You accused me, earlier, of being a fan of Fisher, and claimed that Dembski rejected Fisher.

This is false. Dembski does not reject Fisher, and indeed, as I pointed out, his entire CSI concept depends on Fisher, as commented on by Meyer. If you read (or re-read) this paper by Dembski:

http://www.designinference.com/documents/2005.06.Specification.pdf

you will find that Dembski considers an alternative to Fisherian hypothesis testing – Bayesian testing – and rejects it. His rejection of “non-design” as a null hypothesis, and therefore of his “design inference” is entirely Fisherian.

You ask, rhetorically: “Hypothesis? What hypothesis?” I answer: the hypothesis that ID is true, or alternatively, that the Theory of Evolution is true. However, I will modify my statement – I do in fact think that trying to estimate the probability that a hypothesis is true is quite fruitful in many circumstances, but not in the case of the hypotheses above. Here is why:

In Fisherian hypothesis testing we do not, and Dembski does not, compute the probability of a hypothesis being true. Using Fisherian statistics, we compute the probability of our data being observed, if our hypothesis is false. This is what Dembski does. And he concludes that the probability of observing what we do, if design if false (non-design is true), given “the probabilistic resources of the universe”, is negligible. Therefore he infers “design”.

Note that he does not compute the probability that non-design is true, not does he compute the probability that design is true.

Those are not what Fisherian statistics allow us to compute, and Fisherian hypothesis testing is what Dembski does.

As you rightly point out, Fisherian hypothesis testing is severely flawed. One simple flaw Dembski rectifies – that of deciding on an appropriate alpha value. Far greater (tbh, I think the issue of an appropriate alpha is the least of the problems with classical null hypothesis testing) is the problem that it doesn’t give us the probabilities we actually want.

Like you, I would like to be able to compute the probability that my hypothesis is true, not the probability that I will observe the data I do observe if my hypothesis is false. And we can actually do this provided we can supply a prior probability. We can ask: “what is the probability of our hypothesis being true, given data D”? But the answer, using Bayes’ theorem, requires us to provide a prior – our prior belief (i.e. before examining the data) that our hypothesis is true.

It is therefore a useless approach for estimating the probability that either evolution or design is true, because we have no frequency data on which to estimate our degree of belief, although it is an extremely useful approach, for example, for computing the probability that the hypothesis “this person has cancer” is true.

That is why I reject it as an approach, not for any hypothesis but for the hypotheses that Design, or The Theory of Evolution are true.

As a way of testing hypotheses derived from the theory of evolution, however, it’s pretty good.

But Fisher won’t serve either. Using Fisher, all Dembski has done is to show that the patterns we observe in biology are prohibitively unlikely under the null hypothesis of “Chance”, where “Chance” appears to mean “spontaneously assembled itself into one of a small subset of an astronomically large number of possible patterns under circumstances in which every other pattern was equally likely”.

I agree, and every biologist agrees, that Chance, so modeled, does not account for biology.

That does not allow us to reject every other hypothesis except design. It certainly does not allow us to reject Darwinian evolution. That is Dembski’s mistake.

“You ask: “How would you calculate or assess the realistic odds of naturalistic evolution? …” And I have answered: I wouldn’t. I have told you that I do not consider that such a calculation is an appropriate approach to hypothesis testing, even if it were possible, which it is not.”

Hypothesis? What hypothesis?

So, that seems to be end of this debate, at least for now. And the conclusion is quite simple — the naturalist atheistic evolutionists are determined to ignore any practical results of modern science, namely of evolutionary computing science of the last 100 years or so, especially as ALL the results clearly go against their expectations and wishful thinking — practically proving that their pet theory of Epicurean or Lucretian random unguided processes is utter irrational nonsense!

It is only too sad that many good Christians still get duped by such blatantly primitive theories and empty hypotheses.

Atoms swerving by mere chance, Kolmogorov complexity, algorithmic probability, or randomly assembled genes? No, mere pipe dreams. Enough said.

“Everything that depends on the action of nature is by nature as good as it can be, and similarly everything that depends on art or any rational cause, and especially if it depends on the best of all causes. To entrust to chance what is greatest and most noble would be a very defective arrangement.” (Aristotle, Nicomachean Ethics, I, 1099b22)

“As you undoubtedly must know, the null hypothesis was another “design” of eugenicist Fisher, and your beef with Dembski is about another such statistics trick of your pet science author Fisher, namely, I presume, with Dembski’s statement, such as in his Design Inference, (section 6.4) where Dembski is critical of Fisher — “This is the conceptual difficulty that to this day remains unresolved in Ronald Fisher’s theory of statistical significance testing.” This keeps looking more and more like another serious flaw or bad purposeful design in Fisher’s crumbling statistics methodology. I am sure future will tell.”

