:
I know he and Marks are still appealing to “conservation of information”, but they no longer seem to be saying anything about conservation of CSI
Yes, I think Dembski has probably backed away from the LCCSI (quietly) although a few years ago he reaffirmed (at UD) his Design Inference argument. That reaffirmation is only supportable if he also has the LCCSI in place, or else uses P(T|H). As you have argued in the latter case the Design Inference adds nothing if P(T|H) itself works.
Dembski and Marks have been pushing a Conservation of Information Law in their recent papers. I think they connect it to theology (“In the beginning was the Word”). However they can’t use that to argue that natural selection doesn’t work — the most they can argue that way is that if it does work it is transferring information that already exists in the shape of the fitness surface. To argue that natural selection doesn’t work you need to use Behe-like arguments as with P(T|H) or use the No Free Lunch Theorem, which also does not work for the case of fitness surfaces that arise in the real world.
P.S. Did you see that Mung quoted Hubert Yockey citing you?
Thanks for pointing that out, I had missed it. It is hard to tell what role that quote plays in Mung’s argument. Perhaps it is supposed to discredit Theobald’s methods by quoting me against Theobald. If that is the intent it misses the mark — Theobald used likelihood and Bayesian methods, not parsimony methods. (I have access to Theobald’s paper on line and checked).
Mung:
No, Mung, I am not saying that (and neither is Dembski). First, H is not a number, as Jerad has patiently been trying to explain to you. Second, read what I wrote in presenting Dembski’s argument:
“Low” does not mean zero.
Mung responds:
Okay, Mung. You think Dembski’s idea is stupid. We get that. Why don’t you fight it out with him, then?
I wrote:
Mung:
It’s not a strawman. You thought H was the denominator of a fraction, and that setting H to 0 would mean dividing by zero.
I wrote:
Mung asks:
Mung, instead of demanding that we educate you about ID, why don’t you read Dembski for yourself? His answer to your question appears at the bottom of page 25:
In the spirit of Lizzie’s rules, while I agree with you that this is humorous, it is a legitimate mistake for someone unfamiliar with statistics.
That being said, now that the error and underlying lack of understanding has been pointed out, Mung does have an obligation to retract his claims (and insults) and educate himself before participating further in the discussion.
What’s more funny to me than the mistake itself is the fact that Mung was so sure of himself that he accused Jerad of being drunk and me of lying, when the actual problem was his own shallow understanding of Dembski.
No matter how often that happens, Mung never learns to pause and consider the possibility that he is wrong before hurling insults at someone.
Mung:
Zachriel:
Mung:
No, you can’t. See this.
Mung, why accuse Zachriel of being “resistant to logic and reason” when he is just trying to teach you some science?
Zachriel:
Mung:
Mung, Zachriel’s comment is a sly dig at KF, whose response to every evolutionary argument seems to be “But you haven’t addressed OOL!” Zachriel is pointing out that your argument undermines KF’s.
Mung:
No, because P(T|H) is not zero, as Jerad and I have explained to you several times.
Who said the ‘probability’ of H is zero?
I wrote:
Mung:
Are you a woman? If so, I apologize for calling you ‘he’, but that’s hardly ‘misogynist’. In fact, how could anything I’ve said about you be construed as misogyny if I thought you were a man?
Yes, and here’s the missing context:
I was expressing Dembski’s view, not my own. As Jerad already explained to you.
It’s an accurate characterization of Dembski’s position. Why would I regret it?
His comment about a “divide by zero” indicates that he still doesn’t understand that P(T|H) is a conditional probability, not a function over a fraction. Methinks he could benefit from Wolfram: http://mathworld.wolfram.com/ConditionalProbability.html
Mung,
How many years have you been involved in the ID debate? It amazes me that in all that time you’ve never bothered to learn anything about Dembski’s CSI argument.
I’ll repeat it yet again: P(T|H) is not zero. Reread what I wrote earlier:
Dembski is saying that we can “eliminate H as the explanation” if χ is greater than 1. But χ is greater than 1 for values of P(T|H) other than 0. So Dembski is saying we can eliminate H as the explanation even if P(T|H) is not zero.
You apparently think that Dembski’s idea is incredibly stupid. That’s fine, but why are you arguing with us? Fight it out with him if you don’t like it.
The symbol between T and H in P(T|H) is not a division sign, but means “given” in a conditional probability.
http://en.wikipedia.org/wiki/Conditional_probability
H is a hypothesis, not a probability. For Dembski, H is usually a uniform probability distribution. We then assess the probability of T given H.
Sure you need to know. Water can melt below 32°F.
Simple science experiment! Throw salt on your icy sidewalk. Or make some homemade ice cream!
Heh. Think we’re trying to make *valid* inferences.
Good for you! On the other hand, we’re wrong a lot of the time.
Indeed, most scientific hypotheses are wrong. A hypothesis isn’t very interesting unless it takes a chance at being wrong.
Meta:
Another thread, On the Circularity of the Argument from Intelligent Design. Now we have two threads on Uncommon Descent and three or more threads on The Skeptical Zone, intertwined in a complex manner. Interesting.
No wonder we’ve gone plural.
Mung, to Jerad:
Mung,
Dembski is arguing that if the probability of something is low enough, we can eliminate it as a possibility, even if its probability is nonzero:
Note his use of the word ‘eliminate’, despite the fact that he is dealing with nonzero probabilities.
I understand that you think Dembski’s idea is ridiculous, stupid, irrational, and contradictory. You’ve made that quite clear. I encourage you send him an email, written in your usual gracious style, informing him of his shortcomings. But why are you arguing with Jerad and me when it’s Dembski’s idea you find ridiculous?
I would appreciate a direct answer to this question: Have you read any of Dembski’s books? If so, which ones?
Also, why do you refuse to read Dembski’s short 41-page paper? It’s available online, it’s free, and it answers the questions you’ve been pestering us with. If you have trouble understanding it, I’d be happy to help you out, but only if you do the work of trying to understand it first.
It’s not my job to spoon-feed ID to you. Read the paper, and if you have specific questions about parts of it that you find confusing, then post them and I’ll see if I can help you.
Plural and zen!
Seems we have evolved into a democracy. Until Lizzie returns we commenters are at liberty to start new threads. If you think another thread will be helpful, I at least would be happy for you to post one.