Author: Christophe Theron
Date: 17:50:05 05/01/00
Go up one level in this thread
On May 01, 2000 at 18:44:04, Robert Hyatt wrote:
>On May 01, 2000 at 10:11:51, Mogens Larsen wrote:
>
>>On May 01, 2000 at 09:32:07, Robert Hyatt wrote:
>>
>>>It isn't like the Nunn test itself doesn't skew results, remember. IE the
>>>opening book of the engine is eliminated, which means all the book learning
>>>of the engine is eliminated. Now other forms of learning are eliminated. This
>>>doesn't exactly sound like a "useful test". IE would it be interesting to take
>>>two automobiles, remove one spark plug from each, drain the transmission on
>>>both cars... (wait, one car doesn't have a transmission, just a direct drive,
>>>oh well, drain the other) and so forth.
>>
>>I agree with you most of the way, even though I'm trying to be less
>>confrontational (fat chance) :o). What is your take on the one or two machine
>>issue with ponder off. Will the autoplay feature (and various other differences
>>from one to two computers) of two computers with ponder on, skew the comparison
>>with one machine and ponder off? And if yes, how significant would it be? If
>>not, be gentle about my ignorance :o).
>
>I believe that ponder=on and ponder=off are two different things. I don't
>know how it will affect the results, because I don't spend any time testing
>against anyone with ponder=off. But I do know that ponder=off is _not_ tested
>very well in my code because I consider it pointless to use it except for
>testing/debugging. If it affects others like it does me, then perhaps the
>games are 'valid'. If they have tested ponder=off to any degree, then the
>tests are not valid.
Apparently, the way Crafty plays with ponder=off is not that bad at all. I mean,
maybe you consider that you didn't test it, but it could be that after all
Crafty is not more handicapped by ponder=off than other chess programs...
>>I must admit that the purpose of the test is starting to allude me as well, but
>>maybe I'm just not paying attention.
>>
>>Best wishes...
>>Mogens
>
>
>I think a lot of the 'testing' is all about trying to prove that Crafty can't
>play chess. IE you start out with a conclusion, then you try to find a way to
>make that conclusion happen...
Bob, I don't see this experiment as trying that Crafty cannot play chess. I find
it interesting because for the first time results of games played with and
without pondering, and at various time controls, are published.
It does not really matter that Crafty is part of the experiment.
Look again at the results, Bob. In my opinion, the ponder=off results show that
Crafty's search scales as well as Fritz search with longer time controls.
That is a very interesting point, not something against you.
Christophe
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.