I’m no enthusiast for Fisherian statistics, but Dembski is. So is Stephen Meyer, who cites Dembski’s enthusiasm enthusiastically in his “The Signature in the Cell” (pp 179 and following). Dembski’s whole CSI notion is based on Fisherian statistics:

http://www.designinference.com/documents/2005.06.Specification.pdf

That’s one of the many things wrong with it.

I agree with you that the Fisherian approach has had its day in the sun, and while it is still the workhorse of everyday statistical testing, is conceptually flawed, well beyond the single flaw Dembski identifies, namely the arbitrariness of deciding on an appropriate alpha value (typically 2 sigma for biological sciences, 5 sigma for physics).

You ask: “How would you calculate or assess the realistic odds of naturalistic evolution? Do you have any of your own calculated numbers, estimates or guesses? Or at least what order of magnitude would such a probability have to be for you to make it a truly rational choice for you?”

And I have answered: I wouldn’t. I have told you that I do not consider that such a calculation is an appropriate approach to hypothesis testing, even if it were possible, which it is not.

Regarding the OoL discussion that ensued: To a great degree, probability arguments miss the mark on OoL issues for the intractible reason that such calculations can only (by their very nature) deal with the physicist’s reformulation of the term “information”, often referred to as “physical information”. But life (very observationally) does not gain its biofunction from physical information, it comes instead from recorded information – which in turn creates physcial entailments that must be satisfied in order for such information to even exist.

The search for purely material answers to the OoL therefore must take into account what is repeatedly and coherently observed to the true, and begin to propose mechanisms which can satisfy those physcial entailments.

http://www.uncommondescent.com/intelligent-design/upright-biped-replies-to-dr-moran-on-information/

To be fair, Dembski’s point about “specification” is that to have “CSI”, a pattern has not only to have low frequency (“high complexity”) but it also has to have “compressibility”, and you then compute the number of equally or more compressible patterns out of the total possible number of patterns to compute the probability of one of those patterns turning up.

That is fatally flawed too, but not simply because each one of a large number of possible patterns is highly improbable.

And yet, it happened!

What you are doing is precisely the typical runaround and self-contradictory double-speak the opponents of naturalistic evolution have been getting from the evolutionists for the last 100 years. I asked you some pretty direct questions and you gave some pretty clear answers — it’s all there for others to re-read.

The topic of this thread is the universal scale and how the magnitude of numbers used in evolutionary computing reflects on the rationality of those who believe in such numbers. I am arguing that these numbers or probabilities are precisely how a rational person ought to determine the rationality of his/her belief — we do it in everything else from economics, politics, medicine, legal system, other sciences, etc. — it is the numbers that prove a new drug, that convict a tax cheat, a speeding or a drunk driver, or an election fraudster.

As you undoubtedly must know, the null hypothesis was another “design” of eugenicist Fisher, and your beef with Dembski is about another such statistics trick of your pet science author Fisher, namely, I presume, with Dembski’s statement, such as in his Design Inference, (section 6.4) where Dembski is critical of Fisher — “This is the conceptual difficulty that to this day remains unresolved in Ronald Fisher’s theory of statistical significance testing.” This keeps looking more and more like another serious flaw or bad purposeful design in Fisher’s crumbling statistics methodology. I am sure future will tell.

Anyway, as important as it may be to argue the correct philosophy of science and statistics methodology, (especially if there is a suspicion that there may have been a deliberate foul design built into it), such arguments get pretty arcane to the point of boring and unimportant for all the non-specialists and for the host of average Joes and Janes out there. (I am sure most people, even scientists, cannot even follow Dembski’s criticism of Fisher in his Design Inference.)

But, luckily, we do have also the numbers, and yes they have been calculated and recalculated and accepted, and they seem to fit that data more or less, and we know today that no single realistic self-organizing or self-adaptive evolutionary algorithm has been found despite serious effort and some heavy-duty computing of the last century and decades. The massive failure of all these evolutionary programs, in conjunction with the estimated mind-numbing probabilities in biology and genetics is what should be able to convince even the Joes and Janes, especially when they have a nice universal scale model which will put everything into the right “perspective” of what such numbers with huge exponents really represent!

So reflecting on your last sentence:

“My point is that I do not think those numbers represent the quantities that those who calculated them seem to think they represent.”

Assuming that you have now admitted that you know about these numbers, and since you seem to know better what these numbers do or do not represent than the very experts who have suggested them, let me ask the same questions again:

How would you calculate or assess the realistic odds of naturalistic evolution? Do you have any of your own calculated numbers, estimates or guesses? Or at least what order of magnitude would such a probability have to be for you to make it a truly rational choice for you?

