Conflicting Definitions of “Specified” in ID

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.

261 Replies to “Conflicting Definitions of “Specified” in ID”

  1. keiths


    dFSCI – the term being defined

    d – digital
    F – Functional
    S – Sequence
    C – Complexity

    So dFSCI – a sequence of digital symbols exhibiting some minimal level of functional complexity (or something like that).

    I think I got that right, gpuccio?

    Now how is that definition circular? I honestly do not see it.

    No, Mung, it’s “digital functionally specified complex information”, and the acronym is not the definition.

    If you want to know what dFSCI is, why not read one of the many comments in which gpuccio defines it?

    If you want to know why it’s circular, why not read one of the comments in which Mark, Zachriel and I explain that fact?

    Why do you expect other people to do your work for you?

    You’ve called people ‘morons’, ‘idiots’, and ‘liars’ when you don’t even understand the topics being discussed.  Do you understand how foolish you’ve made yourself appear?

  2. Alan Fox

    Understand your exasperation completely but could I suggest that comments that contain both substantive arguments and personal remarks could be split into separate comments and the personal stuff could go into the sandbox thread.

    I do it too and I don’t have the excuse of not being here from the start. Not that there’s anything wrong with being a latecomer! 

  3. Joe Felsenstein

    gpuccio has now given (comment #579 in the Jerad thread at UD) a summary of the argument. There are a few notable features:

    1. The compressibility criterion:
    Much of the debate in this thread has been about Dembski’s apparent connection between specification and compressibility. I have already explained why I think that compressibility can certainly be used as specification, but is higly unwarranted for design detection because of its low complexity. 

    As Dembski had stated that compressibility was only one of the possible ways to make a specification (in No Free Lunch),  gpuccio can still argue that his definitions are compatible with Dembski’s. (Elsewhere, as in his 2005 online paper on Specification, Dembski made compressibility more central).

    2. gpuccio then makes “function” the criterion for defining specification. He lets this be any kind of function that the observer can define objectively.

    In a genetic algorithm model, could I define specification as fitness? I suspect not. gpuccio earlier said all genetic algorithm models were front-loaded with Complex Specified Information by virtue of the fact that they had been designed (by us) to have the genotypes be able to reproduce. And that any high degree of function (fitness) that occurs in such a model is already there.

    In fact, in a genetic algorithm model those high values of fitness are not properties of the population at the start.

    Anyway we’ve been over that argument and gpuccio has ruled out GAs.

    3. OK, so why is it that gpuccio thinks that high amounts of “function” cannot arise by natural selection?

    f) Don’t ask me why only conscious agents can empirically generate dFSCI. I don’t know why. I have some theories, but I really don’t know. It is rather simple to imagine why non conscious systems cannot empirically generate it. But it is really difficult to understand how consciousness can bypass the difficulty.

    So we are not to ask gpuccio why he thinks his argument works. It’s an “empirical” argument. And we are not to try to see whether it works in a GA. The GA is argued (on insufficient and erroneous grounds) to already have the FCSI present.

    From the remainder of the arguments here, I suspect that in any model we might make to see whether dFCSI can arise by modeled natural selection, gpuccio would say that any that did arise is not what gpuccio meant by dFCSI, precisely because it did arise in that model, and that means that it is ruled out as being dFCSI. Hence the allegations that gpuccio’s argument is circular.

    It sounds very much as if gpuccio’s argument is that dFCSI can be recognized if there is an extreme amount of function, but that all models that might investigate the claim that it can only arise through conscious design are ruled out as showing dFCSI.

    I disagree that those models already show the high function. GA models refute gpuccio’s supposed empirical observation that high function cannot arise except by conscious design. And ruling them out is arbitrary and not a valid argument.

  4. onlooker

    Anyway we’ve been over that argument and gpuccio has ruled out GAs.

    Indeed, but he has done so without answering a number of open questions regarding how he justifies ruling them out.  He has particularly studiously avoided any commentary on Tierra, which seems to meet his requirements for modeling natural selection.


