Here is a game example. What should White play on move 5? Opening explorer says d3. Computer engine says fxe5 is best and d3 is an inaccuracy.
Specifically the opening explorer states:
d3 was played 272 times, 47% White win, 22% draw, 31% Black win
Another example: is White retreating the bishop here really a blunder? Or is it more of a missed win?
Engines value the bishop pair that is why they see d3 as a lemon.
When you are starting out you don't know that.
What I don't understand is the chess.com opening explorer is based on master games. So if d3 is a lemon why do master level players play it?
What master games? Black would not play 3. ... Nc6 because of 4. fxe
So how would a master game find themselves pondering 5. d3 or say 5. Nf3
I can see 5. d3 at club level of course because black is in trouble anyway.
Are you sure the openings explorer is master games and not just games played?
There was a bug in the opening explorer recently and maybe it hasn't been fixed yet. The opening explorer, based on the master games database, says 3...Nc6 was played 13 times.
There is an example of a specific game labeled 4. Bb5 ½-½ Behting Carl - Karel Hromadka (1924.??.??)
But when you look at the game
https://www.chess.com/games/view/16683
Nc6 was played on move 2 not on move 3, and Bb5 is played on move 3 not on move 4.
Chess.com's engine is pretty weak. You can safely disregard the engine's opinion if the moves are relatively close. Also, I'm pretty sure that in the above position every good-looking move is a book move, the engine is just talking nonsense. Also, apparently the opening explorer has a lot of bugs too.
When you are analyzing your games, you should focus on tactical sequences in the middle game where pieces are won and lost.
Interesting game (https://www.chess.com/games/view/16683), but to me (speaking as a neophyte) it seems that white played to draw not to win.... Sorry if the comment is off-topic.
Interesting game (https://www.chess.com/games/view/16683), but to me (speaking as a neophyte) it seems that white played to draw not to win.... Sorry if the comment is off-topic.
Yes, Behting Carl is not the most famous of players so was probably looking to secure a half point against a better player. The game was almost 100 years ago and not exactly a classic.
Another example from my last game: The game is mate in 8. Opponent moves king such that it is now mate in one. Analysis engine rates the king move as good.
Depends on the skill rating
Computer wont make a mistake so if a move is a blunder or innacuracy there is a reason why and your future higher rated opponents will be able to use that in their adventage. If you know from your previous games that one specific move isnt good as you thought so, try to replace it with the best move that computer suggested and see what happens.
Don't forget that the opening explorer doesn't tell you what move has been played in this position but how many times the resulting position has been reached in all possible transpositions. So it doesn't understand moves made in the wrong order. That is not necessarily the sensible way to do it although but it is the way it works. Consider the following position:
Opening explorer seems to suggest that this position has been reached 189 times in Master Games in its database and 188 times black missed the free knight and played Nf6. Of course it hasn't - 188 times black played Nf6 before bringing then white replied Ng5 and then the black bishop came out to the c5 square. On one occasion a Master made the wrong move and their opponent accepted the free knight and won quickly. I don't think that is a bug - I think that is the way it is designed.
And also ip
And also opening explorer doesnt suggest moves, it just shows what has been played, its on you to figure out how strong each move is.
You can get ideas from the computer analysis, but the computer is not perfect. Sure, if you link several mainframes and have a team of techies standing by the machines beat Kasparov, but we've all noticed less accurate results from chess.com analysis. One otb game I played a couple of years ago ended in a draw agreement. The computer here said I had an advantage of about 2.50, but simply playin the recommended best moves led to a repetition of our final position after about 10 moves, and no other moves led to any different result. Perhaps the computer meant that there were more, and more serious, mistakes my opponent might have made, but both of us A-class players clearly saw the draw.
In the openings, many times the program will say another move is better than your choice, but if you understand what you are doing with your choice, that move will work out better for you. (Assuming that the difference in the moves is less than 3/4 of a pawn.) In the middle game, aside from gross tactical mistakes, the same principle applies. The machine's first choice may only be valid if you can also play the machine's first choice for the next ten moves, whatever your opponent might do. Having a plan and understanding what you are doing is more important than making a move that may be .35 better and losing the thread of the game.
I analyze all of my games afterwards but sometimes the analysis seems to go against common sense. It may be because I am a relative beginner and don't fully understand the ideas behind the best moves, but it is clear to me that the analysis is not the be all and end all. I'll give a few examples:
I have seen videos from some top players who say they played a certain move even though the engine doesn't like it much, they are almost apologizing for their move despite having very sound reasons for playing it, and having won games with it.
When I look at the line associated with a best move, I often think "never in a million years would we have played that." Perhaps at the higher level those lines are closer to how top players play but they are often just a series of strange moves.
If I have played a few games against someone, or if I have studied several of their past games, I have a reasonable chance of putting them on their next move. The computer never tries to do that.
The person with the highest accuracy does not always win. In maybe 2 or 3 percent of games I have seen the cleverer player wins with the lower accuracy percentage.
Does anyone have any guidelines on reviewing computer analysis? When the computer says this move was an inaccuracy or a mistake or a blunder, when should we think "I must stop playing that" and when should we treat the feedback more skeptically?