Good Math, Bad Math

As usual, Casey Luskin over at DI’s media complaints division is playing games, misrepresenting people’s words in order to claim that that they’re misrepresenting IDists words. Nothing like the pot calling the kettle black, eh? This time, he’s accusing Ken Miller of misrepresenting Dembski
in a BBC documentary.

Let’s first take a look at what Casey claims happened in the documentary:

A reporter recently sent me an anti-intelligent design BBC documentary with the outlandish title “A War on Science.” In it, Darwinian biologist Ken Miller is shown purporting to refute irreducible complexity in the bacterial flagellum by citing the type 3 secretory apparatus, giving his usual misrepresentation of irreducible complexity. But it gets incredibly worse. Miller egregiously twists the basic arguments of leading ID theorist, mathematician William Dembski. To paraphrase Miller’s argument (Miller’s exact words are given ***below), when cards are dealt out in a game of poker, the hand you get is unlikely. But obviously that hand wasn’t intelligently designed. Therefore, unlikely and non-designed things happen all the time, so evolution can happen even if it’s unlikely, and we should never infer design. This completely misrepresents Dembski’s arguments.

And now, let’s look at the words that he pulled out of context, and see what was actually said in the documentary. The following are a direct transcript of Miller’s words, in context:

“One of the mathematical tricks employed by intelligent design involves taking the present-day situation and calculating probabilities that at the present would have appeared randomly from events in the past. And the best example I can give is to sit down with 4 friends, shuffle a deck of 52 cards, and deal them out, and keep an exact record of the order in which the cards were dealt. We could then look back and say ‘my goodness, how improbable this is, we could play cards for the rest of our lives and we would never ever deal the cards out in this exact same fashion.’ And you know that’s absolutely correct. Nonetheless, you dealt them out and nonetheless you got the hand that you did.”

Ok. Quick show of hands. Does anyone read the last paragraph, and think that it in any way refers to Bill Demsbki and his nonsensical specified complexity? Anyone? Anyone?

The problem here is that Casey doesn’t get math, at all. He sees the word “probability”, and he has no real idea of what it means, but he knows that something about things being improbable is part of his idol, BillyD’s writings. So that must be what it refers to, right?

Except that it’s remarkable clear that Miller’s argument is not talking about Dembski. It’s talking about other ID babblers like David Berlinski, who make a direct argument that
life must be designed because otherwise it’s too improbable.

Here’s a quick hint for you Casey, my boy…. Before claiming that someone is making a bad
mathematical argument, or twisting one of Billy’s arguments – try checking to make sure that
they’re actually make the same argument that you think they’re making. Because if you’re going to
go around claiming that people are deliberately misrepresenting your point of view, when in fact
they’re making an entirely different argument, and the only reason you didn’t know that is
because you don’t understand either argument… Well, it makes you look like a total
idiot.

Of course, you’re probably used to that.

Comments

  1. #1 HWSoD
    March 22, 2007

    this is just funny.

  2. #2 Jonathan Vos Post
    March 22, 2007

    What is the probability that a total idiot KNOWS that he’s a total idiot? We calculate the a posteriori probability, given that N people have told him he’s an idiot, by Bayes’ Law, as follows…

    [his eyes glaze over]

  3. #3 Orac
    March 23, 2007

    I actually saw that documentary (BitTorrent is a wonderful thing; I encourage everyone to get a hold of this documentary and watch it). If anything, the producers went too easy on Dembski and bent over backward to be fair.

  4. #4 Jonathan Vos Post
    March 23, 2007

    Interesting experiment suggests what a single mutation can do, in the context of an adaptbale nueral system, in extending the capabilities of a mammal’s senses:

    Published online: 22 March 2007; | doi:10.1038/news070319-12
    Mice made to see a rainbow of colours
    All you need to see more is more pigments in the eye.

    http://www.nature.com/news/2007/070319/full/070319-12.html

    Lucy Odling Smee

    Simply by inserting a piece of DNA that codes for a human eye pigment into the genome of a mouse, scientists have introduced a rainbow array of colour to the dull mix of yellows, blues and greys that normally make up a mouse’s visual world.

    This suggests that the mammalian brain is very flexible and can interpret signals not normally encountered. It also hints that just a single genetic mutation could have added reds and greens to the visual palette of our ancestors tens of millions of years ago.

    [truncated]

    … the mouse brain is perfectly capable of deciphering a stream of new information from the eye. “This is a beautiful demonstration that fairly sophisticated processing can arise from a single change at the front end of the visual system,” says Williams.

