Sunday, February 03, 2013

The real problem

I'm busy to acquire the knowledge how to play the various endgames. And that is good, since I need that knowledge. But my real problem is the transition towards the endgame. Without endgame knowledge, I cannot even hope to ever fill in this gap. But having this endgame knowledge acquired is no way a guarantee that I will learn how to make the transition from the middlegame towards the endgame. And that is where the real problem lies for me.

Take for instance this diagram from a game that I played yesterday:

Black to move.
[FEN "3r1bk1/1pq2p1p/4r1p1/2P1p3/PQ6/2R1B2P/6P1/5RK1 b - - 0 27"]
I'm black. I'm a pawn ahead. I certainly outplayed my opponent sofar. As is so often the case. But I have no clear idea how to play this position.
In the past I solved this problem by offering a draw. Even people with a much higher rating would accept that. But now I am no longer in time trouble in this kind of positions and now I no longer offer or accept draws, I must learn how to play this. But I have no clues somehow, since I always avoided this kind of play.

I played 27.  ... Qc6 here, in order to block the c pawn, thus blocking the white bishop,  and to improve the mobility of the queen. According to Ivanhoe I give 0.61 pawnpoints away. Without noticing it or understanding why. I can't play this if I haven't the right patterns. These are the 16 best moves in this position (see table below). The reason I didn't play 27. ... e4 is that it frees the white bishop.

I managed to play another 15 moves, giving all advantage away and finally losing the game. Without noticing even how. The game went on:

27. ...  Qc6 28. Rb1 Ra8 29. Rc4 Ra7 30. Qb5 Qxb5 31. axb5 Ra3 32. Bf2
e4 33. c6 bxc6 34. bxc6 Re8 35. c7 Raa8 36. Bg3 Rac8 37. Rb8 e3 38. Rxc8
Rxc8 39. Kf1 f5 40. Rd4 Be7 41. Rd7 Kf8 42. Bd6 1-0

I finally was in time trouble due to the fact that I played with so little clues, and my play started to detoriate seriously so I resigned.
What are the issues in this position I should be aware of? Somehow this is a blind spot for me.

1. c5 cant move, no need for blocking ( and you are not suppose to block with a queen...).
e5 is passed, pushing creates space, the rooks are already behind and controling the d file. e4 is mobelising the black queen, reduces the mobility of the white queen and Rc3. After Rd8-d3 the blockade at e3 will fall sooner or later...

Suggestion: play (blitz) it as white against an engine several times

2. I agree, it is difficult to understand e5-e4, and it probably is even more difficult to find it without the help of a program.

Looking at Qc6 - well that move is not much worse than the 2nd best move of the evaluation list you showed us.
Missing a very good move isnt so tragic. It is rather that you lost more and more ground after Qc6. It must have been the beginning of a series of not so good moves.
I would think that e5-e4 indeed frees the bishop and white, and I also would not like to push my pawns before I traded off a bit more pieces (at least the queens).
The whole position looks like white has some compensation for his pawn, and a program can only give these evals, because it has much deeper insight of the various variations. In such a position, a normal human cant look much deeper than 1-3 moves ahead. It looks very dynamic, and only programs are fearless.

Of course we should learn from our blunders and inaccuracies. But what you could also have a look at is: had there been a possible way, where you would not win a pawn, but have a bit of a better game, keeping the initiative with long term positional advantage? (long term positional advantage is: your opponent has a double pawn, of a weak isolated pawn, or a bad bishop versus a good knight in a blocked position).
Karpov said, that if he had the choice of going into probably favorable tactics or avoiding tactics and maintaining a microscopic advantage he would think long about his decision: no doubt he would go for the microscopic advantage instead of entering the blur of tactics.

Then again, this made sense especially for him of course, because all his opponents were weaker than him (later came kasparov). If you are stronger, than it is always good to avoid tactics. On the contrary, if you are weaker, the best chances lie in the random blur of tactics assuming that your chances are worse anyway.

