Monday, March 24, 2014

Collector’s Item

White to play
Kramnik - Karjakin, Candidates (2) 2014

I wonder if yesterday we saw the end of Kramnik's career as a credible challenger for World Championship. Still five rounds to go, of course, so too early to completely write him off, but a point and a half is an awful lot of ground to make up. Even for a man at the top of his game you’d think that a big ask and Vlad is very much not that right now.

If that is it for Kramnik in terms of playing at the very highest level it was kind of him to check out with a rook ending. Another case study for 'active rook is often a decisive advantage' folder you might say. Mind you being multiple pawns up probably helped Karjakin a bit too.

Anyhoo, rook endings are last year’s obsession. Let’s get on with the ISEs.

Today’s, courtesy of Kramnik - Karjakin from the first cycle, is an unusual twist on a standard theme. It also reminds me to ask a question that’s been puzzling me for a while.

Rook takes minor piece on bishop three. As ISEs go, this type is not too hard to find. Usually, though, it’s a knight on the kingside.

It’s not too difficult to work out why. Chessers castle short way more often than they go long and - unless you’re my friend and fellow blogger Morgan Daniels - knights are pretty much going to be loafing around on f3, c3, f6 and c6 from the get go.

So rook takes bishop on the queenside like Kramnik - Karjakin is quite the rarity. Although do feel free to prove me wrong and point out similar sacs in the comments box if you know of any.

There may be no great mystery about why you get rook takes knight on the kingside on kingside much more frequently than rook takes bishop on the queenside, but one thing I do find curious is when these exchange sacs happen. Theoretical ISEs - moves that are not only thematic, but which  happen at a specific point in a given variation - tend to come from Black. So you get ... Rxf3 in the French Tarrasch, ... Rxc3 in the Sicilian Dragon and others (TISE III), ... Rxb2 in the King’s Indian (TISE IV).

When White plays Rxf6, say, it’s more likely to come in the middlegame. Like Kasparov - Anand, Tilburg 1991 (TISE II) for instance, or Kosten - Gordon from Torquay last year (TISE VI). Where White does have a theoretical ISE - Rxh5 against the Leningrad for example (Standard ISEs) - it seems that Black pretty quickly decides  to avoid it.

Why would that be? Is it simply that starting with a small disadvantage means Black needs to work harder to muddy the waters or is there something else?

2014 ISE count: 31
TISE Index

Sources for ... Rxf3 in the Tarrasch ISEs:-
Andrew Martin: The French Defence (Audio Chess)
Nigel Short: The French Defence
Neil McDonald et al: Chess Pub, French Defence section
Viktor Moskalenko: The Flexible French
Simon Williams: The Killer French DVD


John Cox said...

I think the increased prevalence of ...Rxf3 rather than Rxf6 in the opening is because of the specifics of the way the f-file gets opened early doors; the most likely way is for one side to play P-K5 and the other reply P-KB3, and White tends to get a pawn to e5 more often than Black gets one to e4 because he goes first and gets to put his pawns in the centre.

Anonymous said...

I think the assumption about the relative rarity of rxc6 sacs is probably correct. I did some stucture searches in chessbase. There are more sacs in lines where black castles queenside more often (French, Scotch) and the name Shirov is conspicuous. But no real surprises.

The only thematic rxc6 sac I can think of is in the Bf4 Grunfeld, but the compensation there is central dominance rather than an attack.

Paul C

ejh said...

How are we off for theoretical sacrifices of the exchange by allowing Bxa1 or ...Bxa1?

Anonymous said...

How are we off for theoretical sacrifices of the exchange by allowing Bxa1 or ...Bxa1?

How about lines of the Reti or Catalan where Black takes the c4 pawn and defends it with b5? That can result in the loss of the Rook on a8 as we saw in in an Aronian game in the 2014 Candidates. That usually results in a material imbalance of Bishop and Knight against Rook and a pawn or more, so not a pure exchamge sacrifice.

Another exchange sacrifice example was Mcshane v Aronian from the Tal Memorial in 2012. We all know that advancing the queen side pawns to support a captured c4 pawn can be dubious. The idea is that you play a4, provoking c6. You then have axb5 cxb5 and the cheapo Nxb5 exploiting the pinned pawn on a6. What Luke was able to demonstrate was that taking the Knight on b5, allowing Rxa8, was, in the circumstances where he played it,a very strong exchange sacrifice.


Jonathan B said...

The opening of the file being important makes sense. So you get ... Rxc3 in the Sicilian but not Rxc6 in the English because the c-file isn’’t always going to be open (and when it is Black’s not likely to have castled long).

ejh said...

I'd kind of want to see some actual variations, though, if we're saying that such-and-such a line is theory rather than something that somebody played once. (Which is a discussion in itself, I suppose. Is "theory" anything that has ever been played is a recorded game? Or anything that has been evaluated in a theoretical work, even if it hasn't been played?)

Anonymous said...

I think RdC is correct again, a structure search on ositions with white ba8 and pawns on b5 and c4 gets lots of hits.

At least a hundred master games in the a6 Catalan, and eqivalent Reti. I don't see ant very popullar line, but several positions with a few gm games such as:

1. Nf3 d5 2. c4 e6 3. g3 Nf6 4. Bg2 Nbd7 5. d4 dxc4 6. O-O a6 7. Nbd2 b5 8. Ne5 Nxe5 9. Bxa8 Qxd4

Paul C