Saturday, April 14, 2007

Faulty problem on CTS



This is a problem from CTS (Chess Tactics Server). White just played Rc6. You are black. What should you do? And as all regural CTS-solvers know, you only have about 3-15 seconds (assuming you dont want to end up losing rating points) to decide whats the best move for black? And for those who dont use CTS (oh why dont you woodpushers use it????) I must inform you that in CTS only one move is regarded as a correct one.

You all probably spot very quickly that white's bishop and rook can be forked, and white can't defend them both. So Nd4 looks like a logical way to continue. In addition, some might think that "hell, after Nd4 white still has a rook, so why don't I just play Rxf3 and after Kxf3 I'll still have that fork." This thinking is also very logical but it misses a simple point: white does not have to recapture the rook after Rxf3. White has in-between-move Rxe6!

"Very cool" might one think. So after all, is there only one correct solution (Nd4)? Nope, because after Rxf3 Rxe6 black also has an in-between-move...Rxg3!!. Now after Kxg3 fxe6 or hxg3 fxe6 black has a trivial win. He is pawn up, and simply by pushing his queenside pawns he can force white king to leave the kingside, and then black king can easily pick up all the white pawns on the kingside. Very logical, very simple. Black wins.

So, Rxf3 and Nd4 both win. Now you are back to square one: which one is the correct solution? You have no way to know, so you just have to take a guess. But if you are lucky and you don't see Rxg3!! at all then you might think only Nd4 wins and you choose the correct one. But if you are unlucky (and in this case "unlucky enough" equals GOOD ENOUGH ) enough to see Rxg3 then there is a big chance you end up losing lot of rating points although you found a winning move.

2 comments:

wormwood said...

both are winning, but the difference is Rxf3 wins only a pawn, but Nd4 a piece minus counterplay for white (...Nd4 Rc7). for a computer that's a 'better' win. but for a human, the rook might stir up all kinds of nasty trouble.

I just checked with fritz. the Rxf3 is just the simple pawn endgame win that can't possibly go wrong. but oh boy, the counterplay from Nd4 is way more complicated. I'd take Rxf3 any day before that!

here's how fritz thinks it's gonna go:

[Result "*"]
[FEN "8/p4pkp/1pR1n1p1/4P3/5P2/3r1BP1/P5KP/8 b - - 0 1"]
[SetUp "1"]

{--------------
. . . . . . . .
p . . . . p k p
. p R . n . p .
. . . . P . . .
. . . . . P . .
. . . r . B P .
P . . . . . K P
. . . . . . . .
black to play
--------------}
1... Nd4 2. Rc7 Rd2+ 3. Kf1 Nxf3 4. e6 Nxh2+ 5. Ke1 Nf3+ 6. Kf1 Rd6 7.
Rxf7+ Kg8 8. Re7 h5 9. Kf2
*

Jusah said...

yeah. I would also take Rxf3. there is no point to let white have his rook. also after Nd4 it will certainly take longer to mate white.