Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Check with Eduard

Author: Mark Young

Date: 05:09:58 06/26/01

Go up one level in this thread


On June 26, 2001 at 07:14:55, Chessfun wrote:

>On June 26, 2001 at 06:25:16, Mark Young wrote:
>
>>On June 26, 2001 at 00:17:21, Chessfun wrote:
>>
>>>On June 25, 2001 at 22:01:57, Mark Young wrote:
>>>
>>>>On June 25, 2001 at 21:54:07, Mike S. wrote:
>>>>
>>>>>On June 25, 2001 at 18:15:41, Mark Young wrote:
>>>>>
>>>>>>The results are bogus anyway, I can sit at home and win games as he did....Let
>>>>>>me run the computer against Eduard....I bet the results would be much different.
>>>>>>
>>>>>>Why does he not play a 20 game match, the computer will learn what he is doing
>>>>>>and pick a different way of playing against 2.Na3 Then he is toast.
>>>>>
>>>>>What matters then, is the single game (each) with a brilliant win against the
>>>>>program, and not, if the learning feature may avoid repetition (or if some games
>>>>>may be lost beforehand). That's not the point, but that these games can happen
>>>>>at least once on each computer. I would be glad if I were capable of winning
>>>>>such games regularly (I have some, but very few old one's).
>>>>>
>>>>>Furthermore, why call the results bogus, unless you have evidence that these
>>>>>games aren't reproduceable or possible? That's not quite fair IMO.
>>>>
>>>>The point is we are talking about games under tournament conditions, not games
>>>>sitting at home at blitz times, with no controls. Anyone can sit, play with the
>>>>program, and produce games like this, but its not the same when you don't have
>>>>control of the screen, program, and the settings of the program.
>>>
>>>
>>>To me I see a different point.
>>>Try playing a GM 50 times and see how many you'll win.
>>>Forget the time controls for a second as IMO Eduard could easily
>>>repeat this at tournament controls as I feel I also could.
>>>Computers are known for being better at blitz than GM's simply log
>>>onto ICC and have a look. With a computer once you find the path to
>>>the win in most cases the path remains open. Simply play out of book
>>>asap if you win the computer in all liklihood will repeat it's same mistakes.
>>>
>>>Try that against a GM.
>>
>>You tell me a GM who is willing to be exploited like we can the computer
>>programs, and I might be able to produce a draw or a win also.
>>
>>Can I program holes in the human GMs book to let me FOOLS mate him. :)
>
>You can't "fools mate" any current program. Most program books are clearly
>good enough and most strong humans who play them play book lines to about 15
>moves.

Now there are *rules* on how you can exploit the programs....I see.

>
>>The point is sitting at home cooking up ways to exploit the computer is not what
>>I would call tournament conditions.
>
>The same cooking can be done and then played out under tournament conditions.
>
>>BTW: Eduard only posts his wins, and you don't know what his record would be
>>under a fair test playing the computers under fair match or tournament
>>conditions.
>
>Naturally, I would assume that in winning these games first you must play
>a bunch that don't pan out. But the same can be done at longer controls and once
>acheived it will repeat.
>
>Sarah.



This page took 0.01 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.