    The result also hints that perhaps the right kind of photopigment is all that’s needed for humans to get owl-like night vision, or to see ultraviolet colours.

  5. #5 Nelly
    March 23, 2007

    Have you watched the video? It took me all of 2 seconds to search for the film’s title and find it on Google video, and watch it. If you actually watch the movie, the scene appears around time index 37:55, and you can see that in the whole scene Miller is talking is about Dembski. Dembski is even shown rolling dice during the scene, and Miller is obviously talking about Dembski. That obliterates your first point that this isn’t about Dembski. What about your second point that he misinterprets Miller? If Luskin is wrong, then what IS Miller saying???. You don’t tell us. From actually watching this video it seems pretty obvious that Miller’s talking about Dembski and saying that Dembski detects design by finding a merely improbable event. Miller doesn’t need to use the words “specified complexity” for us to get his point.

  6. #6 Mark C. Chu-Carroll
    March 23, 2007

    No, I haven’t watched the video. I’m going by the transcript that Luskin himself provided. And it’s very clear what Miller is saying. He is not talking about specified complexity – he’s talking about the old classic probability argument. You know, the one that I provided a link to in the article?

    Go to any creationist site, and search for the word probability. You’ll get hundreds of arguments – all of which are variations on “Life is so improbable that it couldn’t have happened without the intervention of a designer”.

    Remember that “Specified Complexity” isn’t the only argument that Demsbki ever uses. It’s his favorite, becaue it’s constructed in a way that makes it look impressive. But he’s also come up with things like his “universal probability bound”, where he argues that anything with a probability of less then some number can’t possibly happen. (He’s provided several different numbers for the UPB.)

    Ultimately, even the specified complexity argument reduces to a probability argument – which is why Demsbki invented his UPB. The SC argument ultimately claims that the probability of something posessing the qualities of both specification and probability is *so* improbable that it’s impossible.

    Casey is just, as usual, playing games. Quoting things out of context, to create an excuse for his faux outrage. Compare the section that he quoted – which makes it appear that Miller is talking about SC; then read the entire quote, where it’s clear that he’s *not*, and then try to tell me that Casey’s actually being honest.

  7. #7 Jud
    March 23, 2007

    Let’s go a bit further to deal with the Dembski argument, as quoted by Luskin. The Dembski quote says the key difference between mere complexity and the really special kind is that the special kind “conforms to the right sort of pattern to eliminate chance.” (So the “right sort” of complexity to “eliminate chance” says chance mutations and undirected selection can’t drive speciation. Tautology, anyone?) One of the examples Dembski gives is an arrangement of stones that spells out a coherent English sentence, essentially the “monkeys typing Shakespeare” argument rephrased.

    It seems to me (probability experts in the audience, which I’m definitely not, tell me if I’m wrong) this is simply a special case of the argument that Miller destroys. Instead of the example of dealing any hand of cards, let’s take the special case of being dealt a royal flush. I’m sure the odds are astronomical, though I’d guess it’s happened. But the cards only form a “coherent pattern” because it’s been so designated as part of the game. (After all, if you could win with any old five cards, gambling at poker wouldn’t be much of a challenge, would it?) Particular sequences of letters appear in Shakespeare because millions of people reached prior agreement that these were “words” forming “sentences” in “English.”

    The problem with Dembski’s example was pointed out with admirable simplicity by PZ: “I don’t see perfection around me.” Once one puts aside the mindset that humans are some sort of paragon, the “goal” of speciation (it’s easy to do, think: guys have nipples, people get sick, have wars…), it becomes evident that humans aren’t like some pattern of stones spelling out a coherent sentence (much less Shakespeare), but are simply one of the forms of life that happen to be extant on this particular planet at this particular time.

    Sure, we’re complex (more than is strictly good for us, as anyone with appendicitis would likely agree). But barring the equivalent of God’s message-in-a-bottle that says we’re the output of His manufacturing process (Sal Cordova has postulated that He has left such a message in the form of “DNA steganography,” and is waiting for research to locate it), I can’t think of scientific evidence that would show this complexity is the “right sort” to “eliminate chance.”

  8. #8 tgibbs@bu.edu
    March 23, 2007

    It is certainly applicable to Dembski’s “universal probability bound” argument. Dembski sometimes gives lip service to the fact that probability of evolution by natural selection would have to be calculated along evolutionary pathways, but of course he doesn’t know how to do this (since it would require knowledge of the fitness in the contemporaneous environment, not only of every step along the specified pathway, but also of every other possible evolutionary pathway that could lead to a functionally equivalent outcome). So in the end, to argue that the probability that he cannot calculate must be low, he always seems to fall back on hand-waving permutation arguments that are essentially the same as “proving” the impossibility of poker based upon the astronomically low probability of any given hand.

