Computer analysis is disheartening

I'm just an average player, but I devote a lot of time to studying the game and I'm trying to improve.  One way to improve, of course, is to get a computer analysis of your games when you're done; it's one of the advantages of paid membership here.

I'm always deflated when I analyse games where I thought I played well.  No matter how soundly I defeated my opponent, no matter the material or positional advantage, it always says a good 12-20% of my moves were inaccuracies and blunders. It always puts these question marks and double question marks and question mark/exclamation point combinations after many of my moves.  It never puts an exclamation point.  It never says, "Pretty clever, there! Good job!"

Heck, half the time the moves I thought were clever it says were stupid.  It's disheartening, really.  And forget "Computer Workout" here--every move I make it tells me I've made things worse.

I know how you feel.  I don't think the computer can give you the "Good job" comment because it assumes/expects that you will make the best move everytime.  And as for the actual analysis, I somewhat worry about the mistakes and worry about the blunders, but not really the inaccuracies.

For a time i actually thought live staff members did the analysis of the games !! LOL

What is your rating? Let's assume it's something pretty average like 1400-1500, and you're critiquing a game of someone rated 800 playing against someone who's 760. That's probably about the ratings gap of you and a 2200 computer engine. How often are you going to go, "Wow great move there Mr. 800. You were really aggressive and sure put a lot of pressure on Mr. 760." To you, a lot of their moves are going to look ridiculously bad.

Ah, dont sweat it. The computer analysis of one of my games told me I made a mistake, and then the analysis it showed of the position put the other player in the position of losing their rook.

its true, it says little advantage and sometimes its a checkmate position (but i play for free, and the computer is 2000 rating...), lol!

Hi jswilkmd, I was once in the same situation as you. I bought a copy of Fritz and analysed my games and boy was I bad. Yes, it is tough and gets you down but if your persist with analysing your own games you will be rewarded!

At the time I decided to aim to make moves which would show up in Fritz`s top 5 moves. Gradually I began to make moves which were in the top 5. I done this for about a year and went through all the games I played seeing if I was making any improvement. True enough, I was making progress! As you get better you can aim to make moves in Fritz`s top 3 or even the top move when it comes to tactics. You will get better but it is a harsh reality when you miss moves that could swing a +/= position to a -+ position in favour for you. (Now i am able to regularly make fritz`s best move!)

If the top 5 moves is too much for you, why not try the top 10 and gradually aim to get down to making top 5 moves and so on. It will take time but if chess means that much to you, you will be able to do it.

One issue with the computer analysis engine is that if it thinks it has a particularly strong move, even a good alternative(by another program) is labelled a mistake, and quite often it calls master moves an inaccurary in opening theory.

If you want to be rewarded at least some times Fritz 11 actually gives out the odd ! to moves. Here's an example:

Another problem I've noticed with computer anaylsis is when playing fairly positional battles when not too many dynamic elements vary from move-to-move, once the computer gets its heart (umm gears) set on a specific move, it will proceed to mark all aberrations as errors or blunders even with minimal fluctuations in +/-.

The main issue with computer analysis is that the computer focuses mainly on hard tactical aspects rather than strategical ones (Tactics means exercise, exercise, exercise - strategy, on the other hand, is far more subtle; in fact, computers can't really grasp it). I'd have a strong player help me analyse my games and tell me ideas and plans that could've turned out better.

But don't be discouraged. Computers don't have a heart.

I know one 2300+ OTB player who plays mostly CC where he’s rated over 2500. Of course at that level all players use engines to blunder check their analysis, NOT to select their moves. Those players routinely win against players who rely exclusively on engines to make their moves. The reason they win anyway is because, as this player said, about the only thing an engine evaluation is good for is the material situation. Just try playing over high class GM games using an engine and you will often see the engine evaluate the GM’s move at a low score only to drastically change a few moves later when it "realizes" the GM’s move was stronger. Again that’s because GM’s are evaluating the position on "intuition" and are able to make judgments not readily seen by engines or the engines aren’t placing enough emphasis on positional considerations. The short of it is unless you’ve made a tactical blunder take the engine’s evaluation with some skepticism.

The long games I play in real I always put them in my database and the computer may analyse them. For the games I play here I dont use the computer, only to check after a game the opening (if it was a strange one) or an intresting endgame

Take the Eniamar vs NN game from Eniamar's post.

After 44... Rxg8, although 45. Qxa5 wins easily too, a simple way to win is to play a heretic 45. Qe6+ Rg6 46. Qxg6+!? Kxg6 47. Kg3 transposing to an easily won pawn endgame. Although a computer might treat this as a tactical blunder (losing queen for rook), a human might see this as "eliminating counterplay".

I had a game analyzed here and it was pretty disappointing. In fact I have never used it since. A standard Fritz type of software would be much more useful. What would be helpful to me, which I have never seen on a chess site, is if I could play blitz games and then immediately(on the same screen) quickly go over the game with a computer. Of course it would have to be set up so you could not use the computer during the game. I would want it to use the moves "already entered" to save time. I could learn where I missed tactics, etc. then move on to the next blitz game. I think this would be really helpful. You might find patterns where you consistently miss ideas/tactics. I just don't have the time to take my blitz games and plug them into the computer software, etc. Anybody else ever thought this?

I congratulated n opponent last week for his clever rook sacrifice that turned a game around. Then I got the game analyzed and found that the sacrifice was a blunder! It may have been a blunder if played against a 2500 rated player but it was a pretty good mave against a 1750 one.

Icanfight-- you can do that, babaschess with FICS allows you to load an engine and post-mortem your game. The software prevents you from using any engines during play however.

The point about engines on openings moves is a little misleading. For opening analysis; one should be using the openings book! Only then would you refer to fritz`s analysis.

John Nunn wrote about how to use computer engines to improve your chess and its all in his book ''Secrets of practical chess''. I will let you all read it yourselves :)

Ahhhh, computers donnn know how to play, they only calculate it out, millions and millions of combos to the very end.  You wish to be complimented by a computer? hheh? pleezzz

yes, i would like to see that change.  I want to see at least one or two exclamation points on my games.  I know I'm a bad player, but every so often, I do play a howitzer.

I disagree. Computers CAN calculate it out but that does not mean that the mistakes humans make are always down to bad calculation.