Engine Evaluations

    When analyzing games for this Blog it is not uncommon to run into engine evaluations that I simply do not understand. In many cases running a Shootout (the engine plays blitz games against itself) will help confirm the evaluation. However, and this is an important however, some of those games can last 80, 90 or 100 or more moves! For a game between humans this is meaningless. Stockfih, and a lot of other engines for that matter, are so good that most humans do not know why their evaluations are what they are. But, are they always correct? 

    One comment I read seems about right: “...stop worrying whether (an engine) likes a position or not (because) it is far more important to have a position you are comfortable with rather than one where you have no clue how (to play).” 

    This problem is addressed by Kostya Kavutskiy, an IM from the United States with a 2300+ rating, in the following informative video. Just for fun you might like to set up the position he shows at 4:20 in the video just to see how you equipment performs. On my laptop after 20 minutes Stockfish 17 running on three cores gives the best move as 1.Bxb5 with a 5.49 score. When I played 1.Bxh7+ (which was not in its top 3 choices) it evaluated the move at 5.33 almost immediately.