Saturday, September 29, 2007

Creationism is like gremlins bowling in your attic.

This is one of the best Qs of the Q&A. I was REALLY proud of this young lady-- Not only was she well spoken, but she got Dembski to say some funny stuff :)

Young lady-- How is positing a supernatural cause or a designing intelligence reducing, rather than widening, escalating levels of improbability?

Dembski-- Weh, whe, well, when you say 'escalating levels of improbability', youre using word magic there. A designer is inherently improbable...?

Young lady-- Youre leaving the laws of natural science behind and positing an intelligent designer. To me that seems a lot more highly improbable...

Dembski-- Youre using improbability as some sort of subjective belief.

Young lady-- In relation to what youre saying, it seems just as improbable that there is some supernatural being directing evolution than to say maybe we dont understand everything just yet, but perhaps there are biological causes.

Dembski-- Well, I mean, your throwing these probabilities around. *rabble rabble*

Young lady-- So how do you even up with a probability of an intelligent designer? Dont you have to weigh the two probabilities?

Dembski-- Why does that probability even come in?

*audience laughs in disbelief*

Young lady, also in disbelief-- You have to know the probability of a designer if youre going to weigh the two probabilities!!

Dembski-- Whe whe whe we can do a Bayesian analysis! Ahhh, your and my probabilities of a designer are very different. So it would be better to have a methodology that didnt require probabilities for a designer.

*audience laughs again*

Dembski-- Weh we can do a Bayesian sort of thing, given to me it is infinitesimally small I mean, you know, I mean... Here is a standard example.. weeh ehhh beyhh... we need to do this for the sake of the audience. Eeeh you hear some sound up in the attic, you know? Sounds like gremlins, ehh, bowling. Bowling. Gremlins bowling. Thats ahh, would be ahhh, you know, that would explain it. Eh eh, highly unlikely that you know. If there are gremlins up there, bowling, then that would explain the sound. But whats the prior likelihood of there being gremlins? Its highly improbable. So even though if there were gremlins it would explain that, the high improbability that there are gremlins means dont give that explanation a second thought. I think thats how the design hypothesis works for you. For others it doesnt work that way. Okay? So in a sense, if were going to decide this issue, its not going to be on the basis of assigning some sort of prior probability of design. You have to look at the actual improbabilities of this evolutionary system. The thing is, nobody that I see has been trying to do that sort of move. I mean, Richard Dawkins does not try to look at the prior improbability of god, hes trying to say "Look, its highly probable you get these systems because when you do the analysis, Darwinism is a strategy that can climb Mount Improbable", thats his whole point!

I think some of you more familiar with Dembskis 'math' will have fun with this exchange. But Im not, so Im going to let you all play with that aspect of this Q.
  1. What I liked about this exchange is that it once again demonstrates that IDC is a negative argument. THEY dont have to provide probabilities of a designer. They just have to declare an evolutionary system impossible and declare victory.
  2. Dembski compared the probability of his designer to the probability a noise in your attic is caused by gremlins bowling. I dont think I could have said it better myself.
  3. Richard Dawkins does indeed address the improbability of a designer. Page 113, 'The God Delusion.' He addresses the very retardation Dembski "eeehed" and "wehed" and "beyed" about the misuse of 'improbabilities' by Creationists.
  4. ID Creationists are not eloquent. That last schpeal by Dembski was one of the dumbest things Ive heard escape the lips of a 'professional.' "Bayesian sort of thing", ugh.


deadman932 said...

Gack...Gremlins Bowling = ID Designer...brain stunned, neocortex rebelling...frontal lobes recoiling in horror.

Ah... relief found in sheer fuckin' laughter.

That is just a goddamn delight to read, ERV. Cheers, Deadman_932

The DHC said...

a fellow critic of Dumbski, sweet baby jesus!. you should check out my latest post, i feature him in it. you'll be satisfied with my pettiness

Anonymous said...

I see the Double-Standard, Yellow-Bellied, Fork-Tongued Warbler has come home to roost.

Gremlins in the attic. Sheesh.

Basically, Dumbski just admitted that unless you accept a designer (i.e. the Xain God) as an axiom, you can't deduce a designer from the evidence.

He just denied his own postulate.

Anonymous said...

I support Dembski till the gremlin thing--we live in a 1 of 1 universe, so probabilities in some way don't make sense. On the other hand, the good lady has a point about the scope of new questions/explanations that enter when you add in a deity. What's the usual phrase? It fails on... parsimony? But. That'd be too much for him to admit eh? You only worry about silly stuff like that for real science.

In any case though, that's not supposed to be fatal to his argument, because he has independent evidence for a designer, remember?? This is where he's supposed to whip it out.

Anonymous said...

What a great question, and a delicious moment seeing Dembski stumble. For once, I'd like to ask a similar question using IDiot terminology: wouldn't the creator be irreducibly complex?

Unknown said...

You really need to cut D some slack. He's turning into an emotional wreck by the sounds of things.

The poor guy is doing the best he can; not much admittedly, but he's hardly been given much to work with. I mean, there he is, on stage, doing his best to defend a position that's just totally wrong, with no evidence, no scientific evidence, warped logic and no facts to speak of and yet he still somehow manages to form a sentence that resembled an argument.

