Re: [asa] Re: ID without specifying the intelligence?

Ted has written a very good response to Peter from the historical
perspective, but I'd like to add to this with a practical answer - leaving
aside the label "Methodological Naturalism", and just consider what is
scientifically useful.

Ted wrote:

| The problem is that non-natural causes are "science stoppers."

I would put it differently from that, in saying that appeals to what I would
call Universal Explanatory Mechanisms ( I'll give it an acronym UEM to make
it sound plausible), is a "science stopper", in that once the appeal to the
UEM is made, nothing else scientifically useful can proceed from it, in
particular no further predictions can be made from the theory.

Here's a concrete example for Peter. I know Peter is a programmer, so I
hope others on the list not so inclined will forgive the indulgence.

Suppose I came up to you and said I'd tossed a coin 100 times and had
produced the following sequence of heads and tails:

hthththththththththththththththththththththththththththththththththththththththththththththththththt

Now, I'm guessing that if you examined this sequence carefully, you'd
immediately spot the pattern. And if you were a C programmer, you might
reason that you could produce this sequence as output from a very short C
program, rather than tossing a sequence of coins. The program might look
like this:

main(){for(int i=0;i<50;i++)printf("ht");}

The first thing that strikes you is that the string of letters that make up
the program is considerably shorter than the sequence of heads and tails it
generates. ( A string of 42 characters produces an output 100 characters
long).

The question you might reasonably ask yourself is "Just how simulated
sequences of 100 coin tosses could be generated by a C program of length 42
characters or less?".

The answer is really not a lot. I can see immediately how to do all heads
or all tails. There might be a way of squeezing it down to get repetitions
of sequences of 3,4 or 5. An extremely ingenious C programmer might find a
dozen different ones in that length.

But that's it, for sure no more than 100. Now how many sequences are
possible for coin-tossing. Clearly 2^100 or around 10^30.

So it would be reasonable to assume that there is something pretty
remarkable that I tossed a coin 100 times and came up with one of the
handful or so sequences that could be generated by a 42 character long C
program.

You would be scientifically justified in concluding that I had lied to you
when I said I tossed the coin.

Now, suppose the next day I told you I'd tossed a coin 100 times and had got
the following sequence:

hhthhhthhhttttthhthhthhthhhtttththhththhhtththththhthtthhhhttttthhtttthttttthttttthhhhhhthththttttth

Now you would look at the sequence and be unable to detect any pattern. But
you might note that the entire sequence could be reproduced by the following
C program:

main()
{
   char c[] =
"hhthhhthhhttttthhthhthhthhhtttththhththhhtththththhthtthhhhttttthhtttthttttthttttthhhhhhthththttttth";
   for(int i=0;i<100;i++)
       printf(c[i]);
}

Now note that the string of characters of the C program is LONGER than the
string itself. In fact I think most people, whether C programmers or not
would see that the program contains an embedded copy of the sequence of
heads and tails and then proceeds slavishly to print them all out, one at a
time.

How many different 100 coin-toss sequences could a C program of this length
(about 150 characters) produce. Well clearly such a class of C programs
(150 characters long or more) could produce any of them - just substitute
the desired sequence in line 3.

So the fact that such a program exists that can replicate my sequence of
heads and tails is of no scientific interest whatsoever.

The difference is that the second class of C programs (with length 150
characters) is a UEM - Univeral Explanatory Mechanism; they can reproduce
any sequence.

So there's nothing remarkable in the existence of such a program. Nor does
it have any use scientifically. There is nothing in the program that will
tell me how the sequence might continue.

By contrast, the 42 character C program is clearly not a UEM for 100
coin-toss sequences - it can only reproduce a handful of them - and
furthermore they have predictive value - it's not hard to figure out from
the algorithm how the sequence might continue.

So in my second sequence, you've really no way to deduce that I was lying
when I told you I got it from tossing a coin 100 times.

To get to the theological point here, which I've already made about
"revelation" (formerly of a female alien biologist), if I REVEALED to you
some instructions - to download a trial version of MATLAB from
www.mathworks.com and then to type in the 57 character string:

rand('state',42); x='ht'; disp(x( round(rand(1,100))+1));

from the command line, then you'd have a pretty convincing demonstration
that I'd lied when I said I had got the sequence by tossing a coin. (Again
because 57 character Matlab commands aren't a UEM for 100 coin toss
sequences.

Other examples of UEM's that have all cropped up in the last few days are:

A sufficiently intelligent alien to design us.
A multiverse with sufficient numbers of them that some unexplained highly
unlikely event is made much more probable.
An omnipotent creator.

All of these UEMs are possible, but the appeal to them to explain something
is scientifically not valid. Leaving aside whether it's "methodological
naturalism" it's more a question of "is the explanation any use
scientifically"?

For a model to be any use to a scientist it must be able to explain only a
limited subset of the possible datasets. Any further confidence in the
existence of your UEM must come from revelation, not from scientific
theories.

Iain

To unsubscribe, send a message to majordomo@calvin.edu with
"unsubscribe asa" (no quotes) as the body of the message.
Received on Thu Sep 13 14:11:14 2007