  9. #9 secondclass
    March 23, 2007

    Nelly, I watched the video, and your statement is false. The way the video is edited and narrated, it looks like Miller is responding to Dembski, but nothing in Miller’s words gives that indication. If I’m wrong on this, please provide a quote from Miller to the contrary.

  10. #10 secondclass
    March 23, 2007

    Furthermore, Mark is correct in stating that Dembski’s argument boils down to a probability calculation. It isn’t the probability of the actual event, but of what he calls in TDI the saturated event. If the probability is less than 1/2, then design is inferred. The “specification” of the actual event plays a role in defining the saturated event, but that’s it. So technically specification isn’t a separate criterion. No matter how you slice it, Dembski’s argument boils down to looking at a probability and saying, “This is too low, therefore design.”

  11. #11 Davis
    March 23, 2007

    …mathematician William Dembski…

    Dembski shouldn’t be allowed to keep using ‘mathematician’ as a title, since last I checked he doesn’t seem to be doing anything remotely math-related these days. PR, obfuscating, and lying certainly don’t constitute mathematics in my book.

  12. #12 Blake Stacey, OM
    March 24, 2007

    Davis:

    I agree. How about we try this on for size: “Professional liar William Dembski. . . .” Has a kind of ring to it.

  13. #13 Jonathan Vos Post
    March 24, 2007

    In the Legal world, some people are referred to snidely as “professional litigants.”

    Someone so accused has three standard answers:

    (1) But my cases all have merit;

    (2) So is my opposition;

    (3) If you really think so, try filing a motion to have my current case (they use the term “instant case”) deemed Frivolous. If you lose, you’ll end up paying me more, anyway…

    I mention this because I see a parallel between ID people and professional litigants. Of course, one thing that they and we don’t agree on, is what is a “law” of Nature?

  14. #14 Jeremy Mohn
    March 25, 2007

    “Here’s a quick hint for you Casey, my boy…. Before claiming that someone is making a bad mathematical argument, or twisting one of Billy’s arguments – try checking to make sure that they’re actually make the same argument that you think they’re making.”

    I did what Casey Luskin didn’t bother to do–I emailed Dr. Miller. His response is here.

    Who wants to calculate the probability of an apology or a retraction?

  15. #15 fnxtr
    March 25, 2007

    part of his idol, BillyD’s writings

    Does this mean Luskin refers to Dembski as “Billy Idol”?

  16. #16 daenku32
    March 25, 2007

    Simply because I remember saying something similar that Miller said (except I did it circa ’02), I would kind of like to ask a general question regarding probabilities..

    I haven’t studied probability theory but it just seems like you definitely need a set, a constant, and a variable. Berlinski’s claims would eliminate the constant from the requirements.
    In the card example, the deck would be the set, whatever the order the cards that are dealt initially would be the constant, and attempts to replicate the constant by subsequent deals would be the variables that then would have the odds counted. If you count the odds prior to the subsequent dealings, then you would simply be doing the initial dealing “virtually” rather than physically.

  17. #17 truth machine
    March 25, 2007

    If Luskin is wrong, then what IS Miller saying???. You don’t tell us.

    Perhaps he assumes that we, unlike Luskin, are able to comprehend simple sentences of English, and don’t need to be told what someone is saying beyond simply quoting him. Consider Luskin’s “paraphrase”:

    But obviously that hand wasn’t intelligently designed. Therefore, unlikely and non-designed things happen all the time, so evolution can happen even if it’s unlikely, and we should never infer design.

    And yet nowhere in text that he is “paraphrasing” are there any instances of the words “intelligent”, “design”, “evolution”, “infer”, “should”, or “never”. What Miller is saying is what he says (see his precise words above), not Luskin’s complete misrepresentation of it.

  18. #18 truth machine
    March 25, 2007

    Oops, he did use the word “never”. Well, I guess that proves that Luskin is right after all, heh heh.

  19. #19 truth machine
    March 25, 2007

    It’s talking about other ID babblers like David Berlinski, who make a direct argument that life must be designed because otherwise it’s too improbable.

    Only indirectly. Specifically, he is talking about the confusion between a priori and a posterior probabilities, which is just one of the mistakes Berlinski makes, but is also made by others who don’t take Berlinski’s line.

  20. #20 Tyrannosaurus
    March 26, 2007

    Over at the PT Jeremy Mohn posted a response Miller emailed to him with permission to post it. In it Miller explains about the BBC video last year and the Disco Institute’s claims. Take a look is worth reading and comes directly from the source.

