Author: Serge Desmarais
Date: 14:54:06 09/09/98
Go up one level in this thread
On September 09, 1998 at 03:43:21, Dirk Frickenschmidt wrote: >On September 09, 1998 at 00:00:16, Serge Desmarais wrote: > >Hi Serge, > >thanks for clarifying. > >I strongly recommend the implementation of an ICC-option to computer chess >programmers after having made my first weeks of ICC experiences (having played >some more chess.net games before). I think Bob Hyatt and Bruce Moreland (among >others) are completely right to put so much emphasis on getting test results >from (mostly) human playing on ICC. I believe the effort would soon pay for any >programmer, since the implementation of new or changed program code could be >tested faster and better this way, even if you get mostly blitz results. Still I >am convinced that these give you first significant hints, how your program >version is doing and what are crucial points you still have to work on, from >opening to endgame... > >Kind regards >from Dirk > Just that I am not sure about how good/valid blitz games testing can help to improve things? And computers/programs are already better at playing blitz games than about any human, not to speak about bullet games. Personnally, I would like the programmers/companies to organize more matches against good GMs involving only long games (40 moves in 2 hours and then 20 moves an hour). The problem is that the GMs won't play for free (just for the advancement of knowledge/science). And these are expensive. Another way of getting these kinds of long games is by registering the computers/programs in strong tournaments. Only problem is that the FIDE's rules allow any player to refuse to play against machines without being penalized. I would have to check if that rule is still valid. But I must say that I have an account on ICC only for chess programs (Braincan) and it gives an idea (at least about how to compare against other programs, but again I don't have the top hardware (upgrading all the time costs a lot of money!). At least, the games are fun to watch. Serge Desmarais > >>On September 08, 1998 at 11:46:46, Robert Henry Durrett wrote: >> >>>On September 08, 1998 at 03:17:02, Dirk Frickenschmidt wrote: >>> >>>>Hi all of you, >>>> >>>>to get some more impressions of Junior5's blitz strength, I played >>>>some games on ICC against strong human and computer opponents (the >>>>latter often running on much faster hardware). >>>> >>>>I used a simple 200MMX and 24Mb Hash, having to play manually by >>>>switching between Junior and "Blitzin" (the ICC chessboard interface) >>>>in Win 95. As I am a comparably slow switcher (I saw others do that >>>>with considerable speed) I lose at least 3-4 seconds/move for manual >>>>switching, leaving only the rest to Junior, thus slowing the program >>>>down additionally compared to programs answering automatically as well as >>>>against humans who have the whole time just for thinking and giving in the >>>>moves on *one* board. >>> >>><snip> >>> >>>It seems to me that there must be some way for people who ligitimately put their >>>computers on ICC should have some way to avoid this manual process. Do all >>>"computers" on ICC have to put up with this inconvenience? If so, isn't there >>>some way that a computer-savvy person could write some program to do this >>>automatically? Maybe a special version of "Blitzin" could be written which >>>would be intended solely for use with computers. It could interface directly >>>with the computer, removing the human "middle-man." >> >> >>I think it is not possible. The programmers of the commercial chess programs >>would have to implement this as an option. But their main goal is for people to >>play against them/analyse positions or games with them and, with the auto232 >>compatibility, have them play against another program running on a different >>computer. A few have an "engine vs engine" option too. They can also be used as >>a database manager/viewer. But an ICC-option could be interesting too. Some >>programmers have made test version that runs automatically on ICC, while their >>commercial one can't (WChessX is one). The question, for a ICC connection to be >>implemented, should be IF enough users of that program would use/benefit from >>it. Note that Chessmaster 6000 has an Internet connection option, to play >>against other CM 6000 programs, but not on ICC, I think. >> >> >>Serge Desmarais
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.