  5. keiths

    Jerad writes:

    That’s what keiths is saying. He says: you much be sure that T could not have arisen via Darwinian/natural/unguided processes before you decide, for sure, that T did not arise via Darwinian processes. That P(T and H) = 0 which makes P(T|H) = 0.


    No, that’s not what I’m saying. I’m saying that according to Dembski, we are entitled to rule out unguided processes as an explanation even if the probability is not zero. The probability has to be low — lower than what Dembski calls the “universal probability bound” — but it does not have to be zero.

    When I say that according to Dembski,

    You have to know that something couldn’t have evolved before you attribute CSI to it.

    …I’m saying that the probability has to be below the “universal probability bound”. Mung would understand this if he or she would bother to read Dembski.

    By the way, Mung, you labeled me a ‘misogynist’ for criticizing you the other night. That implies either that you are a woman or that your comment was a complete non-sequitur. (Actually it was a non-sequitur either way, because criticizing a woman hardly amounts to misogyny). Would you like to inform us of your gender, so that we can use the correct pronouns when referring to you? If so, let us know. Otherwise I’ll return to using the generic ‘he’.

  6. keiths


    In case you missed it, my comments have been about keiths’ misrepresentation of Dembski and your follow-on attempts to rescue him from his foolishness.

    He’s claiming you have to know P(H) = 0 before you can make a design inference, which is just absurd.


    You can repeat that a thousand times and it still won’t be true. See my comment to Jerad above.

    Why don’t you read Dembski’s 41-page paper and save yourself some future embarrassment? (Unfortunately, it’s too late for you to retract your current statements.)

  7. keiths


    It’s kind of hard to recognize this as a joke…
    liar! misogynist!

    …when the rest of the thread is full of comments from you like
    You are either ignorant, or a liar, or both.
    You got me. Only a complete moron would say that.
    And today’s Junk for Brains winner is, onlooker!

    It’s like when a convicted pedophile makes a joke about molesting children. It’s not funny, and nobody listening is sure that it’s really a joke.

  8. Toronto

    Mung: “So now you see we can only represent 4 possible states, not 8, or fully half the information capacity, with just that one added aspect of ‘necessity.’ “

    What you have just said is that the “improbability” of a string N bits long is NOT necessarily 2**N, but could be much less due to a “necessity mechanism”.

    This makes KF’s “islands of function” more likely to be “reached” than IDists have previously claimed.

    gpuccio won’t like this either. 🙂


  9. keiths


    I think there’s only one of his [Dembski’s] books I haven’t read. and it’s on my Kindle. That book, for your information, is The End of Christianity.

    Not only have I read his books, I’ve read numerous books he’s only edited or contributed to. I’ve met him in person. I’ll put my library up against yours anytime.

    What’s your point?

    My point is that I’ve never encountered someone who had read most of Dembski’s books yet didn’t understand the concept of the universal probability bound, and how Dembski uses it to rule out events having small but non-zero probabilities.

    Perhaps you need to read those books a couple more times. Better yet, start with this paper and keep rereading until you get it. If you have trouble with it, post your questions and I’ll see if I can help you understand.

    I’ll put my library up against yours anytime.

    Despite knowing nothing about my library? What an odd thing to say.

  10. Zachriel Zachriel

    Mung: And here I thought you had to divide T by H to get P.

    Jerad: Generally the P(T|H) is going to be very, very small but not 0.

    Mung: For the probability to be very very small implies that we’ve divided the numerator by the denominator to arrive at the value. 

    Mung: But we can calculate P(H), right? 

    Not really. H is not a single value, but a chance hypothesis, such as a uniform probability distribution.  We always try to give the best interpretation to what people write {so if someone writes P(H)=0, we would read this as a probability distribution that always returns zero}; but you seemed to be confused on P(T|H), just as you were on the freezing temperature of ice. 

    Mung: Moreover, there is no logical reason why some other process should not be able to generate dFSCI. Let’s call that hypothetical process “X”…

    I know no such “X”…

    It’s all empirical, scientific.


  11. keiths


    That final quote comes from gpuccio, not Mung. 

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.