  21. #21 Noel Grandin
    March 27, 2007

    Hmmm. The problem with this kind of argument is that effectively it says “it happened, and I have an explanation, therefore, it happened the way I said”.
    Which is cute, but doesn’t actually provide any useful information.

    Firstly, probabilistic arguments against evolution are valid because proponents for evolution argue that the process is likely.

    Secondly, we’re not interested in the a posteriori likelihood here. We’re interested in the pure probability of the event happening.

    And yes, in the absence of actually being able to repeat the experiment several times to generate a sample statistic, this is a valid argument.

    Historians and forensic specialists use them all the time.

    Otherwise criminals would go, “yeah, I know it seems weird that this gun just dropped out of the sky into my hand, but honest, that’s the way it happened”.

  22. #22 Mark C. Chu-Carroll
    March 27, 2007

    Noel:

    I wasn’t aware that proponents of evolution argue that abiogenesis is particularly likely. We do argue that it happened, but “likely” and “likely enough to have happened at least once in a galaxy of millions of stars and billions of years” are not the same thing.

    And the thing that I’m arguing against isn’t so much the *concept* of a probabilistic argument; it’s the specific *form* of that probabilistic argument. To make a probabilistic argument, you need to be able to define the component probabilities, and put them together properly. But the probabilistic arguments of folks like Berlinski and Dembski are formed from numbers that are, at best, wild-ass-guesses, and they’re smashed together without regard for whether they’re actually independent or not; and they set arbitrary conditions (like Berklinski’s idiotic “exactly one possible initial self-replicating molecule”, or his “template” requirement) to further skew the numbers; and then they create arbitrary thresholds for “impossibility” (like Dembski’s “Universal Probability Bound”.

  23. #23 Xanthir, FCD
    March 27, 2007

    To expand on Mark’s words, one of the important bits that would make a valid probabilistic argument that is missing from DI arguments is an analysis of the probability of life itself, not of life-as-it-happened-on-Earth.

    This, currently, is impossible, as we don’t have enough understanding of just what is possible in self-replicators in general, which obviously is the requirement for life.

    If you ask what the chance of Earth-life appearing on Earth is, then you’re talking about an a posteriori probability instead, and using it to attempt to infer design is just as bad as calculating the chance of that exact deal of cards.

  24. #24 Torbjörn Larsson
    March 27, 2007

    Firstly, probabilistic arguments against evolution are valid because proponents for evolution argue that the process is likely.

    Biologists who use evolution theory in their daily life couldn’t care less whether abiogenesis is likely or not, since it is totally outside the domain of the theory. Evolution describes how existing life behaves.

    And abiogenesis researchers probably don’t care less whether abiogenesis is likely or not because they are currently interested in how the process could work. In other words, we don’t know enough to calculate a likelihood.

    Those that are really interested in this likelihood are exobiologists and ETI researchers who considers the possibility of life respectively intelligent life in the universe. It would have been nice to have a geological record stretching back to the window of ~ 10^9 years when life started on Earth. Unfortunately we have not.

    So it seems we don’t know how many times life got started until it took hold and survived. But we do know that global extinction events are common, wiping out a large part of species. And we also know that the early solar system underwent some radical changes and had many more bodies zipping around.

    Considering that RNA seems to be a later development, and DNA for certain is so, it is even possible that existing life had several independent progenitors that interacted and mixed before the common genetic code evolved.

    in the absence of actually being able to repeat the experiment several times to generate a sample statistic, this is a valid argument.

    Not really.

    As I noted, abiogenesis researcher seems primarily satisfied when they have shown a possible pathway. That is perhaps the best we can expect them to do, but it is also sufficient knowledge here.

    We will likely never know the historic pathway taken, unless it happens to be unique, which would be a real surprise. (Perhaps you can calculate the probability for that? ;-)

    Exobiologists are looking at a repeated experiment, namely the probability for earth-like life to arise on earth-like worlds. It is really difficult (iow, needs finetuned values) to suppress the possible numbers to not have life arising somewhere in the universe, which btw we have observed. :-)

  25. #25 mgarelick
    March 28, 2007

    To make a probabilistic argument, you need to be able to define the component probabilities, and put them together properly.

    Yes, yes. I think the most glaring vulnerability of Dembski’s EF is that in order to estimate probability you have to know something about what you’re estimating the probability of. Since we know nothing about design, with respect to the natural world, the only way we can draw conclusions from the EF is to give design the status of ~p. But that, I think, amounts to an argument that if the probability of p is low, anything you can imagine can take on the status of ~p.

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.