3. The endgame knowledge that you should know is that if both rooks and the queens disappear, a +1 pawn bishop endgame should be won by you. If you can go to +2 pawns with a rook on either side, the endgame should be won by you, otherwise rook endings tend to be drawish. Queens endgames are uncertain but are often decided by whose pawn is closer to queening: White's c-pawn is on his 5th rank. And the a/c pawns present a queenside majority which is far from your king's ability to help blockade. Your e-pawn is on 4th and closer to his king which can help blockade. But this is still middlegame. The computer line ends up being a tactical line that wins White's greatest advantage: the space provided by his c-pawn. The computer line gives you a rook outpost on e3 and tactics against the king on the g1-a7 diagonal such that you can temporarily lose the exchange but get it back. In the end, the computer line wins White's plum on c5. One great thing that White achieved during the actual game was reconnecting his a- and c-pawns into a duo against your c-pawn. They were better kept apart. If he allows you to capture one, he may have to give you time to capture the other. It's true that advanced pawns are difficult to assess: weak or strong. But your pieces eventually get pushed to the edge. That is what you should have done to him. Controlling the center is not just for the opening. His compensation is not just the c5 pawn, but that his pieces have more space and the center. And if you could have figured out the tactical capture of c5 or even bigger game, then tactics generally override positional considerations. In the endgame, space and center become a little less important in the general sense, but still important in key squares and lines and pathways to queening.

4. @Aox,
that is a good idea, playing it as white. I will try that.

5. @Munich,
if I mange to loose these kind of positions systematically, then a more crude fix is needed than to avoid it.

6. @Soapstone,
Thanx, that's very enlightning. I'm going to dig deeper in this position for some time, until I get the feeling I get the hang of it.

7. errr - I forgot to include the word "not" about karpovs decision making:
"he would NOT think long about his decision: no doubt he would go for the microscopic advantage"

We cant always avoid tactics. But it is a general good rule, that you must not grab the very first weak pawn that is in sight.
You can leave his weak pawn being weak (instead of taking it) and strengthen your position further. Dr Nunn said once LPDO = "lose pieces drop off".
On the other hand, sometimes of course you better take the pawn as long as it is available.

The nature of tactical stikes is, that you need to overstretch your position a little bit yourself in order to reach out for the material gain. Tricky tactics usually work due to unusual looking moves your opponent didnt see.

A very dangerous point in your game is usually the next 1-2 moves after a successful tactical outlash you just made.
The position you showed us looks a bit passive for black.
So my question is not to be taken lightly: could you have avoided in the first place going into a dynamic position where you have little clues about what to do?

You are equally right that you need to learn how to play this position, but that does not make wrong what I am saying, too.
We would need to have your full game to know more.
And sometimes the very best move a program would play isnt the very best move for a human a-class player. For practical reasons.
I dont think Qc6 was a bad move, because chances are that after e5-e4 you would lose, too, just because after e5-e4 you might need to find a series of good moves, and 2nd best moves wont do the job. That is no problem for a program, but it is a problem for a human. Also e5-e4 doesnt look good to me if you were tight on time, too. Again that is nothing a program had to fear, but a human has to take time into account, too. So Qc6 - well it looks safe to me without throwing your material gain over board and without being to commiting/binding.

8. @Munich,
There was no choice than to take the pawn. It was a defensive measure. In fact I was prepared to sac a knight for two pawns, just to stop his gain of space in the center. With a neat tactical trick which I saw after some more thinking, I managed to get the piece back.

It is not so much about this specific game. It is about signalling a fundamental flaw in my play. Not everyone can win this position. But loosing it needs a special kind of blind spot.

I never grab pawns lightly. I have played too long gambits to underestimate the side effects of a pawn grab.

9. Endgames are like middlegames, but with less pieces, deeper calculation and less branching.

I would love to get away with, as Black in this position, some sort of exchange sac like ...Bg7 and ..Rd4, opening the way to White's king. Two problems with this. Rb4 will refute this, and White doesn't need to accept the sac anyway.

The Reason ..e4 is the solution to the position (I didn't think of it either) is not so much because it makes so much friggin sense, which it does anyway as Aox pointed out, but because it refutes White's next dream move, which would have been Rb4, ganging up on the b7 pawn.

I do have problems missing obvious moves or making obvious sense of the position like Aox just did. He's obviously stronger than me at general chess principles to be thinking the way he described above.

...Qc6 did seem wrong to have the queen as blockader though, and it's closer to stepping into the firing line there as well.

Great position to have analyzed together. :-)