AylerKupp: <Penguincw> That's why you need to be careful with statistics. Surely RPP vs. R would win in the very large majority of cases but everything depends on the actual position. And in the Endgame Explorer you cited there were 2,668 games listed with 546 (20.5%) of the games drawn which is not an inconsiderable amount. And John Nunn in his book on rook endgames indicates that the most common exception to the case of RPP vs. R winning is when the pawns are blockaded as was the case in this game. Dvoretzky in his endgame book makes a similar observation, indicating that the best chances for a successful defense exist when the (inferior side's) king blocks the (superior side's) pawns as was also the case in this game.There were even 9 games listed (0.3%) where the lone rook won! This made me curious so, having some time on my hands, I looked at those games listed and here is what I found. I used the on-line Nalimov tablebases at http://chessok.com/?page_id=361 to determine what the game's outcome should have been.
<amaurobius> Yes, wonderful things, tablebases.
1. Colle vs Ahues, 1930. Shown as the inferior side winning but in reality the superior side (White) won.
2. F J Sanz Alonso vs Granda-Zuniga, 1991. First I thought this was the famous M Ortueta vs J Sanz, 1933 in which Sanz's pawns overcome N+R which if you haven't seen it, you must. But in this game it also shows the superior side (Black) winning.
3. K Urban vs D Rogic, 1997. Yes, the inferior side wins but I suspect that the superior side (White) lost on time since the final position is a draw.
4. A Breier vs J Gottschalk, 1998. Another suspected loss on time since the final position is a draw even though White is 2 connected and unblockaded pawns up.
5. Piket vs Anand, 1998. Incredible blunder by Anand (Black), probably in time pressure, putting his rook en prise while 2 pawns up, allowing the inferior (but not in the final position!) side to win. The game is otherwise a win for the superior side.
6. A Shaley vs K Mueller, 1999. Weird. White is up 2 pawns and the final position is a win for White. Yet the game score is 0-1 with a cryptic note for the final move 58.Kf6 which says "Forfeited. 1-0 for rating." Whatever that means.
7. Gelfand vs Shirov, 2001. A blindfold game which perhaps explains the superior side (White) putting his rook en prise in a drawn position.
8. R Kempinski vs V Sergeev, 2005. An apparent loss of time by the superior side (White) after first losing a pawn in a won position. And the final position was still won for White even though he was only one pawn up.
9. Y Visser vs J Smeets, 2006. The oddest, funniest, and saddest game of all. First the inferior side (White) puts his rook en prise in a lost position. In time trouble, Black doesn't notice that he could capture White's rook with his rook, makes a king move, and allows White to capture his rook, leaving him in a lost position. Presumably enraged at himself Black plays on, first losing both his pawns and playing on until mate, forcing White (rated 2480!) to demonstrate that he's capable of winning a KR vs. K endgame. Per the kibitzing in the game the loser (Smeets) "stormed out of the playing hall closely followed by a huge bang. Apparently he karate-kicked a door open."
So, we have to remember to be careful with statistics, particularly with small sample sizes, since things may not always be what they seem. And never give up on a seemingly lost game.
Kibitzing
