Monday, May 13, 2013

Sixty Memorable Annotations

#17: Alekhine - Capablanca, Buenos Aires World Championship 1927

White to play

The tablebase tells us the fastest win starts with 82 Rg7, from a human point of view of course 82 Re7 is much more sensible, preventing the Black king from joining the action.

Jonathan Hawkins, Amateur to IM (Mongoose Press, 2012)

Nalimov tablebases, God I love them. That six-piece tablebases are freely available to anybody with a 'pooter and an internet connection has got to be the greatest learning tool of the century. For those of us who are trying to hack our way through the mysteries of rook and pawn endings, anyway.

It took me a while to work out how best to use them, though. Their cast-iron certainty can, paradoxically, mess with your understanding if you're not careful.

So you've got a choice between Rg7 (which leads to mate in 24) and Re7 (mate in 29). Does that mean the first move is better? In a sense, yes, but knowing which move leads to the quickest win doesn't necessarily mean you've grasped the essential logic of a position.

Drawn positions are even worse. That's where you have to be really careful with Nalimov, I think.

The look of a man who reckons he's better than any tablebase

So let's take another look at Aronian - Carlsen, Moscow 2006 (see Purposeful Shuffling from a couple of weeks back and its predecessor Random Rook Endings VII). A rook ending that should have been drawn but which our teenage Future World Champion managed to lose.

At what point does that half-point go adrift?

28 ... a5

36 ... h4

43 ... Rb5+

51 ... Rb1

60 ... Kf8

67 ... Kf8

69 ... Ra1

72 ... Ra8

73 ... Ra7+

74 ... resigns

Nalimov isn't - yet - any help in the first couple of positions and while Angus tells me that seven-piece tablebases have now been created they're not yet available online so we can forget the next three as well. From move 69 onwards, though, we can be absolutely sure. Check it and you'll see that it was only after Black's very last move, the awful 73 ... Ra7+, that the position becomes winning for White.

So it's whilst it's true to say, as a kibitzer at does, that

Neither of the players left the drawing path until 73...Ra7+ ....

it's misleading. What you could never find out if you rely solely on a tablebase, is that in no way was Carlsen's first mistake that losing final move.

There are paths and there are paths aren't there? The one Carlsen chose took him close to the edge of a cliff. From there it only took one slip and he tumbled over.

Nalimov might tell you it makes no difference, that one draw is as good as any other. For a human, though, it's obviously best to stay as far away from the precipice as we can. We'll come back to this next week.

Sixty Memorable Annotations Index
Rook and pawn Index


AngusF said...

"while Angus tells me that seven-piece tablebases have now been created they're not yet available online...".

Uh, the tablebases are available online though only (so far as I know) to customers of certain ChessOK products. They can only really be made publicly available through an online facility as they take up 100Tb of disc space. More info here.

John Cox said...

It strikes me that almost the worst move was 69...Ra1. Didn't Philidor teach us to play 69...Re2. White can hardly do anything at all then - still, nothing wrong with ...Ra1 if it reaches the defensive position you've studied, I suppose......

Jonathan B said...

Indeed. More or this tomorrow.