Author: Robert Hyatt
Date: 15:02:43 08/28/98
Go up one level in this thread
On August 28, 1998 at 16:30:40, Moritz Berger wrote: >On August 28, 1998 at 14:13:32, Mark Young wrote: > >>Like the earlier version of Junior, Junior 5 seems much more adept at playing >>other computer programs. After watching the end of the Yudasin match with Junior >>5, I see the same pattern of behavior I did in Junior 4.6 against humans, very >>strong and solid play then out of the blue a move that self destructs its >>position. The moves are not tactical mistakes, just gross positional blunders. I >>know that all programs have their strengths and weaknesses, but this pattern >>with Junior is very disappointing to see because the program seems so very >>strong in all other respects. > >... you might have a point, but Junior 5 is strong and aggressive enough that no >human could afford to just wait for (rare) positional blunders. Here's a random >sample, the last 3 games I played today with Junior vs. assorted Craftys on ICC. >I like Junior's refreshing style very much. > >Esp. the draw vs. data shows nice king safety terms, especially at Blitz it's >remarkable how well Junior managed to get itself out of Crafty's attack. > I don't mind getting beat up on ICC, because it happens regularly. However, the impression above is not quite "reality". IE if I do a search for crafty vs dustbin, I find the following (limiting it to the last two months which is all that I see until games back in 1997): aics% search +=crafty dustbin Your search found 14 games: [8 0 6 = (white win,draw,black win)] 0 2538 dustbin 2784 crafty 0-1 [ br 5 3] D37 Res 07-Aug-98 1 2893 crafty 2550 dustbin 1-0 [ br 5 3] C63 Res 17-Jul-98 2 2554 dustbin 2889 crafty 0-1 [ br 5 3] C68 Res 17-Jul-98 3 2885 crafty 2558 dustbin 1-0 [ br 5 3] B02 Res 17-Jul-98 aics% search -=crafty dustbin Your search found 13 games: [8 0 5 = (white win,draw,black win)] 0 2570 dustbin 2586 crafty 1-0 [ sr 60 3] C65 Res 19-Aug-98 1 2585 dustbin 2539 crafty 1-0 [ sr 30 3] E67 Mat 03-Aug-98 aics% search crafty dustbin draw aics% Your search found 6 games: [0 6 0 = (white win,draw,black win)] 0 2562 dustbin 2780 crafty = [ br 5 3] C47 Rep 07-Aug-98 My only comment is that what you wrote gives a far different impression than what I see on ICC. Yes "crafty" has more compute power than what you are using. However, that "power" didn't come easily and took a lot of programming effort. I find similar results against singacrafty, and against fitter, which doesn't match with the "impression" you give above. This isn't intended as a "flame" but is a note that your data has some *big* holes in it. Maybe your impression of Junior slightly "jades" your observations and you remember the good wins but tend to forget the losses. Also note that "crafty" isn't usually the "strongest" crafty on ICC. Because "crafty" is the development version and quite frequently has serious bugs because of new changes. try your luck with singacrafty some more, or with fitter. They are hell on wheels...
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.