Tuesday, June 03, 2008

The Test of Time, The Test of Computers

The puzzles chosen for the "Do you know the Classics?" chapter of Yakov Neishtadt's interesting (especially for £1) book Test Your Tactical Ability have, states the author, withstood the test of time: all of them, in fact, having been chosen from at least thirty years before the book's initial publication in 1981. The position below is from that chapter and dates back even further, to a 1904 clash in Nürnberg between the great Aron Nimzowitsch (white, to move) and his lesser known opponent playing with the black pieces Leopold Hoffer. The solution provided in the book is given immediately below the diagram, so if you prefer not to see it just yet, take a few minutes to try to work out what's going on before looking.

White to play,
and . . . ?

Now, the solution given by Neishtadt is the beautiful 1.Be8!! Raxe8 (1... gxf6 - 2.Ng4!; 1... Qxf2 - 2.Bxf7+ Kh8 3.Ng6 mate) 2.Qh6!! gxh6 3.Ng4, after which black resigned. Beautiful. But, alas, not true. A few minutes with my computer undoes the test of time, and reveals the solution to be more or less entirely flawed. See if you can spot what decades of delighted chess enthusiasts must have missed, a defence for black that leaves him with a more or less won position . . .


Anonymous said...

Is the answer 1...Raxe8 2.Qh6 gxf6 3.Ng4 Qxf2, by any chance?

ejh said...

Would that save (or win) the game?

Tom Chivers said...

Yes, well done Jack. According to Crafty, black is just winning by miles. Black can also take on f2 with the queen in the 1...gxf6 2.Ng4 line, it says...

What a shame!

Anonymous said...

I've seen this one before, even so I didn't find 1 Be8 this time either, so only analyzed 1 Ng4. I concluded it draws. Fritz confirmed my evaluation but also found plenty of things that I missed.

(a) 1...Rfc8? (Funny I looked at this defense first, it's what 1 Be8 is designed to stop, I suppose.) 2 Nh6+ Kh8 3 Qxf7 (3 Qe5 [Fritz] is faster) Rg8 4 Qxg8+

(b) 1...gxf6 2 Qh6 Qxf2 (2...Ng5= [Fritz] I overlooked that one.) 3 Nxf6+ Qxf6 4 Qxf6 Rad8 (trying to win. 4...c3 5 Kxh3 c2 forces white to give perpetual) 5 Kxh3 Rd5 6 e4 Ra5 7 e5 and white gets a perpetual.

(c) 1...Nxf2?! (risky) 2 Nh6+ gxh6 3 Bd1! (found by Fritz, obviously) Qa3 (3...Qa5-c7+; 3...Qb1-b8+) 4 Qxh6 Qd6+ 5 g3 Qxg3+ (silicon monster indeed, but black might as well have played line 'b') 6 Kxg3 Ne4+ 7 Kh4 Nxf6 8 Qg5+ Kh8 9 Qxf6+ Kg8 10 Kg5. Cue S&BCCer comment about trying to defend this against Robin Haldane.

I wonder what the Deutsche Schachzeitung for 1904/1905 had to say. They usually had good analysis.