Today the solution is 37.....Rxe4. Now if 38. Rxe4 Bd5 pinning and winning the white rook.

click for larger view

Instead of 28.Rf1, White can try 28.Bd5! exchanging his weak Bishop. Crafty then gives as best for both sides 28...Bc5+ 29.Kg2 Bxd5+ 30.Rxd5 Rxf4 31.Kg3 Ra4 32.Rd2 Rda8 33.Kf3 Ra3+ 34.Ke4 and Crafty can now only find R3a4+ 35.Kf3 Ra3+ 36.Ke4 R3a4+ with perpetual check. If Black plays too aggressively he can actually lose: e.g. 33...Rxa2?? 34.Rxa2 Rxa2 35.Rc8+ Kg7 36.Ne8+ and mates in three.

Appearances can be very deceptive. Playing through this game I thought that White was losing at move 17.

click for larger view

(Eliskases v Borges 1946 17?) Eliskases finds the fine resource 17 Qa3! avoiding the positional blunder 17 Nd6 Rd8 18 Nxc8? giving an inferior ending without compensation (Black controls both the d4 and d5 squares). So ...Rd4 is a powerful positional Black threat. If White plays Rxd4 Black has a powerfl passed Pawn and, With major pieces still on the board, the B of opposite colours favour the attacking side - namely Black.

Black to play (37...?) "Very Easy"

Black wins a piece with 37...Rxe4 38. Rxe4 Bd5.

The move number might help to explain it. If the time control was at move 40, one or both players could be playing quite quickly. It can sometimes take some time to realise that you have a lost position. <beenthere240> explains this well in his post.

I thought that the weak move was 21 f4, because when black did gxf4, he transforms his bad bishop into a good bishop, because the 'g' pawn was blocking it.

Can anyone one give ur opinion about my analysis and about 20 e4 being bad?

thank you.

click for larger view

Rybka 3:

click for larger view

Rybka 3:

<[-0.57] d=19 21.Rd3> bxc5 22.Qxc5 Rd7 23.Rad1 Rb8 24.b3 Rbd8 25.f4 h5 26.Kh2 h4 27.fxe5 hxg3+ 28.Kxg3 Bd5 29.Nc4 Bxc4 30.Rxd7 Rxd7 31.Rxd7 Qxd7 32.bxc4 Qd2

click for larger view

Rybka 3:

<[+0.22] d=21 20...f5> 21.Rad1 e4 22.Qe3 Qa5 23.f3 b6 24.fxe4 Qxc5 25.Qxc5 bxc5 26.exf5 Bxa2 27.Bxc6 Rab8 28.Kg2 Rxb2 29.fxg6 Rxd2 30.Rxd2 Be6 31.Rd3 Be5 32.Nb7 Rxd3 33.exd3 c4 34.dxc4 Bxc4 35.Na5 Be6

click for larger view

(Black to play). Crafty now repeats moves with Ra4+ Kf3 Ra3+ draw. But your suggestion of 34...Rxh3 wins comfortably after 35.Kd5 g5 36.Nb7 Be3 37.Rd3 g4 38.Rcc3 Bf4 39.Rxh3 gxh3 40.Rxh3 Rxa2. The dust has cleared and Black is two Pawns up. My source for these moves are the Crafty links given below.

After 35 Kd5 White has tactical chances, but these only come alive if Black makes the mistake of exchanging on d6 prematurely creating a passed Pawn, and/or allowing the white K to penetrate. If White attacks f7 with Rc7, ...Rd8! is a formidable defensive/attacking move, and ...Re2 attacking the e5 Pawn is also dangerous (there are tactics involving a Rook pseudo- sacrifice on e5 followed by Bxd6+ winning the R back with the N as interest). Meanwhile the B guards f2, so f7 is never under serious attack. So the White "attack" never gets going.

So how come Crafty missed this unless forced to find it (by starting from the above position)? Here are my views/guesses. First,
this online implemenation of Crafty is primarily designed as an ENDGAME TRAINING tool. It expects to be defending positions as Black, so it is not averse to drawing (unless it can see a clear plus). It happens to play middle games very well and with a sharp eye for tactics, but this is a bonus. Secondly, it seems to evaluates positions only either 11 ply or 13 ply ahead (I am not sure which). It also has a restricted link to a Nalimov table base (looking ahead about 35 moves). Thus it plays KQ vs KR perfectly, it plays most tablebase RPP v R perfectly etc., but it plays complex KNN versus KP endings weakly both in attack and defence.

37...Rxe4 nets the bishop, because 38.Rxe4 Bd5 is a deadly pin.