Author: Ed Murak
Date: 22:47:08 12/14/05
Go up one level in this thread
On December 15, 2005 at 00:00:02, Will Singleton wrote: >[d]r1b3k1/pp1p1p1p/2pBq1p1/8/8/8/PR3PPP/3Q2K1 b - - bm f6 > >Black is in serious trouble due to lack of development. >Only one move will prevent disaster. > >Solution times, p4 2.8ghz: > >Zappa 1.1 8 sec >Rybka 30 sec etc. Yes, nice moral. Develop thy pieces. Capablanca used a much simpler example towards the end of Chess Fundamentals, but this one is amusing too. I am weak but it is straightforward to see that black is being strangled. He has two pieces out of play. A second glance shows that things are "absolutely urgent" for black. White has the concept move Re2 which is absolutely obvious in its mortal venom. I think some programs may waste time considering as best moves things like 1. ...Qd5? which of course _even_ I don't consider. Why? Not because the endgame is lost after the exchange (just the opposite, black's pawn majority is such as to make doubling & isolating a very welcome price to pay for removing mate threats), but because of the zwischenzug 2. Rd2 to deflect the WQ, and then 3. Re2 to clear the table. After 1. ...f5?, the first move a human could consider, 2. Re2 and the roof falls in on black. I wonder if any engine can find the shortest (forced) line to checkmate for white from here in reasonable time. 1. ...g5? could be popular with computers as well but suffers from the same defect as the previously considered move. A human who has seen why 1. ...f5? fails won't spend much time with 1. ...g5?. Our own killer heuristic, maybe? The divine spark is seeing that 1. ..f6 gives black a hitherto unavailable resource, denied him in any of the previous variations. To see why 1. ...f6 saves the day, we should examine why 1. ...f5? does not. This I think is how many players at my humble level could proceed- 1. ...f5? 2. Re2 Now things are desperate. The BQ is attacked, the "obvious" 2. ...Qf7 to cover the f8 square is busted by 3. Qe1 - and the dratted white bishop stops the usual way of preventing an 8th rank invasion (K-shuffle). Black can't lose a tempo so the only move for him a human would consider here is 2. ...Qd5 as this attacks the WQ and WB, and the threat of one of these could provide a handy tempo if there is even one breathing space in the onslaught to come. 3. Re8+ is the clear and only continuation, anything else would be nuts for White (an oscillation 3. Rd2 and then back to e2 is pointless; all other White moves lose). Time for a diagram. [d]r1b1R1k1/pp1p3p/2pB2p1/3q1p2/8/8/P4PPP/3Q2K1 b - - 0 3 Black has two moves. (a) 3. ...Kg7 permits the immediate 4. Qa1+ and the game is up. (b) 3. ...Kf7 4. Re7+ [d]r1b5/pp1pRk1p/2pB2p1/3q1p2/8/8/P4PPP/3Q2K1 b - - 0 4 and now whatever black plays, 5. Qa1(+) and black is in an evident mating net. By this time, my patzer-brain has absorbed that Qa1(+) in all variations is the key move ensuring that black is dead. So how do I stop the threat of White being able to play Qa1 without loss of tempo? Simple. Back to the board in front of me- [d]r1b3k1/pp1p1p1p/2pBq1p1/8/8/8/PR3PPP/3Q2K1 b - - 0 1 1. ...f6 which, by blocking the long diagonal (a1-h8), takes the sting out of a future Qa1. If White tries the same attack as before, all moves are forced on each side. 2. Re2 Qd5 3. Re8+ Kf7 4. Re7+ Kg8 (as ...Kf8 loses the queen to the exposed check) Now we are at- [d]r1b3k1/pp1pR2p/2pB1pp1/3q4/8/8/P4PPP/3Q2K1 w - - 0 5 And we see that, due to the pawn blockage on f6, white cannot play 5. Qa1 now, as black would then just gobble up the White bishop. So White is forced to play 5. Re8+ and satisfy himself with a draw. That is how a veritable beginner such as myself could both arrive at 1. ...f6 and see it is a draw, in almost the same time as RYBKA on an average computer but faster than other programs listed. The whole thing is about seeing the sideways Qa1 threat right from the start position, and then 1 ...f6 forces itself on you as the move that must be made and made immediately. There are many other lines so pardon me for concentrating on the ones a human like me would, rather than on ones an engine would feel a need to explore. RYBKA is a terrific program - I am not surprised it did so well in this test. I really recommend it to all. Humanity could tremble when the endgame "module" is finished during the next few weeks. Sorry I interrupted, please go back to engine testing (as I will), or correct my chessic mistakes if you would be so kind.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.