Friday, September 30, 2011

Game Analysis--Cabablanca v Tartakower NY 1924

I bought a chess book for my son. The Most Instructive Games of Chess Ever Played by Irving Chernev. It was a 1992 edition that is in descriptive notation. So I have taken it as an exercise to periodically take a game from that book and analyze it myself. I will then compare my analysis to Chernev and when I can find it others. As this game is from the NY 1924 tournament, for which I have the book. I also have Alehkine's annotations.

The NY 1924 tournament is very interesting. I was 22 rounds of games among some of the strongest players of the time. The previous world champion Em. Lasker took first, the current world champion Capablanca took second, and the next world champion Alekhine took third. Marshall, Reti, Maroczy, Bogoljubov, Tartakower,  Yates, Ed. Lasker, and Janowski rounded out the field. Bogo lost 9 games, so there has got to be a really great game there. The prize fund was $5475 with expenses of the players $3568 and travel $1940. The dollar sure was stronger back then.

Here is my commentary on Cabablanca v Tartakower NY 1924. I analyzed the game alone first, then looked at the notes of Alekhine from the tournament book. I think I covered most of what Chernev did, but his is better.

Instructive 1

Saturday, September 24, 2011

The Fastest Path vs The Best Path

I am doing two Chess Tempo endgame problems a day to help with two things. One is to get faster at exact endings, and the other is to achieve a deeper understanding of piece interaction in a simpler context.

The Chess Tempo endgame problems have two modes: theory and practice. In theory mode, you fail the problem if you do not follow the fastest path to mate defined by a tablebase. In practice you keep going as long as you pick a winning move, and the extra count against your score. I am using practice mode, because I do not think it is humanly possible to memorize tablebase fast paths, and I want to move quickly in the endgame. So I work on understanding processes and patterns, which I can apply to solve the problems.

Chess Tempo problem 66560 is a good example of this difference.
White to move and win
The key elements for how I decided to play, was that White's king is in the square of the a-pawn, and the White knight prevents the Black king from taking the h-pawn. So White can march his king over and take the a-pawn, and Black cannot protect it without leaving the square of the h-pawn. Then the White king marches back and with the knight forces the h-pawn to queen.

The fewest moves path is to recognize that White can force the h-pawn home without the need to take the a-pawn by the threat of this position:

To achieve this, the knight will need to blockade the a-pawn, and be able to get to f7 before black's new queen can enter the game. From a4 to f7 is four moves, a3 to f7 is three moves, so either would work for the mate. However, it is three knight moves to control a4 and two to control a3, and white needs one move of the king to secure the h-pawn. So the blockade takes place on a3. This is a good counting and visualization exercise.

I would play the clearer but longer(more moves) plan in a game.

Apologies: I could not get my normal PGN viewer to work, today. I could not figure out how to make the board larger on this one. Here is the PGN if you want to take it somewhere else.

[Event "Chess Tempo Endgame"]
[Site "?"]
[Date "2011.09.24"]
[Round "?"]
[White "Us"]
[Black "Problem-66560"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/p4k2/8/7P/3K2N1/8/8/8 w - - 0 1"]
[PlyCount "19"]

1. Ke5 {to secure the h-pawn, Ke4 and Ke3 both work} a5 2. Ne3 a4 3. Nc2 {The
knight catches the a-pawn} Kg7 4. Kf5 Kh6 5. Kg4 Kh7 6. Kg5 Kg7 7. Na3 Kh7 {
now the h-pawn advances and the end is nigh} 8. h6 Kh8 9. Kg6 Kg8 10. h7+ {and
Black must choose between being mated and letting the h-pawn queen.} 1-0

Saturday, September 17, 2011

TCCL 2011 round 1

Twin Cities Chess League round 1:

The good news: We are playing in the siver division, where the teams should be more competitive.

The bad news: Probably fewer masters and experts for me to play.

The "good" news: I got stomped in the first round by a player rated 2359.

He played a Botvinnick system against my KIA. This was the same defense that GM Kudrin used in his demolition of me. I have also seen it while playing the KID. I think I will be able to extract some lessons from this game.

The more bad news: everyone else on the opposing team was rated over 2000,  and they swept us.

Thursday, September 15, 2011

Take the Time to Count.

