From: Paul Nelson <pnelson2@ix.netcom.com>
>Hi Richard,
>
>You wrote:
>
>> Paul, thanks for the references, but unfortunately I don't have
>>access to a university library. Is any information about this work
>>available online?
>
>I don't know. But if you send me your surface mailing
>address in the UK, I'll mail you copies of the articles,
>along with other relevant materials (e.g., a long unpublished
>paper by Peter Rust dealing with what Dembski later
>called CSI).
Thank you. I'll send you my postal address in a separate email. However,
before incurring the effort and expense of sending them, I suggest you check
with your colleague, Dembski, whether these really are valid applications of
the Design Inference. If they're not, then they are probably of no interest
to me.
>> Also, I wonder why it is that Dembski did not cite these
>> papers in his reply to me on this subject. Perhaps it's because
>> the fit to his Design Inference is not sufficiently close for them
>> to be considered an application of it.
>
>Again, I don't know -- although from many conversations
>with Bill, I know that the Scherer/Rust-type calculations
>figure in his thinking, at least as rough approximations to
>the sort of probabilistic estimates one might attempt with
>the origin of biomolecules.
>
>> What we really need is for Dembski to give his seal of
>> approval to a particular calculation. Until then, we will be
>> left guessing about whether this or that calculation really is
>> a valid application of the Design Inference or not.
>
>You shouldn't need Bill Dembski looking over your
>shoulder to apply TDI. Try it yourself.
Dembski has claimed to have detected ID in nature by means of his Design
Inference. (At least, I think he has--please correct me if I'm wrong about
this.) He therefore has a responsibility to cite the application of the
Design Inference on which this claim is based. Until then, why should we
take the claim seriously?
Furthermore, it's not clear how the DI should be applied to the origin of
biological systems, since all of Dembski's examples deal with much simpler
cases. I want to see the details of the DI in action. What exactly are the
relevant chance hypotheses? How have other chance hypotheses been excluded?
What is the specification? What is the "side information"? I doubt I'll find
the answers to these questions in the papers you've cited, since they
weren't written with the DI in mind. I'm not impressed at being told, in
effect, here's a method, here are some simple examples, now go off and apply
the method yourself to the much more complicated case in which we are really
interested!
In any case, the method of the Design Inference is unclear. Until Dembski
clarifies it, it isn't possible for anyone else to know whether they are
applying it correctly. If you disagree with me, and claim that the method is
clear, then you should have no difficulty in answering the following
questions. I have already sent them to Dembski, and am awaiting his reply.
Perhaps, as his colleague in the CSRC, you could encourage him to expedite
his reply. ;-)
[start quote]
1. In TDI, you make a distinction between "design" and "intelligent agency".
The method you name the Design Inference only detects "design", and you give
no additional criterion for distinguishing "intelligent agency" from mere
"design". It seems to me, therefore, that one of the following must be the
case:
(a) A conclusion of of "intelligent agency" follows automatically from the
detection of "design"; or
(b) Your method cannot detect "intelligent agency".
Could you please clarify which of these is the case, or else explain what
additional criterion is used to distinguish "intelligent agency" from mere
"design". Could you also please confirm that "intelligent agency" refers to
the same concept as the more commonly used term "intelligent design".
2. To make an inference of design, the probability of a specified event (E)
must be shown to be small under "all the relevant chance hypotheses that
could be responsible for E" (TDI p. 50). TDI doesn't contain a clear
definition of the term "chance hypothesis", but it seems to me that the
hypothesis of evolution by random mutation and natural selection is a
relevant chance hypothesis for the origin of any entity which is claimed to
have evolved. In fact, you imply as much in the final paragraph of your
article "Explaining Specified Complexity" (Meta 139, 1999/09/13). However,
Wesley Elsberry, and most others with whom I've discussed this, claim that
this does not qualify as a chance hypothesis. Would you please settle this
argument by saying which of us is correct.
3. In addition to the Design Inference, you also describe an Explanatory
Filter. Would I be right in thinking that the EF is just the procedural
application of the Design Inference, and so must conform to the logic of the
DI? The two are not alternative methods, correct?
4. In TDI, you define the Design Inference in terms of probabilities, but
elsewhere you talk of specified complexity and Complex Specified Information
(CSI). It seems to me that your method for detecting specified complexity or
CSI is exactly the same as application of the Design Inference as described
in TDI, with only the nominal difference that the probability bounds and
calculated probabalities are transformed by taking the logarithm to base 2
and negating, i.e. I = -log2(P). Could you please confirm that this is
correct, or else explain what the difference is between these methods.
[end quote]
By the way, in a response to FMAJ1019@aol.com, you mentioned that "The
magnetic patterns on your hard drive exist in nature." Perhaps I should
clarify that, by ID in nature, I mean ID in the origin of biological systems
(excluding of course those with which mankind has tampered). The presence of
ID in man-made objects is not what we are interested in here.
>Here's a puzzle for you. Every e-mail message you've
>sent me is an example of CSI. The messages are real
>physical patterns, encoded on magnetic media (although
>they might have been encoded in sand, brick, ink, God
>knows what). What natural cause accounts for these
>patterns?
>
>Or is it reasonable for me to infer an agent, namely
>Richard Wein, at a computer somewhere in the UK?
This is the sort of simple example which Dembski restricts himself to in
TDI. When applied to biological systems, the issue is far less clear.
Nevertheless, even this case is problematical. I think you would have
difficulty formulating and justifying a specification in this case. Perhaps
you would care to try. I consider Dembski's approach to specification to be
fatally flawed. But, if you can come up with a convincing specification for
this case, you may be able to persuade me that I'm wrong!
Richard Wein (Tich)
This archive was generated by hypermail 2b29 : Fri Sep 29 2000 - 17:50:55 EDT