Saturday, July 07, 2007

Hitting the wall

Last week I had a nice chain of reasoning. Today I tried to play out the position from yesterday against various chess engines and immediately I hit the wall.
When I trade off to a rook ending, I'm not able at all to win this. On the contrary, in most variations I manage to lose. I have no clear picture what to do in this position. It is a side-variation from the game in the book, which means that I'm left on my own.

This is the position:

White to move.
The plus pawn looks promising.

I went to my endgame books for advice, but they only treat endings that are much closer to theoretical endgames.
This is the advice I could distill from my 16 endgame books:
  • Play actively, often even at the cost of material.
  • Trade the rooks when the remaining pawnending is won.
The other three "rules" I found had no relation with this position.
This advice is very flimsy. Even more since "actively" isn't exactly defined.
It's obvious I hit the same wall as Takchess.
It is pretty amazing that there aren't any books which treat the most important part of the game. There are two options that arise:
  • Study rookendings from annotated mastergames. But then again, when you try to play this out against the computer you will hit the very same wall when the computer deviates from the book. The only thing you can hope for is that the annotations give some clue what to head for.
  • Think for myself.
Bummer, I'm afraid I have to resort to the last option.


  1. In keeping with the importance of learning general rules of the endgames because of the uniqueness of most endgames,I found this interesting. I searched this position in my 1.100,000 Fritz game database and there were 0 matches. With so few pieces, I felt there would have to be a match.......

  2. I would already be happy when I knew what to move, the rook, the king or some pawn.

  3. very beautiful post.

    thank you...

    after 288 games, 95% of them bullet, in the last 14 of 15 days, it is time to step back, do some review, and analysis, then start at 3/8 in view of 2/12...

    did you include endgame strategy as among those books, 14, was it?

    time for bed. and take care. excellent work temposchlucker.

    warmest, david

  4. Well, you can see it as followed. white's rook attack the c-pawn which is isolated. Black has two choices. Defend this pawn with his king or defend it with the rook. By laterally defending the pawn with the rook black will severely restrict the mobility of his rook. Moreover the c-pawn will never be pushed forward while being under attack by the white rook. So the king support will be needed to advance this pawn if you have laterally defense of the pawn. Also the mobility of the black rook is severly restricted to the a and b file.
    First thing that white does is "improve the king position". To where? to the Q-side ! the pawns need support of the king in their advance. If the black king attacks the white rook, It goes down to the third rank. Idea: to make b4 possible in the future while keeping a3 protected. This also holds the black rook to the a-file.

    Moreover it will give white the opportunity to swing the rook to the king side when necessary. It is further vital that the central open file does not give black the chance to infilitrate with his rook. White can in most cases challenge the black rook on the central file as white has the stronger position.Therefor white has the stronger rook and if the pawn structure remains intact black will not exchange down the pieces. The Pawn majority on the Q-side should be enough for a win.

    So black will probably try the following: attack the white rook. push c-pawn one square forward and open up in this way the 7th rank for his rook. Where possible he might try to prevent the white king of reaching the Q-side pawns. White can use his rook as shield once brought down to the third rank as the white king will walk along the second rank to his b-pawn.

    of coure black will try to mobilize his k-side pawns but this should be met easily.

  5. Montse,
    I'm impressed. I came up with the same things, it only took me 10 hours before I got some inspiration. I'm not used to this way of thinking about the game at all. Which means that there is a lot room for improvement.

    I hope I'm able to formulate a few guidelines how to approach such positions.

  6. it took me about 30 seconds, and I suck at endgames.

    I think your problem is in planning, not necessarily in your technique, and you just need to play a lot of CC to overcome that. well, of course, in a way that's very similar to using 10h on this position, son I guess it should work as well.

  7. WW,
    it's not quite clear to me what the result is of your 30 seconds. All the idea's that Montse came up with?

    What do you mean by "planning"?

  8. in that time I saw there was no obvious tactics (excluding a possible rook sac for black c-pawn if white can advance queenside pawns far enough), that the black rook is tied to defend c-pawn and black king is cut off from the pawn duo. black can't penetrate on kingside without rook helping, and the rook can't abandon the c-pawn. so it's pretty obvious to walk the white king up to queenside and queen the pawns. black can't really do anything to stop that.

    by planning I mean constructing such plans as this one. I think it's a skill, just like tactics, calculation or visualization, and can be trained only by doing it. tacit knowledge. which is where a steady diet of CC comes into picture.

    you've said over and over again that you find yourself in positions which you don't know what to do in. you try to combat this by searching for some rigid rule based system which would in some magical way 'solve chess' (in a broad sense). but I don't think that's possible at all. if it was, machines would've played great positional chess decades ago, far before they had any kind of serious number crunching power to brute force it all. and anyone could play good chess just by obeying some set of rules a gm gave us. but we all know it just doesn't work that way.

    and as usual, the grain of salt disclaimer applies to everything I just said.

    aw crap, my monitor shut down third time during this. it's dying, and a new one is already in coming. I'll just stop before it shuts down again...

  9. WW,
    you are quite right. I have a huge problem with planning as you define it. I'm much too critical on my plans. The idea's you mention where rejected by me since they don't work. That took me already quite some time. But often it is better to work with a bad plan than with no plan at all. Hope is underway though as you can read in my latest post.

  10. WW,

    The essence of this rook endgame will be that you will have one pawn left on the queen side and three pawns on the K-side. You will need to create a second weakness to win this position and this will be at the kingside. For instance saddling blak up with a backward pawn which would be an enormous liability.