The game mentioned in the title ended when the computer malfunctioned. The evaluation had reached +9, or 9 points for White. I copied the game before the computer shut down, and uploaded the partial game to chess dot com. The last two moves are 66. g3+ Ke5 67. Bb3 Ne7 . In the endgame White has 10 points and Black has 9 points; it looks like a difficult endgame to win, and Stockfish at depth=59 was taking a long time to play. The partial game is available from the URL: https://www.chess.com/analysis/library/4abAFQc2tW?tab=analysis .
The chess dot com accuracy scores are 98.6 for White and 96.5 for Black, which is high for a game ending in a “clearly winning” position (according to Stockfish) . The chess dot com game review shows a mistake for Black at move 61… Bg7? . I’m running a 6-hour per side match from the position after 61. Rc1 to see whether the result is decisive and what the accuracy scores will be.
I’ve been wondering whether there was some luck involved in the stronger Stockfish getting a win, for example perhaps the weaker Stockfish made more inaccuracies than on average. That is why I’ve started a second game: Stockfish depth=59 vs Stockfish depth=23, which hasn’t yet finished.
The match from the position after 61. Rc1 has ended, with a win by checkmate for White. See the URL: https://www.chess.com/analysis/game/pgn/4gNNyfjqnN?tab=review . The game review finds inaccuracies at 57… Be7 and 63… Bg7, however I’m not convinced these are actual inaccuracies. I don’t know about 57.. Be7, but as for 63… Bg7, a new calculation at depth=45 shows as first choice 63… Bg7 at +5.31 and 63… Nc4+ in second at +5.98 ( the game review’s choice of best move). After further calculation up to depth=63, 57… Be7 is a genuine inaccuracy ( eval +3.33) and 57… b3 is indeed the engine’s first choice (eval +2.14), so the game review got that one right.
Interestingly, for Black the opening gets an accuracy score of 100.0, the middlegame a score of 98.5 and the endgame a score of 86.9 . I noticed a slowdown of the depth=59 calculations in the endgame especially at 66. g3+ in the first game of Stockfish depth=59 vs Stockfish depth=23. The low score of 86.9 accuracy for the endgame of game 2 makes me wonder if endgame accuracy scores are reliable.
The second game of Stockfish depth=59 vs Stockfish depth=23 (starting from the position after 7. Nd5) is ongoing. The second game ended in a clearly winning ( +5 according to Stockfish depth=59) position for White, but it took two tries. On the first try, the stronger Stockfish made a mistake resulting in a 0.0 or equal position. Game 2 can be found at URL : https://www.chess.com/analysis/library/3PymjYD8tW?tab=analysis .