You say:

“Your logic of accepting naturalistic evolution is outright ridiculous! This is not how any rational science-savvy person would or should accept any findings of science — just because you disagree with somebody who proposes a theory and happens to criticize your pet belief… you conclude that that ” is why evolution works” !?”

But that’s not how I accept the findings of science! You seem to have read into my post something that was not there. I accept the findings of science as all scientists do: provisionally, and only to the extent that they fit the data.

“Since you don’t know any numbers, a quick illustration of what they were dealing with ought to suffice — just to spontaneously evolve a bacterium the odds suggested by experts were in the order of 1 in 10 to the power of 40,000! (Yes, that is the exponent of forty thousand!) And the same numerically mind-numbing mysteries are found everywhere in biology and genetics.”

I didn’t say that I didn’t know any numbers. I know quite a lot of numbers that have been calculated.

My point is that I do not think those numbers represent the quantities that those who calculated them seem to think they represent.

the part of the paper of Shallit and Elsberry that I found particularly damaging was the long mathematical discussion of information and complexity, which seems to show that Dembski used these terms in inconsistent ways and made statements about them that are in some cases unsupported and in some other cases false. I see nothing in the article by Luskin that addresses these criticisms.

But why hasn’t Dembski himself issued a detailed point-by-point answer to Shallit and Elsberry, especially their criticisms about information and complexity? In the world of scientific research, if one’s technical papers are publicly criticized as being invalid, inconsistent, incoherent, etc., by competent people giving precise and detailed arguments, it is regarded as practically a duty to respond. It doesn’t do to say that the criticized papers are several years old and in the meantime you have advanced beyond those earlier papers. If the older papers have serious mistakes, those must be explicitly acknowledged in an erratum, or in subsequent papers. Have any of Dembski’s writings since the Shallit-Elsberry article appeared admitted that any of the technical criticisms therein are correct, or else rebutted them? Again, I am talking about the technical mathematical criticisms.

There are times that a scientist wouldn’t bother to issue a retraction or erratum — for example, if he makes a relatively minor mistake in some paper that no one has paid any attention to. I have never heard of a case in my field, however, where a scientist has been publicly accused by other experts of serious mistakes in work of importance and where he did not respond in detail. A physicist who did that would lose all credibility. What would happen is that experts would not even bother to read his papers any more. I assume that it is the same in all scholarly fields.

Anyway, why are you and Luskin defending this work, rather than the author of it? And why defend it here? What is required by scholarly standards is that Dembski respond (a) himself, (b) to Shallit and Elsberry, (c) in detail. Issuing what amount to press releases in blogs and comment boxes may be good politics, but it does not serious scholarship make.

If Shallit and Elsberry are wrong, then it would be an excellent opportunity for Dembski to gain scholarly credibility by showing this publicly and in detail. He could use it as a “teaching moment”.

This is between Dembski and Shallit & Elsberry. I would love to sit back with my popcorn and watch them argue this out. But it looks like Shallit and Elsberry by default.

“No, I have no calculated numbers, estimates or guesses. There is no way of calculating such a thing, which is why ID arguments…”

Your logic of accepting naturalistic evolution is outright ridiculous! This is not how any rational science-savvy person would or should accept any findings of science — just because you disagree with somebody who proposes a theory and happens to criticize your pet belief… you conclude that that ” is why evolution works” !?

Anyway, you are dead wrong and quite ignorant assuming that there are no such numbers and calculations. The whole history of the 20th century Darwinism was a search for such numbers and therefore for acceptance and rigorous justification, and there are plenty of failed models and research programs and numbers to choose from.

Starting with the numbers of the British eugenicist Ronald Fisher, who so wonderfully “designed” or tweaked the methods of acceptable statistics in favor of evolution, or so he hoped, that most modern science research is now simply “false” and unrepeatable. I guess you have never heard about the so called “decline effect” in science?

http://www.newyorker.com/reporting/2010/12/13/101213fa_fact_lehrer?currentPage=all

Anyway, there is a plethora of evolutionary computing programs which tried to prove random evolution, from childish attempts of mathematical half-wits like Monod or Dawkins to some serious programs, like the famous Santa Fe Institute evolutionary programs, with big names like Gell-Mann and many other prestigious heavy-weights and computing gurus.

Since you don’t know any numbers, a quick illustration of what they were dealing with ought to suffice — just to spontaneously evolve a bacterium the odds suggested by experts were in the order of 1 in 10 to the power of 40,000! (Yes, that is the exponent of forty thousand!) And the same numerically mind-numbing mysteries are found everywhere in biology and genetics.