I had a disappointing drawn in the Region VI Championship. In the following position I played (as black) a combination that reduced to a winning endgame.
black to play
30...Nxe4 31.Nxe4 Rxe4 32.Rxe4 Rxe4 33.Qxe4 fxe4 34.Rxf6 Kxf6 35.Bxe4 Bf5 36.Bxf5 Kxf5 my king is placed much better 37.Kg2 Ke4 38.Kf2 Kd3 39.Kf3 to reach this position.
black to play

Here I messed up badly. I was worried about getting my king trapped on the a-file. I was not in time trouble, I had plenty of time. I messed up the visualization of the position after 39...Kc3 40.Ke3 Kxb3 41.Kd3 Kxa4 42.Kc3
black to play
This is a clear win for black with 42...b5. There is no problem even if  black did not have a b-pawn, because white will have to eventually release control of b4 anyway,. but even though I was not sure this was a win, If I had counted the possible other paths, I would have known that they were at best draws, and this was the best position for me, even if I was trapped and it was a draw.
black to move
This diagram is messy, but the different colored arrows mark moves to make a queen. Without interference, White's red path to queen the d-pawn is 8 moves. Black's green path to queen the c-pawn is 8 moves, but it interferes with white's red path halting it after move 3 (White's cannot move Ke6-d7, because it leaves the d-pawn loose. White's path to queen the g-pawn is 9 moves. Black's counter to that path starts with 39...g5, which is answered by 40.h4 g4+ 41.Kf4 and 8 more moves to queen the h-pawn, but black queening the c-pawn has not started, so it is still 8 moves. The red path for Black counters the white h-pawn plan with a 7 move path, but now the white d-pawn path is 7 moves, because the white king has taken one step. So 39...g5 makes it easier for white to queen a pawn, but no easier for Black. My 39...g5 was a horrible blunder that threw away the win. All resolvable with some counting. Tedious and time consuming, but I had the time. Since all paths but 39...Kc3 were draws or worse, that was the right move, even if it ended with my king trapped.

The key is to visualize each path in turn and count it out.

Addendum: there is a better win after ...Kxb3 41.Kd3.
black to move
41...Kb4! and white will have to abandon the c-pawn. (42.h4 Kb3)

Addendum 2:
For those who are screaming at me, that trapping a king in front of his rook pawn does not work unless the pawn is far advance, I know that now.
black to move
1...Ka3 2.Kc3 a4 and black has the opposition and is released
either to b2 to escort the a-pawn to queen, or to b4 to free and escort the c-pawn. The rook pawn alone in this position would be a draw, because white can get to the queening square, or trap black's king.

Tuesday, September 13, 2011

2011 TCCL starts Friday 9/16

I just took a look at last years TCCL. All the teams in the Gold division except for the Rochester teams are headed by a master. Some have several.

That's interesting

I wish I was ready to start playing the Catalan, but at least I should get plenty of learning material.

Perhaps, I will get lucky and have black on Friday.

I am still going though the games for the Region VI Championship. I blew a couple of endgames (drawing a win, losing a likely draw). I should get a post done this week on the draw. I still don't understand enough about the loss, yet. I did win an endgame, because my opponent got into time trouble and missed the perpetual I gave him.

I really need to work on endgames.

Wednesday, September 7, 2011

Knights are Weird

While daydreaming about topology the other day, I was taken by a singular question.

"How do the chess pieces view the chess board?"

If each move is taken as a single step, if each square that a piece can move to is considered adjacent to the square the piece is on, each piece lives in a very different world from the others.  Bishops do not even see half of the board. On an empty board 14 squares are adjacent to the Rook, and 49 others 50 are two steps away, no matter what square they are on. This is not new ground of course, it is covered often when discussing the mobility of the pieces, but it is another step closer to verbalizing and understanding a vague idea I have had for a while.

When I was teaching my sons how to play chess, I tried to help them visualize how the pieces interact, by calling the squares that a piece can move to the shape of the move. I was reaching for a Euclidian geometric description.

The chessboard appears to be a Euclidean place, 64 squares assembled into a flat 8x8 grid. The Go board is such a place, but the chessboard isn't. It is a network of 64 points, but the connectivity of that graph is different for the different pieces. The graph changes for the pieces as other pieces move. No point is reachable for a piece if a ally piece occupies that point. For the pawn, the graph changes radically as the pieces move, because diagonal moves are dependent on an enemy piece occupying a point.

And the Knight views the board in a wholly different way:
1-red, 2-orange, 3-yellow, 4-green, 5-blue, 6-purple
I have colored the squares based on how many moves it takes for a Knight to reach them. But beyond the red squares (one step) coloring is not enough, because the path to those squares matters. Lets place a white pawn on f2, so the knight can only go through g3.
1-red, 2-orange, 3-yellow, 4-green, 5-blue, 6-purple
Now it takes two more moves to reach e1, e5, d1, d3, h3,g4, b2, and b4 .

I think it is well known that Knights are tricky, but how do we get better at using our Knights? I do not think that this is so clear. I am going to try working on knight endgames. They are not very common (Silman does not even include the N+B mate in his Complete Endgame Course), but I think the clarity of endgame positions will help. OTOH, here is a fun puzzle game called Black Knight, that you maneuver a knight around different board topologys.