Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: I don't buy Nemeth's claims

Author: Uri Blass

Date: 10:40:46 07/14/02

Go up one level in this thread


On July 14, 2002 at 10:01:26, Robert Hyatt wrote:

>On July 14, 2002 at 08:27:47, Uri Blass wrote:
>
>>On July 13, 2002 at 18:15:33, Robert Hyatt wrote:
>>
>>>On July 13, 2002 at 12:46:37, Uri Blass wrote:
>>>
>>>>On July 13, 2002 at 10:47:45, Robert Hyatt wrote:
>>>>
>>>>>On July 12, 2002 at 21:34:43, Telmo Escobar wrote:
>>>>>
>>>>>>On July 12, 2002 at 14:29:17, Louis Fagliano wrote:
>>>>>>
>>>>>>>He seems to have replaced John van der Wiel as the leading computer killer (at
>>>>>>>least on this board).  Rebel managed to overcome van der Wiel's anti-computer
>>>>>>>chess style and beat him in a challenge match.  Since Ed Schroeder has retired
>>>>>>>from playing challenge matches against humans, is there anyone with a copy of
>>>>>>>Rebel Century 4.0 and a fast computer (sorry, mine's only 500 MHz) and would
>>>>>>>want to challenge Mr. Nemeth to a match?
>>>>>>>
>>>>>>
>>>>>> Try to win using his strategies and you'll see.
>>>>>>
>>>>>> I'd like to meet Nemeth in the Fritz 7 server, to have him playing any of the
>>>>>>CB engines I have in my computer.
>>>>>
>>>>>
>>>>>You might be surprised.  There was a player named "mercilous" on ICC that
>>>>>wrecked holy hell on _all_ programs a couple of years ago.
>>>>
>>>>You forget one important thing and that the programs of today have better
>>>>hardware and it make things harder for humans.
>>>>
>>>>Of course you can win after enough tries but this strategy is not going to work
>>>>in most of the cases and GM's are not going to try it when they know that they
>>>>may have maybe probability of 10% to win the machine and 90% to get problems.
>>>>
>>>>Uri
>>>
>>>
>>>Better hardware won't do a thing against "mercilous".  He played 1 0, 2 0 and
>>>3 0 games.  At those time controls, no hardware is fast enough to see that open-
>>>ing the file will get you mated.
>>
>>It is dependent on the position and the software.
>>I am also more interested in slower time control and not in 1 0 or 2 0 games.
>>
>
>So am I.  I don't pay a lot of attention to "blitz" games.  I haven't for
>several years now...
>
>
>
>>I do not like the solution of crafty because it means that the program is not
>>going to capture even in cases when it is good to capture.
>
>It depends.  It has several "qualifications" that must be met.  If the search
>can find a way to break one of the qualifications, it will not make the capture.
>And in cases where the capture is good, it generally means that the opponent
>has simply blundered anyway, if it isn't a real trojan-horse attack.  Not
>capturing will probably not lose the game...
>
>
>
>>
>>I think that a better solution is to do special extensions in positions when the
>>trojan horse sacrifice is relevant.
>
>
>Not so easy.  The capture _starts_ the variation.  Now you have to recognize
>other features (queen on the board, can get to the h-file, program's king has
>castled, opponent's has not, opponent has a rook on the h-file or one can get
>there quickly, etc.  You will be extending a _lot_ even in non-trojan positions.


I agree that it is not easy but if you do the special extensions only in
positions when a move like hxg5 is possible then you are not going to extend a
lot in non trojan positions.

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.