Here Morphy tries a little too hard to force an attack and the game fizzles into a lost pawn endgame. The problem seems to be the weakness of his queenside pawns, so that when black 32...Rd2, Morphy cannot defend both of his a and b pawns. One has to be lost. De Reviere's 31...b5 prevents white from saving his b pawn with Nc4.

Add to this the fact that the passed d pawn has insufficient support and it is not hard to see why black sweeps up.

I have a personal theory about Murphy's Law - "If something can go wrong, it will". Its main purpose is to counter the overconfident who can think that they are immune to all risks. So Murphy's Law reminds us that bad things can and do happen, and that we ought to prepare. We should not assume that something will go right.

But there is a flip side to Murphy's law. It emphatically does not mean that the worst outcome will always happen. That is demonstrably not true. If you drop your bread it does not always fall butter side down. The other queue doesn't always move faster. You-know-what doesn't always happen.

The trick is that we tend to remember the bad outcomes, or outcomes which confirm a personal belief. So it appears that they are more frequent than they really are. Coincidences happen, but no more frequently than statistical chance would suggest. We just remember them more.

click for larger view

Rybka 3: <6-piece tablebases>

When I first read your post, I thought you knew the Queueing Paradox, beloved to probabilists, but upon re-reading, probably not, so here it is: loosely, in fact the other queue usually does move faster. Upon your arrival, the probability is 1/2 that you join the slower of two equally sized queues. Because you spend more time in the slower queue, <if you count the time spent in the queue> (rather than just which queue you join) the other queue usually does move faster!

The immediate 6.d4 was considerable.

7.♕b3 leaves white in an equally bad position.

7...♘c6, and white lacks very good replies.

8.g3 was probably a little bit better, by not disassembling white's kingside immediately, however, black is better.

8...♘bc6 was considerable.

10...♗h3! was the best, I feel.

12.♕d2 was considerable, but white's position is already worse.

12...♘bc6 13.gxh6 ♗xh6 14.♗xh6 ♖xh6 was much better.

A quick draw was 14...♕h4 15.♘xf7 (the risky-looking 15. ♗xf7+!? reaches nothing for white, for example 15...♗xf7 16.♘xf7 ♕g4+ 17.♔h2 ♖f8 18.gxh6 ♗f6! seems to be dangerous) 15...♖f8 16.♖f4 ♕e1+ 17.♖f1

The immediate 16.♗xg5 was obvious to play, it is a little bit annoying that Morphy went for the slow 16.♘a3 instead of this very concrete move.

It seems here I opened Pandora's box, inadverently. As a human being I am tempted to play 16...f6 in order to restrict white's central play, and to prepare ♗f7. Well, the digital monster opts for 16...♘bc6, and its principal continuation is 17.♗d2 ♘a5 18.♖ae1 ♘xc4 19.♘xc4 ♖h6 20.♖f2 f6 21.e5 ♔d7 22.exf6 ♗xf6 23.♖ef1 ♘d5 24.♘e3 ♘xe3 25.♖xf6 ♘xf1 26.♖xh6 ♘xd2 27.♖h7+ ♔e6 28.♖xh5 ♖e8. No organic creature (except maybe very strong ones, such as IMs and GMs) would play such lines, ever. Given that it is highly improbable that I encountered a bug or weakness in Stockfish (especially in this depth, it calculated down to 61 plies), we must agree that the deepness of the ideas of this position vastly exceeds my understanding of chess. Black has good hopes, certainly.

19.♘f4 could be considered.

The silicon monster suggests an immediate 19...d5...

24...♗e2 leads nowhere.

27.♖e6 ♘c8 28.♖h6 ♘d6 29.♘e6+ ♘xe6 30.♗xe6 and white is OK.

28.♗xf5 ♘xf5 29.♖h5 ♘xd4 is also possible, because of 30.♖h8, although black's play is easier.

The silicon monster suggests 30...b5, and in this case 31.♔f2?? fails, because the ♘c4+ double attack is impossible after black is taking the ♙b2. The computer's point is that after 31.♖f6 ♖d1+ 32.♖f1 the black ♖ can go to freely to the e-file via d3 or d5, because white's ♖e1 enables black's ♖ to attack the ♙b2 via d2. If any of you are high-rated players, please tell me your opinion, if this is possible to spot over the board.

31...♖d2+ 32.♔e3 ♖xb2 33.♖h7 =, because of 33...♔d6?? ♘c4+ +-, and 33...♖xa2 ♖xe7 =.

32.♔e2 is calm, and it seems somehing like equality: 32...♔d6 33.b4 ♖e5+ 34.♔d3 ♖xe6 35.♖xe6+ ♔xe6 36.♔d4 leads nowhere, 32...♖e5+ 33.♔d3 ♔d6 34.♘c2 ♖xe6 35.♖xe6+ ♔xe6 is also most probably a draw. But how could black prove superiority without taking the ♙e6? Probably 32...♖e5+ 33.♔d3 a5 is the way, but that is super-calmly avoidable by playing 33.♔d2, preventing or at least delaying the black ♖ to enter the first rank. 33...♖g6 34.♘c2 ♖g2+ 35.♔d3 makes equally not much progress for black. 33...a5 34.♘c2 seems also being defendable. Also 33...♔d6. 32...a5 seems to fail to the ever-present 33.♘c2 - it seems to be a very good move in plenty of the discussed positions! Moves like 33...♖g6, ♖f6, a6, ♔b6, ♘f5, ♘g8 are all seem to be insufficient to squeeze out anything from this position. Thus, I am almost convinced that 32.♔e2 holds equality.

35.♘e4+ is probably the most tenacious attempt to defend. 35...♔e5? leads to a draw with 36.♖h7 ♔xe6 37.♘c5+ with a perpetual; while 35...♔c7?? makes white better after 36.♘d6!, with an eventual ♘b7+, losing the black ♘. After 35...♔d5, with perfect play, it is almost surely winning for black. Possible continuations are 36.♖h7 ♖b4 (36...♘f5? 37.e7! =) 37.♘g5 ♔d6 with lots of technical work to do, and 36.♖h5+?! ♔xe6 37.a4, in which white tries to exploit the open position of the black ♔. But again: both of these lines are suspected to be won eventually for black, however, finding a forced win is non-trivial (for me). The immediate 36.a4?! ♖b3+ 37.♔f4 ♔d4! seems to be the easiest path for black.

39...d4 seems to be more straightforward, but the actual 39...♔d7 also wins of course

Both 44...♖e4 and 44...♖c4 wins, but the latter is a longer and tougher road, as the white ♔ enters the e-file. This is very difficult to calculate properly over the board, the silicon monster gives some insane variations. It is a great fun to discover them, do it yourself with or without computer!

46.♔e2 seems to be the only move that is a draw.

Amazingly, 51...♔c3! or ♔b3! wins for black, because after 52. ♖c8 (with a possible check) it doesn't matter how the white ♖ gets on b8, the black ♖ can defend the pawn from h5. The played 51...♖a2 is thus a draw, after 52.♖c8+ ♔b3 53.♖b8 ♖c2+ 54.♔d1 ♖c5 there is no ♙d2+.