The Weirdest Endgame I've Seen in Years
White is winning here, and it shouldn't be difficult, either. All he needs to do is round up the g-pawn, which requires no special GM technique, followed by running the g-pawn up the board until a Lucena position is reached. Black's king is too far away for White to have any troubles at all, so one would reasonably expect Ricardi, a very solid grandmaster, to collect the full point in 5-10 moves. 71.Rg7 Obvious and good. 71...Kb6 72.Kd6 But what's this? Granted, it doesn't hurt anything, but it's utterly unnecessary at best. White's normal and fully sound winning plan is to grab g4 and promote the g-pawn; there aren't any bonus points to be had for driving the Black king west of the b-file. [72.Rxg4 Rxa7 73.Re4 Ra8 74.g4 Re8+ 75.Kf5 Rf8+ 76.Kg6 Kc6 77.g5 Kd5 78.Re1 and the win is trivial - anyone who knows the Lucena position can win this in his or her sleep. For those who don't know it, here's the winning procedure. First, drag the pawn to g7 (the White king will be on g8). Second, put the rook on the 4th rank; third, bring the king out - the rook will eventually be able to block the checks and the pawn will either promote or cost Black his rook. Thus: 78...Rg8+ 79.Kf6 Rf8+ 80.Kg7 Rf2 81.g6 Kd6 82.Kg8 Rh2 83.g7
End of stage 1. 83...Rh3 84.Re4 And that's stage 2. 84...Rh1 85.Kf7 Stage 3 begins... 85...Rf1+ 86.Kg6 Rg1+ 87.Kf6 Rg2 88.Re6+! (88.Re5?? Rxg7=) 88...Kd7 89.Re5 threatening 90.Rg5, to which there is no adequate defense. 89...Rf2+ 90.Kg6 Rg2+ 91.Rg5
The End.] 72...Ra3 73.Ke5 [73.Rxg4 is an easy winner: 73...Kxa7 74.Rg7+ Kb8 75.g4 Rd3+ 76.Ke6 Kc8 77.g5 Re3+ 78.Kf7 Kd8 79.Kg8 Ke8 80.Ra7 Rg3 81.Ra5 Ke7 82.Kg7 followed by g6 and the Lucena procedure.] 73...Ra5+ 74.Kf6 Kc6 75.Re7 [75.Rxg4 Rxa7 76.Re4 Kd5 77.Re5+ Kd4 78.g4 and so on.] 75...Kd6 76.Kf7 What the...? White is still winning, but WHAT IS THIS? Did White make a side bet that he'd win using the a-pawn? Maybe there's a win to be had by crawling with the king via e8-d8-c8-b8 etc., but there's just no reason for it. 76...Ra3 77.Ke8 Kc6 78.Kd8 Kb6
79.Rg7 Umm, I give up. White has said A, B, C, D and E, but apparently doesn't care too much for the alphabet and starts another project. What I mean is this: White's last 5 moves were all geared towards bringing the king to b8, so why stop now, especially when it wins? [79.Kc8 Kc6 (79...Rxa7 80.Rxa7 Kxa7 81.Kd7 Kb7 82.Ke6 Kc6 83.Kf5 Kd6 84.Kxg4 is only a draw if Black can illegally play 84...Kg6; since he can't, White wins.) 80.Rc7+ Kd5 81.Kb8 Rb3+ 82.Rb7 Rf3 83.a8Q Rf8+ 84.Ka7 Rxa8+ 85.Kxa8 Ke4 86.Rf7+- etc.] 79...Rf3 80.Rxg4 Finally! 80...Kxa7 Okay, we've finally reached the sort of position we could have had 8-9 moves ago, only the White king isn't as well-placed as it would have been had White captured on g4 immediately. 81.Kd7 [One natural way to bring the point home is to combine locking in the Black king with pushing the pawn. Thus although it's not the fastest tablebase move, I'd probably play something like 81.Rg6 so that the pawn can come to g5 and the Black king can't get around White's pieces and in front of the g-pawn. 81...Kb7 82.g4 Rd3+ 83.Ke7 Kc7 84.g5 Rg3 85.Kf7 Kd7 86.Rg8 will be followed by g6, Kg7, Ra8-a1-e1 and so on up to Lucena land.] 81...Kb6 82.Kd6 Rd3+ 83.Ke5 [83.Ke6 is my preference, again to make it harder for the Black king to get to the g-file, but White is still winning.] 83...Kc7
84.Rg8= And after this final error, the position is a draw! After this move, the Black king has time to get in front of the pawn, as happens in the game. It would be wrong, however, to blame the draw on this one error by White; rather, it was the culmination of a long series of inaccuracies and changes of plan. White had lost the thread long ago, and while his winning margin was initially huge, he allowed it to be chipped away a bit at a time until this happened. Simply unbelievable! [84.Rg7+ Kd8 85.Ke6 allows White to keep the Black king cut off and preserve the win: 85...Ke8? 86.Rg8 is mate, while 85...Re3+ 86.Kf7 followed by a general advance of the g-pawn still wins comfortably.] 84...Kd7 85.g4 Ke7 86.Rg7+ Kf8 87.Ra7 Kg8 88.g5 Rb3 89.Kf5 Rb6!
This defensive method is called Philidor's defense (not to be confused with the opening), and although it's not necessary here, because White has a g-pawn (rather than a c-, d-, e- or f-pawn), but it's to be preferred because of its universal applicability. Black keeps his king on the queening file and his rook on the 6th rank until White plays the pawn to that rank; in response, Black plays the rook to the first rank and checks the White king until he can claim a draw, the pawn falls, or White trades rooks and allows a dead drawn pawn ending. As noted, though, Black didn't need to use this method here: with an a-, b-, g- or h-pawn, Black can leaves his rook on the back rank and sleep unless White does something idiotic. [89...Rb8! 90.Kg6 Rc8 91.Rg7+ Kh8 92.Rh7+ Kg8 93.Ra7 Rb8 94.Kh6 Rc8 95.g6 Rb8 96.Ra1 Rc8 97.g7 Rb8 (97...Rc6+ is of course better, but I'm trying to show just how impossible it is for White to win this.) 98.Kg6 Rc8 (98...Rb6+) 99.Rh1 and only now must Black demonstrate sentience with 99...Rc6+ (99...Rb8?? 100.Rh8#) 100.Kg5 and now it's more interesting to determine what Black's losing moves are. (There are only three!) My preferred drawing move is 100...Rg6+ 101.Kxg6 with stalemate.] 90.Kg4 Rc6 91.Kh5 Rb6 [After 91...Rb6 92.g6 is the only way to try to make progress, but in addition to an easy passive-defense draw with 92...Rb8, there's the Philidor method with 92...Rb1 so that 93.Kh6 is easily met by checking the king away, breaking up the mating threat: (93.Ra5 allows White to avoid an eternal series of checks, but just allows a different sort of draw after 93...Rh1+ 94.Kg5 Rg1+ 95.Kf6 Rf1+ 96.Rf5 Rxf5+ 97.Kxf5 Kg7 98.Kg5 Kg8! 99.Kf6 Kf8 100.g7+ Kg8 101.Kg6 stalemate.) 93...Rh1+ 94.Kg5 Rg1+ 95.Kf6 Rf1+ 96.Ke6 Rg1 97.Kf6 Rf1+ etc.] 1/2-1/2