It's not really surprising that the argument is nonsense, his speech collapses all the time and he keeps contradicting himself.

He has to lie to himself just to keep his sanity, because if he didn't lie to himself he's realised he'd wasted his entire life.

In a way, that makes him a coward. A brave man can accept that he's been wrong and move on, and true scientists encourage and value that type of braver. A coward will rather fight to the death even if it means taking his entire company with him to oblivion just to make sure he doesn't lose face by backing down.

Just as an officer who tries to save face above all else will lead an army to a military disaster, so will Dembski lead creationism to public ridicule.

slybird said...

ERV, you skewerings of Dembski have been a delight to read. Keep it up!


ERV said...

xzanron-- Im gonna do a "LEAVE DEMBSKI ALONE!!!" YouTube video.

Im gonna do it.

Glad you all are liking the posts, but this is alllll Dembski.

Hes a piece of work.


Anonymous said...

Dembski accusing others of "throwing probabilities around"... that's rich.

Fred Ross said...

He not only falls into the Bayesian fallacy, he revels in it. How disturbing.

Bayesian statistical procedures are incredibly useful in engineering and in the abstract mathematical problems of statistics (the set of Bayesian procedures associated with all possible priors contains all admissible statistical procedures, so you can often reduce a problem of analyzing all possible procedures to one of characterizing a function space), and there has been a lot of work on things like noninformative priors to push it further, but the moment anyone who isn't a mathematician starts talking about a Bayesian analysis, it's probably time to turn their microphone off.

Brian Seitzman said...

Wait... I'm unclear... he wants to do some kind of Bayesian analysis without assigning a prior probability? Analysis is necessarily subjective? I don't even get what point he's trying to make. It's just words without meaning. I don't even understand what he's trying to analyze, or what he's saying Dawkins isn't analyzing. Bayesian analysis is used all the time in phylogenies to construct most likely trees... he sounds like he thinks you can just do an analysis of the entire history of life as a single data set. I guess that's possible in theory, given a large enough data set, but the computing power necessary would be bigger than anything we have now, and it's certainly not something Dembski could do with a pencil and calculator in his copious free time.

Meh. I don't know why anyone takes him seriously in the first place unless all they read are his own crapulous "articles."

ERV said...

I might have missed a word or two, but thats what he said.

A Bayesian sort of thing.


Ian said...

As a professional mind reader (and someone who has wasted time thinking about what it would take to assign prior probabilities to God), I think where Dembski wanted to go would have been to end by saying that his explanatory filter was a way around having to assign probabilities to miracles. I think he was trying to set the questioner up, lead her into a trap where he could make fun of her. When she didn't follow him down that path he got frustrated and ended up getting caught up in his own trap.

The end result is extremely revealing. In essence, he says that if you already believe in G(od)remlins, you won't find miracles to be nearly as improbable as you would if you don't already believe in G(od)remlins. Quite in keeping, I suppose, with his fantasy of pushing biology out of science departments.

Bob O'Hara said...

Oh what fun wMad is. It's ironic that he claims that one can't use a Bayesian approach to design, because he claimed you can for deciding if a tomb was Jesus'. Read the Bayesian bit of the argument - it's hilarious. Basically he says that the evidence is irrelevant, because he never believed it in the first place.


Sander said...

@Mike o'risal:
Prior probability in flimsy theist bayesian statistics is probably 50%

Also see this article by victor stenger

Anonymous said...

I would like to see a calculation of the probability of - oh, let's say, a bacterial flagellum - if there is an "intelligent designer".

An intelligent designer which is capable of more things than are allowed by the laws of nature. (It doesn't have to be omnipotent.)

My estimate of a comparison between the probabilities of "it just happened by chance" and "it was done by an intelligent designer", is that it is less probable that an intelligent designer would do this thing, out of all of the possibilities open to it, than that it happened in the more limited range of things that natural laws can do.

(If the intelligent designer is omnipotent, then it can do an infinite number of things, and the probability that it would do bacterial flagella is zero.)

Anonymous said...

If we're going to be all Bayesian about it, then it seems clear that Dembski and his ilk assign a prior prob. of 1.0 to the hypothesis that (their exact version of) god exists. As any good Bayesian knows, once you do that, then any and all evidence is irrelevant since it cannot work to budge your prior prob. away from 1.0. So, Behe (for example) was telling the honest truth in Dover when he said that the mountain of research on the flagellum is irrelevant to his conclusion that the flagellum is irreducibly complex.

Bob O'Hara said...

It just struck me that he doesn't even need to assign a prior to the probability of a designer. All he needs to do is to calculate the Bayes factor.

Oh, but wait. Then you have to work out the probability of getting a flagellum if it was designed.


The Factician said...

Folks, you're all missing the most important point in this post:

Im gonna do a "LEAVE DEMBSKI ALONE!!!" YouTube video.

That would be awesome. I would LOVE to see that.

Torbjörn Larsson said...

It fails on... parsimony?

