Author: Christophe Theron
Date: 21:30:34 11/23/00
Go up one level in this thread
On November 23, 2000 at 18:40:01, Bertil Eklund wrote:
>On November 23, 2000 at 18:07:16, Uri Blass wrote:
>
>>On November 23, 2000 at 17:34:30, Bertil Eklund wrote:
>>
>>>On November 23, 2000 at 14:43:57, Christophe Theron wrote:
>>>
>>>>On November 22, 2000 at 16:36:44, Fernando Villegas wrote:
>>>>
>>>>>On November 22, 2000 at 14:09:59, James T. Walker wrote:
>>>>>
>>>>>>On November 22, 2000 at 12:30:31, Fernando Villegas wrote:
>>>>>>
>>>>>>>On November 22, 2000 at 11:51:17, James T. Walker wrote:
>>>>>>>
>>>>>>>>Hello Fernando,
>>>>>>>>I can only tell you there are many problems with all programs. None of them are
>>>>>>>>perfect. As a beta tester you can only report what you find. There are
>>>>>>>>literally hundreds of special positions which many programs do not understand.
>>>>>>>>Tablebases will cure many of these problems but not all. The programmers are
>>>>>>>>aware of many of the shortcomings of their programs. They make the decision if
>>>>>>>>it is important enough to fix or forget. I guess they only have so many hours
>>>>>>>>in a day and they have to decide what is most important at the moment.
>>>>>>>>Jim
>>>>>>>
>>>>>>>OK, Jim, but you must concur with me that B+N mate is not just one problem
>>>>>>>between many, but a clasic kind of position every current program must solve
>>>>>>*****
>>>>>>I agree! I remember when David Kittinger took the B+N mate info out of the
>>>>>>Forte computer (It was in it's predecessor the Super Constellation). I asked
>>>>>>him why and he said he needed the memory for something else. I never bought
>>>>>>another Novag computer after that.
>>>>>>*******
>>>>>>
>>>>>>and
>>>>>>>every tester must look at as a matter of fact. The same with:
>>>>>>>a) king + pawn againts King
>>>>>>>b) King + Rook againts King
>>>>>>>c) King+ pawn in the h or a file
>>>>>>>d) oposition rule
>>>>>>>e) king +queen against king +rook
>>>>>>>... And maybe some more. I think that these kind of situations should be tested
>>>>>>>as a minimal test of proficiency. Should be kind of a rule of testing
>>>>>>>operations.
>>>>>>>Regards
>>>>>>>fernando
>>>>>>
>>>>>>As I said. Tablebases will fix many of these problems. I am a very strong
>>>>>>advocate of tablebases even if their contribution to rating points is minimal.
>>>>>>But I repeat, the final decision as to what a program will or will not do lies
>>>>>>with the programmer-_N_O_T_ the beta testers. Do you believe Christophe does
>>>>>>not know that Chess Tiger will not mate with a Knight & Bishop?
>>>>>>Jim
>>>>>
>>>>>I WANT to believe he just missed it. I do not want to believe he delivered us a
>>>>>so gorgeous program without THAT and with full knowledege. Like to sell a
>>>>>Mercedes without one of the wheels.
>>>>>Fernando
>>>>
>>>>
>>>>
>>>>Yes, that's exactly what I did, Fernando.
>>>>
>>>>I'm sorry, but there are other bits of knowledge that I consider as being more
>>>>essential than KBN/K, and so I started to add those other bits first.
>>>>
>>>>
>>>>
>>>> Christophe
>>>
>>>Hi!
>>>
>>>I haven't seen anyone mention another bug. Gambit don't moves at all after the
>>>opening, start thinking but never moves. I have seen it 3 times after 70-80
>>>games.
>>>
>>>Yes the program is so exciting and fun that I can't complain at anyone of the
>>>beta-testers!
>>>
>>>The only complain against the beta-testers is that only two mentioned that the
>>>Century3 auto-player are extremely hard to get working properly.It only works
>>>well against other DOS-programs.
>>
>>I think that it is old news.
>>Ed told us many monthes ago that Rebel Century has problems with the autoplayer
>>and the problem is going to be solved only when Rebel is going to be a windows
>>program.
>
>The log-file only shows "correct". The problem is that the program stops, don't
>read the moves correctly. If you play against Tiger (white) Rebel quits back to
>DOS.
>
>
>>>>
>>>They could also have told us that the new Century3 is an excellent program that
>>>can compete very well against the other top-programs.
>>
>>It is also old news.
>>There are results of century3 against the top programs in the rebel site.
>>
>>see http://www.rebel.nl/r11-rc3.htm
>
>I don't mistrust his own results. I would prefer results from "independent"
>persons.
>
>There is another well known program that in the past claimed that they won
>against all opposition in game in 15. They presented results of all meetings
>from some hidden source. My and others results couldn't confirm this.
>
>Today they claim to be at least 100 elo better than F6a, without any source at
>all.
>
>Bertil
I can guess who you are talking about. But I think nobody here takes this
seriously.
At least I think that the programmer of this program should not let people say
this.
It's not going to help his reputation.
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.