Annotating Games with Scid and Crafty

Aug 20, 2010, 9:42 AM |

Right after completing a game, I plug the PGN into Scid and annotate it with my freshest thoughts from the game. Use the comment editor window for this, and remember to store your comment before moving on. At this point, I'm not even concerned about analyzing it to see if my ideas are any good... I'm just putting down what I was thinking so that I don't forget when I go back and look later.

It's a good time to get somebody else's thoughts on the game after you've gone over it briefly for yourself. I personally post to's game analysis forum, where anybody can comment. Any useful comments can be added to your initial annotation.

Now, I would recommend you use a chess engine to fine tune your analysis. I use Crafty, which is quite strong. I'm really only interested in lines that it finds more or less immediately, so I set it to spend 1 second on every move. I do a blunder check with a threshold of 1 pawn. The only purpose of this is to highlight missed tactics.

If Crafty reports a blunder, I will first examine the suggested line. If the error is immediately apparent and obvious, I will add it to my existing annotation. If it's obscure or nuanced and I wouldn't expect to find the move in game, I simply discard it. In other words, we're looking for human moves, and if it tells us computer moves we can happily ignore it.

Sometimes Crafty will highlight a move as a blunder, and give a preferred line, but it won't be obvious why. This usually happens when your opponent has missed some tactics. In this case, it's useful to look at the engine's recommendations for your opponent's response. This gives you a chance to become aware of your blunders, even if you "got away with it" at the time.

As long as we know we're looking for human moves, moves that we can expect ourselves or our opponents to make, we can get the best of both worlds by using engine analysis after we articulate our own thoughts and get feedback from other players when possible. If we mindlessly feed our engines game after game and look at hundreds of computer generated lines without understanding the nuances of the situation, we won't learn a thing.

During this whole process, feel free to elaborate on any ideas that come up, and note plausible variations that you might see. You can also run engine analysis on your variations, so you don't end up including lines with horrible blunders. They'd be counterproductive if you ever looked the game over again.

After we're done with all that, we should have a great record of the game with a full, detailed annotation. The process of creating it will hopefully have led to some insights, and it will be a great reference and study tool.

Here's the result of my first attempt, that I created as I was working out the process:

[Event "Live Chess"]
[Site ""]
[Date "2010.08.19"]
[Round "?"]
[White "MarshallScott"]
[Black "tambata79"]
[Result "0-1"]
[WhiteElo "1190"]
[BlackElo "1374"]
[TimeControl "15|10"]
[Termination "tambata79 won by checkmate"]

1.e4 c5 2.f4 d6 3.Nf3 Bg4 4.Bb5+

{Making it up as I go along at this point... Not playing any opening I know of.}

4...Nc6 5.Nc3

{Developing... protecting the pawn, mainly. I'm sure the bishop will have to move soon.}

5...Nf6 6.d3

{Seems natural.}

6...a6 7.Ba4 b5 8.Bb3 Nd4 9.O-O a5 10.a3

{odessian on suggests an alternative:}

( 10.e5 dxe5 11.fxe5 Nd7 12.Bxf7+ )

10...a4 11.Ba2 Rb8 12.Ne2

{I wanted to break the pin, and I thought black might play b4 anyway. Also I might win a pawn if he doesn't take one of my knights.}

12...Nxf3+ 13.gxf3 Bh3 14.Rf2 Qd7 15.f5 g6 16.Nf4 gxf5 17.Nxh3 fxe4 18.Bxf7+

{I think I might get some compensation and expose his king.
Crafty reports a blunder here, based on a line assuming fxe4, when I was intending Nxe4 as my plan. I'm just ignoring it, even though it says I'm losing a piece and a half, just because it's so weird.}

18...Kxf7 19.Ng5+ Ke8 20.Nxe4 Rg8+ 21.Rg2 Rxg2+ 22.Kxg2 Nh5 23.Qe1 Qf5 24.Qh4

{"24. Nxd6+ would have been nice." -PVKeeper10 on}
( 24.Nxd6+! )

24...Bg7 25.Bg5 Nf6 26.Bxf6 Bxf6 27.Nxf6+ exf6 28.Re1+ Kf7 29.Re3

{Kind of a weird move, I guess. Its only purpose was kind of a blockading idea, so I can play c4.}

29...Rg8+ 30.Kf2 Qg6 31.Re1

{Embarrassingly, I'm really not sure what I was trying to accomplish here. Crafty finds a 7 pawn error and suggests a complicated line starting witth Ke2 that results in an even position after taking the rooks off the board.}

31...Qg2+ 32.Ke3 Qxc2 33.Qxh7+ Rg7 34.Qh5+ Kg8 35.Qe8+ Kh7 36.Qxb5

{Crafty notes this loses the game (black has Re7+) and suggests:}
( 36.Qh5+ Kg8 37.Qd5+
( 37.Qe8+ Kh7 38.Qh5+ Kg8 39.Qe8+ {dschaef2 on suggests the possibility of perpetual check.} )
37...Kh8 38.Qxd6 Qxb2 += {HT} )

36...Kh6 37.Re2??

{Knew I was just trying to buy some time until I came up with a concrete idea. Had no idea it would be this disastrous!! Crafty thinks I should draw with perpetual check:}
( 37.Qe8 Rg2 38.Qf8+ Kg6 39.Kf4 Qd2+ 40.Re3 Qxb2 41.Qg8+ Kh6 42.Qh8+ Kg6 43.Qe8+ Kh6 44.Qf8+ Kg6 = )

37...Re7+ 38.Kf2 Qxe2+ 39.Kg3 Rg7+ 40.Kf4 Qe5# 0-1

Crossposted at my Blogspot blog.