Yes. But while parsimony can be expressed in bayesian terms in specific cases, the idea that you should express probabilities for a designer (to compare with a natural explanation) isn't something you would normally do. But Dembski has his explanatory filter where he compares probabilities for rare processes, so he is vulnerable if it is highly improbable.

The surprising thing is that he was so vulnerable to the discussion in general. Bob O'Hara has made quite a find where Dembski use bayesian arguments. I get the feeling that one could get creationists to think over their arguments by pressing them on their ID - Improbable Design.

After all, evolution is a process where each step has a definite probability (predictive mutation and fixation rates). While all they have is an Improbable Complexity design that go *poof*, and how likely is that?

Bayesian analysis is used all the time in phylogenies to construct most likely trees...

It has also been used to construct likely cosmological models with respect to parsimony (number of parameters) vs expressive power (fit).

Torbjörn Larsson said...

we live in a 1 of 1 universe

Not necessarily, but the thing with multiverse theories is that probabilities still don't make sense. :-)

I.e., you can see why finetuning isn't necessarily a problem but a natural solution. And you can find certain ways of doing statistics that makes sense, such as on specific observers ("causal diamonds"). But there is no agreement on how to do it.

OT, but multiverses is in the process of taking a hit AFAIU, which is mildly interesting in this context.

A recent paper in string theory seems to show that eternal inflation (which is the standard model for multiverses) aren't compatible with string theory. You can have inflationary spacetimes, but they won't have time to create an eternally inflating part.

And another paper finds that common string models doesn't give inflation. This could mean that string theory is wrong as inflation is more or less a tested model. But IMHO it is more probably a way to select a few models from the vast landscape. This is something that string theory is urgently looking for, and with model selection becoming more probable multiverse selection becomes less probable.

Neither of these problems makes away with multiverse ideas AFAIU, universes can still inflate one by one. But if we use the bayesian analysis so dreaded by Dembski, multiverses has in all likelihood :-P became less attractive and less powerful as an explanation.

AIGBusted said...

Bill Dumb is so funny!

Anonymous said...

Gremlins bowling in the attic.

MAN. That's fucking hilarious. Please give this guy a microphone and a hostile audience more often. This is the new Pygmies + Dwarfs.

The Rev. Jenner J. Hull said...

If I had taken Dembski's "Gremlin" rebuttal and passed it off as an improv in one of my college theater classes, I would've been seen as a genius.

And, as I've just discovered, it can be ably employed as a scat-style, beat-poetry-reading over a suitable jazz instrumental.

Anonymous said...

I think Dembski might have been refering to a paper by Elliott Sober, "The Design Argument".

Check it out, its the 1st link on his "papers" page:

Torbjörn Larsson said...

G, do you mean because Sober prefers a bayesian analysis?

Sober's analysis involves an unnecessarily complex model.

The biological design argument part is equivalent to Elsberry's et al powerful critique of Demsbki's EF filter. (On Talk reasons, I believe.)

And Jeffery & Ikeda has done a more comprehensive critique of the cosmological design (finetuning) argument based on conventional likelihoods without priors and the making the comparison the Young lady asked for, removing Dembski's pitiful out. (On NCSE site, I believe.) [So perhaps it was better if Dembski didn't use the sharper critique. :-P]

Anonymous said...

This gremlins thing is not original. In the wonderfully nutty comic strip "The Masked Cucumber" (by Mandryka), the eponym hero hears a rumble in the middle of the night, goes up to his attic, and finds a bunch of elephants playing bowling. When he asks them what elephants like them are doing in his attic at 4 am, they look at each other in amazement, burts out laughing and tell him he must have a bee in his bonnet for telling this kind of story.

Oh well, you can draw any conclusion you want.

By the way, I am absolutely disgusted by the way Dembski uses "Bayesian" when he just means "some kind of calculation that you probably won't understand, and that involves apriori probabilities I'm totally free to choose as I please".

Anonymous said...

Your "buddy" just "notpologized"... see links at ATBC and Pharyngula:;act=ST;f=14;t=1274;st=20010

Anonymous said...


He admitted the probability of a designer is personal, i.e., subjective. (Perilously close to admitting that the designer exists only inside the head, not in the real universe.)

He admitted that he's starting from the assumption of a designer, not ending with the conclusion of a designer.

He admitted that, to a rational person, the probability of a designer existing is comparable to the probability of gremlins in the attic.

He admitted he has no interest in analyzing his assumption that a designer exists.

He admitted that Dawkins view is that biologists have analyzed the evidence for evolution and find it highly probable.

How is it again that he thinks that ID is a "science?"

Anonymous said...

I think "Gremlins in the Attic" is a great name for a band.

Anonymous said...

Yes, I read that you don't have audio or video yet. I for one would love to hear or see this. Is there anyway you could get us audio or video or know who could?


Anonymous said...

Dembski is clueless. Bayesian subjectivity is about incomplete information; one can't substitute arbitrary subjective opinion, such as a personal belief in God/Gremlins, for the missing information.

Monado said...

So, on hearing hoofbeats in the forest, Creationists just naturally think, "Unicorns!" and expect the rest of as to go along with their Wild-Assed Guess. (pun intended)