Author: John Merlino
Date: 18:37:04 02/13/01
Go up one level in this thread
On February 13, 2001 at 21:04:43, Stephen A. Boak wrote: >I ran a Game Analysis after the patch (thanks, John Merlino) and found the >following changes compared to CM6000 (I don't have CM7000): > >After each actual move played, instead of showing CM8000's suggested 'best move' >with corresponding score and PV for that best move, CM8000 seems to show the >'best reply' to the actual move played that it is analyzing; however, I'm not >sure of that interpretation and request a full explanation (thanks). > >All programs I own (CM6000, Fritz6, Rebel 9, 10 & Century) show as analysis, >after each played move, a program-suggested best move (which may be same or >different than the actual move being analyzed), best line of play expected after >the suggested move, and score for the suggested best move. > >Some programs I own (Fritz6 and several Rebel versions I have) also show a score >for the actual move played (for comparison with the score for the suggested best >move). CM6000 never showed this, so to see if there was a significant score >difference that might indicate the current move played was not best, I always >compared the current suggested best move score with the best move score shown >after the subsequent ply. > >An example from the CM8000 Game Analysis I ran last night shows that some other >method of showing move analyses/scores may be in use: > >8. d3 ... ;Score: 0.32 8... Bg6 9. Nc3 Be7 10. > ;e4 O-O 11. Qe2 Qa5 12. e5 Ng4 13. cxd5 > ;cxd5 14. a3 (7:06) > ... Bxf3 ;Score: 0.35 9. exf3 Be7 10. f4 O-O > ;11. Nd2 Qb6 12. a3 Rfe8 13. b4 c5 14. > ;cxd5 exd5 (13:02) > >It is obvious that 8... Bg6 is not a suggested best 8th move for White. > >A. It may be the calculated best move for Black in reply to White's actually >played 14. Bg2. > >B. Or it may be part of the remaining moves in a suggested best line of play, >where the suggested 'best move' has been [accidentally?] omitted from the >analysis line? > >What is CM8000's suggested best 8th move for Black? It is obviously not the >shown White move 8. exf3 in the analysis line give after the 8...Bxf3 actual >move. > >C. It may be the 8...Bg6 move, shown in the analysis after the previous move >(White's 8th move). > >D. It may be 8...Bxf3 (agreeing with the text), but it is not shown, for some >reason, in the analysis line. > >QUESTIONS: >1. Were these changes already incorporated in CM7000? >2. Why the changes (are they intentional)? >3. Are they changes to match some perceived standard (and if so, what is that >standard or who determined it)? >4. Is the 0.35 score, shown above, the score for the actual move played, >8...Bxf3, or is it the score that would have resulted if the best move >(apparently 8...Bg6) had been played? > >I happen to like what I am used to--the analysis for a given actual move shows >the computer's suggested best move, line & score, so the user can directly >compare the suggested best move with the actual move played. But, for CM6000, >to determine if a played move was close in score to the suggested best move, I >always looked to the subsequent ply and suggested move/score for comparison. > >If my guess is right (but I'm not sure), CM8000 analysis at a give ply now shows >the suggested best followup line of play (followup to the actually played move). > The effect is that the score shown at each analysis is the correct one for the >move actually made, but that one has to look to the preceeding ply and analysis >to see if the played move is indeed the suggested best one. > >In either scenario (the CM6000 one or the current CM8000 one) one needs to >examine two successive plies of analysis to obtain 'complete' information about >whether the best move was played and what the score difference might be if the >best move was not played. So philosophically the two methods may be >'equivalent', in a sense. > >Thanks in advance for explaining what is going on (whether it be intentional or >unintentional), and any philosophy of the CM8000 development team. > >--Steve In a nutshell, the analysis DOES show "best play" from the CURRENT position in the game. So: 8. d3 ... ;Score: 0.32 8... Bg6 9. Nc3 Be7 10. ;e4 O-O 11. Qe2 Qa5 12. e5 Ng4 13. cxd5 ;cxd5 14. a3 (7:06) ... Bxf3 ;Score: 0.35 9. exf3 Be7 10. f4 O-O ;11. Nd2 Qb6 12. a3 Rfe8 13. b4 c5 14. ;cxd5 exd5 (13:02) means that after White's 8th move is played, "best play" is 8...Bg6 9.Nc3 etc., and this line is evaluated at 0.32 for White. Since Black (instead) played 8...Bxf3 (theoretically NOT the best move, which was 8...Bg6), you will notice that the eval grew in White's favor from 0.32 to 0.35. To answer your questions (as best as I can): 1) No. CM7000 (and previous) did it by showing what "best play" WAS for the current position. In other words, for White's 8th move, the analysis would start "8. d3 Bg6". This was more confusing because if the first move in the best play line did not agree with the actual move in the game, it was more difficult for (our typical) users to work back one move, and then look at the best line from a position that isn't represented on the board. 2) It was decided that, of all possible ways of displaying the data, this was the least confusing. The line can simply be described as "best play" from the current position. 3) No standard. It was just decided this was the best way to do it. It also corresponds exactly to engine analysis output, which made it easier to program. 4) The score that is shown IS for the actual move played, but it also assumes the "best play" line as well. So, to be specific, it is the score for 8...Bxf3. You're correct. It can be confusing either way, depending on HOW you want to look at it. But, combined with the move list in the Game Status window, it was felt that the CM8000 way was better overall. I hope that answers your questions. If not, please feel free to ask more! jm
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.