Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: OT / Re: Upon scientific truth - the nature of information

Author: Dann Corbit

Date: 17:12:36 07/15/00

Go up one level in this thread


On July 15, 2000 at 19:58:38, ShaktiFire wrote:
[snip]
>The chance of throwing 100 heads in a row ( or in fact any unique sequence)
>is = (1/2)**100.  However, perhaps we should be examining not the sequence,
>but the final tallies.  i.e. 43 tails, 57 heads,  irreseptive of sequence.
>
>To do this we use the binomial distribution, which calculates expected
>values, for 101 states starting from 0 heads, 100 tails    to 100 heads,0tails.
>
>I mean the sequences don't matter,,,,the tallied up states do.
>
>I have no idea what I just said.

Actually, I think you are pretty smart, and you make a compelling argument.
However, when you have only a very small number of experimental trials, and each
experiemnt is carried out with different equipment, rules, and conditions and
then you use that to form a judgement, couldn't it be that you got a few
improbable sequences to start with?

Chess players may find problems if they get to play the program one hundred
times or more.  They may exploit the weakness.

Again, I don't know that this will happen.  Only that it is possible, maybe even
probable.

If Kasparov had one thousand practice games against Deep Blue, do you think he
would have done better or worse or the same?




This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.