Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Junior - Crafty NPS Challenge - a user experiment

Author: Robert Hyatt

Date: 07:14:09 11/25/03

Go up one level in this thread


On November 25, 2003 at 08:29:27, Sune Fischer wrote:

>On November 25, 2003 at 07:13:48, Uri Blass wrote:
>
>>On November 25, 2003 at 06:47:16, Sune Fischer wrote:
>>
>>>On November 24, 2003 at 23:18:35, Robert Hyatt wrote:
>>>
>>>>>Whether we get 2% draws or 98% draws says nothing about what happens in the
>>>>>remaining 98% respectively 2% of the games, and that *only that* is what we are
>>>>>interested in.
>>>>
>>>>That's a problem, IMHO.  IE I get sick and lose one set.  Am I _really_
>>>>worse, when we have played 1000 sets all to draws?
>>>
>>>Definitely, but probably bot by a very large margin, however the question isn't
>>>about margins.
>>>
>>>>>>  Particularly since we are dealing with
>>>>>>humans and computers that can "get sick".  Suppose on a normal day we
>>>>>>can only draw, but I get sick and lose 6 in a row.  You conclude you
>>>>>>are better.  You are wrong.  The 1000 draws are much more representative
>>>>>>of how we compare than the 6 wins/losses, in this case.
>>>>>
>>>>>You are mixing up the two question because you feel that being 0.001 better is
>>>>>being equal, and it isn't in a mathematical sense.
>>>>
>>>>If we played at the same level _every_ set, game or match, I'd agree.
>>>
>>>Good, so at least we must be agreeing now as far as the engines go!? :)
>>>
>>>>But
>>>>humans don't do that.  with 1000 draws and 1 win I would _not_ say the person
>>>>with the 1 win is better, in any way...
>>>
>>>Do you think statistics care whether the subjects are humans or computers?
>>>
>>>When you've said A, you must say B. :)
>>>
>>>-S.
>>
>>I think that we can say nothing only based on the results of one match.
>>
>>With deterministic machines and no learning and no book it is possible to get
>>A beats B 100-0
>>B beats C 100-0
>>C beats A 100-0
>
>Yes but that also violates the assumption of independent and stochastic trials.
>
>>The games can be always the same and if you look only in the result of one match
>>you can get the wrong conclusion.
>
>The only conclusion you draw is an estimate of who is better, not a very strong
>conclusion.
>
>The interesting point here is that the following game sequences (assuming the
>assumptions hold):
>
>engineA: ½½½½1½½½½½1½½½½½½½½½1
>engineB: ½½½½0½½½½½0½½½½½½½½½0
>and
>engineA: 111
>engineB: 000
>
>has the same probability of A being better.
>To me that sounds very plausible, logic even.
>
>


Sure, but now what about humans?  those three 0's could be
back-to-back games.  The human could be sick.


-S.
>
>>Uri



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.