Author: Robert Hyatt
Date: 11:15:38 08/27/01
Go up one level in this thread
On August 26, 2001 at 16:38:08, Amir Ban wrote: >On August 26, 2001 at 03:58:38, Uri Blass wrote: > >>On August 25, 2001 at 22:48:58, K. Burcham wrote: >> >>> >>> >>> >>> >>>i analysed this game, and the score stayed at 0 most of the game, up until >>> 68. ...Ra8 >>> >>> >>>this is the position one move previous. 68. Ke5 >>> >>>in this position deep junior6 eval is +.22 at depth 24. >>> >>> >>> [D] 3r4/4R2P/6p1/3NKbk1/8/8/8/8 b - - 0 1 >>> >>> >>> >>>now in the game, deep junior7 chooses 68. ...Rd8-a8. >>>this was a six point blunder for deep junior7, and the game is over >>> with this move. >> >>I susect that it is an operator error because Junior7 never considered Ra8 as >>best and has no problem to see +3 for white even at small depthes after Ra8 >> >>Here is some analysis >> >>Junior 7 - Blass,U >>r7/4R2P/6p1/3NKbk1/8/8/8/8 w - - 0 1 >> >>Analysis by Junior 7: >> >>69.Ne3 Bh3 70.Rb7 Re8+ 71.Kd6 Kh6 72.Rb4 >> = (-0.12) Depth: 11 00:00:00 100kN >>69.Nf6! >> = (0.23) Depth: 11 00:00:00 277kN >>69.Nf6! Rh8 70.Re8 Rxh7 71.Nxh7+ Kh6 72.Re7 Bd3 73.Kf6 >> +- (3.15) Depth: 11 00:00:00 512kN >>69.Nf6! Rh8 70.Re8 Rxh7 71.Nxh7+ Kh6 72.Re7 Bd3 73.Kf6 >> +- (3.15) Depth: 11 00:00:00 525kN >> >>(Blass, Tel-aviv 26.08.2001) >> >>Uri > >This is blitz remember (7 minutes/game including operator time), and at move 69 >the engine already shows about 0 sec left and moves instantly. > >The blunder doesn't matter because Junior would have lost on time anyway. > >Blitz is great fun, but don't read science into the results. A lot is up to the >operator, and Goliath owes a lot to its maniacal operator Erdogan Gunes. > >I'm still trying to figure out how to do time division for these conditions. >With 3-4 sec average operator time, the choice is between allowing the engine >zero thinking time and a sure lose of time at move 80. Goliath, for example, was >setup for 3 min/game. > >Amir One reasonable choice: use 3 seconds per move until you have used 60 seconds on _your_ clock. Then back off to 2 seconds for the next 60. Then drop to 1 second and stick there. Then there's the operator time. I lose _zero_ time on my opponent's moves, When he picks up a knight, I type "N", when he puts it down I can type the destination square (Ne5) before he can press the clock. When he hits the clock button, I hit the return (this lets me avoid entering a bad move if he changes his mind before hitting the clock). So nothing lost there. My only loss is in transferring a move from Crafty to the board. I watch the analysis carefully, and am always ready to make the last move it displayed. So that when it beeps, I only have to be sure there was no last-second fail high, and then I quickly move the piece and hit the clock. I used to play humans, using Cray Blitz, set as above (3/2/1 secs per move) using a _real_ chess clock with 5 minutes, and I can only remember losing one game on time out of maybe a hundred played at the last ACM event. I certainly was able to go way beyond 100 moves in every game, at least... The 1 minute off of your clock is also important... as if you ponder correctly you won't burn much. This means that you average 3 seconds per move when you mis-ponder, more when you get it right and wait for the opponent...
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.