Author: Robert Hyatt
Date: 21:04:48 02/02/03
Go up one level in this thread
On February 02, 2003 at 23:22:57, Will Singleton wrote: >On February 02, 2003 at 23:11:21, Brian Richardson wrote: > >>On February 02, 2003 at 21:18:45, Will Singleton wrote: >> >>>Bob indicated that he didn't mind if other comps auto-kib in ch 211. It looked >>>like three or four comps were doing that today. I personally think it's fine, >>>the more the merrier. Sometimes the lines are very different, and thus >>>interesting to compare. >>> >>>Any dissenters? >>> >>>Will >> >>I would also like to see other engine's evals... >> >>Hopefully with a verbosity variable (ala Crafty) to skip output >>until several million nodes have been searched to some >>reasonable depth. >> >>Also, the + always for white vs + for side on move is annoying. >> >>I personally prefer + for side on move, since I tend to think of + >>as "better", rather than having to translate to a color. >> >>I'm not asking anyone to change their orientation, but when kibbing >>it would be helpful to know what a "+" score means. >> >>Brian > >My preference for kibbing on third-party games would be to use + for white, and >- for black. That is how both crafty and amateur do it now. When you have a >program commenting on a game in which it is not participating, then it makes no >sense to switch its orientation depending on the side to move. *That* would be >confusing. > >So, I suggest for ch 211, please kib in white/black orientation. > >Will Here is my take on the +/- stuff... 1. during an "observed" game. I don't see how you can kib anything but +=good for white, -=good for black. Otherwise the questions are continual. And when black is ahead, it is confusing. 2. annotating games. Same idea. What sense would it make to comment on white's 12th move and give a score of +1.2, and then when commenting on black's 12th move and finding that white made a mistake, again giving a score of +1.2, which is now good for black. This would make annotations really tedious to read. 3. During a normal game, in light of the above, I chose to go to +=good for white in all cases. No confusion at all. It did take me a while to get used to tuning in to a Crafty game and seeing a score of -3.5, until I remembered that Crafty was black and that is a _good_ score for black. But now, it is totally natural, and I have no problems with observed games, real games, annotated games, or anything else. I think the key is consistency. And +=good for white is a normal standard. IE wouldn't it be ugly to see this: 36. Rxb2 Rxb2 +- and think white is winning, when +- is given from the side to move instead? Everyone is actually used to +- mean white is winning, so + scores for white winning are pretty "comfortable" for most that are watching... I know many will disagree. But it is hard to argue for "side to move" scores for games a program is not playing in, but commenting on. And if you are going to do that, and also play server games and kibitz/whisper scores, the questions will be non-stop because they see +=good for white today in the DJ vs Kasparov game, and +=good for side to move when your program is playing and everyone is watching. That is why I made the change a few years back, even though it was confusing to _me_ for a while. But now I don't get very many "Is + good for white or for side to move?" any more...
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.