Bubba

The way that I see it the only real intellectually honest question to ask about Dembski is how much longer he'll be the idol of the Christian right before people get bored and move on.

Eventually people will have to see that NFL doesn't provide any real evidence for anything. Plusm to even try to understand it you have to think so only a minority of creationists will try to use his theorums anyways. I figure it takes about 20 or 30 years before creationists are just plain so embarassed about an arguement that they won't use it anymore. I figure maybe a good 5-7 years before people start really giving up on NFL.

Bubba

theyeti

That's right, it was a personal communication. Sorry if I confused anybody. However, he does make a similar coment in NFL, basically to the effect that the math takes precedence over the biology. Of course he doesn't seem to realize that math is only so good as it correctly models reality. If I can find the quote, I'll post it.

Oh yeah, Langan's a fool.

theyeti

Principia

Actually, in NFL (you know what I mean...), Dembski seems to take a complete 180:

p. 219, No Free Lunch

Let me see if I get the argument:[*]Don't know if T-urf13 is CSI[*]Don't have the "information trail" that documents the displacement of CSI[*]Don't have to trust mathematics over biology reflexively[*]Conclusion: T-urf13 might have come about via chance and necessity and poses no challenge to CSI

[The intermediate steps are left as an exercise for the reader, of course]


[ November 16, 2002: Message edited by: Principia ]</p>

theyeti

Here we go:

I think it's on pg. 219. It's not quite as bad as the other quote, but it's telling nonetheless.

theyeti

P.S. Here's my take on Dembski's statement:

Dembski has defined his CSI as something that can’t be generated by natural means, so therefore the mathematical inability for CSI to form naturally flows reflexively from its definition, and not from any potential data input. The whole point of mathematical models in science is to try to simplify what we see in reality so that we can improve our ability to make predictions and test hypotheses. If a model does not match what we see going on in nature, then we scrap it, not because that model is necessarily wrong, mathematically speaking. We scrap it because it’s irrelevant to nature.

Added in edit: Whoops, looks like I cross-posted with Principia.

[ November 16, 2002: Message edited by: theyeti ]</p>

Valentine Pontifex

Well Dembski thinks his team is in the NFL when in reality Baylor's football team would crush them 91-0 if he got lucky.

theyeti

I'll just go ahead and post my entire thoughts on Dembski's section on T-urf13. This was written as part of a larger article, which has not yet been finished (or even started to a significant degree). So forgive the odd references. (Dembski's words are in quote tags.)

First of all, it must be pointed out that we need not only consider “homologous” proteins, but rather we must consider every sequence within the entire protein sequence space that can perform the given function, even those that are radically different from URF13. This will certainly include many more than just those that are “homologous”, and Dembski errs by neglecting them. (Dembski seems to be misusing the word “homologous” here. Homologous proteins are those whose sequences are so highly similar that they’re thought to be derived from a common ancestor. As noted elsewhere, proteins can be arranged into extended families whose members often differ by 50% or more in amino acid sequence and yet carry out similar (but not always identical) functions. The term really isn’t applicable to proteins formed de novo (unless they incorporate already functional domains), only to their varying descendants. We will assume that Dembski simply means those proteins that are highly similar in sequence and structure to URF13, and therefore carry out an identical function). But more importantly, what Dembski has done in the above passages is to show that the formation of a novel functional protein need not be improbable at all! The very same reasons that he uses to declare that URF13 does not contain CSI can be applied to most any other protein. The ease with which a functional protein can be generated, even with more or less random sequences, shows us quite clearly that there’s no reason to think that any protein contains CSI. For example, Keefe and Szostak (2001) screened a protein library of 6x10^12 random sequences and found four novel ATP binding proteins that are unlike each other or anything else found thus far in nature. Just how many more kinds of functional sequences that library contained is anybody’s guess, but Keefe and Szostak conclude that about 1 in 10^11 proteins with randomly generated sequences will be functional, and their study only used 80mers. This beats Dembski’s universal probability bound by 139 orders of magnitude. This also brings up an important point: a random sequence need not perform a specific function as Dembski seems to assume, rather it must perform some function in order to be preserved as an addition to functional complexity. So Dembski concludes:

To interject briefly, Dembski knows good and well that CSI can’t be generated because he has defined it that way. If we see, out in nature, that new proteins are evolving (and we do) then it must mean that it wasn’t all that improbable in the first place, and therefore not CSI. So all he has to do is to explain why it wasn’t improbable and he’s ducked the bullet, but it totally begs the question of whether or not CSI exists in nature. Trying to “reduce the improbability” is exactly what he does next.

In this way he’s made a preemptive strike against any and all such cases of observed and inferred novel protein evolution (which will undoubtedly be increasing rapidly in the coming years with new molecular techniques). No matter what we find, he can a) claim that it wasn’t too improbable, and hence not CSI, or b) claim that it was “stealing” CSI from some inscrutable time in the past. Occurs within a cellular context? No joke, all evolution within the last 3.5 billion years has been within a cellular context. One wonders at this point if CSI can be detected at all, and if there’s any amount of evolution that would be considered CSI.

For example, Dembski insists that the protein must be made from random sequences, and that the incorporation of previously functional domains or other non-random sequences is “stealing” information from the past. Never mind that the addition of a novel function to a living system, regardless of how achieved by nature, would be considered an increase in “information” by any reasonable definition (not that you find many reasonable definitions among the creationist crowd). It doesn’t count as CSI according to Dembski, because CSI is not about functional complexity, it’s about probability. Unfortunately for him, the vast majority of novel proteins have almost certainly originated at least in part from the duplication, fusion, and/or shuffling of already functional protein domains (though often with random sequences thrown in as well). Not CSI you say? Well then, scratch off the overwhelming majority of known proteins. If CSI exists anywhere in nature, it’s doing a darned good job of hiding itself. Perhaps most telling is the final “blow” against this inconvenient counter-example:

I trust mathematics just fine, it’s the mathematician that’s suspect. Declaring that empirical reality is to take a back seat to a mathematical model is perhaps the ultimate preemptive strike. It doesn’t matter what we find in nature, Dembski has insulated himself by declaring that his CSI criterion is untouchable, and can’t be disproven by any amount of research. It should be clear by now why this is so; Dembski has defined his CSI as something that can’t be generated by natural means, so therefore the mathematical inability for CSI to form naturally flows reflexively from its definition, and not from any potential data input. The whole point of mathematical models in science is to try to simplify what we see in reality so that we can improve our ability to make predictions and test hypotheses. If a model does not match what we see going on in nature, then we scrap it, not because that model is necessarily wrong, mathematically speaking. We scrap it because it’s irrelevant to nature.

theyeti

KC

In that regard, Dembski and Langan are very much alike.

Cheers,

KC

[ November 18, 2002: Message edited by: KCdgw ]</p>