The exceptional rook pawn, redux

The universe is so weird. Two days ago, a glitch with the web server meant restoring the site from a backup, and having to redo the formatting for three posts.
One of those posts was the following about the quirk of king-plus-rook-pawn vs. king endings. Wouldn’t you know it — grandmasters Nakamura and Dominguez played one at the US championship today.

[FEN "8/7p/8/8/8/6k1/8/7K b - - 0 1"] { Click or tap on a move in the game text for a popout display board.[#] In the ending of king-plus-rook-pawn vs. king, if the defending side’s king can occupy the queening square, it’s a draw. Whether Black contrives for the pawn to reach the seventh rank with or without check (it matters for the other six pawns; those pawns want to reach the seventh without check), the white king can’t be budged. } 1... h5 ({ }1... h6 2. Kg1 h5 3. Kh1 h4 4. Kg1 h3 5. Kh1 h2 $10 {}) 2. Kg1 h4 3. Kh1 h3 4. Kg1 h2+ 5. Kh1 Kh3 1/2-1/2 * You must activate JavaScript to enhance chess game visualization.

[FEN "8/8/8/8/7p/8/5K2/7k b - - 0 1"] {[#] The rook pawn is exceptional in another way. Since there’s no “i”-file for the black king, it can’t make room for the pawn. } 1... Kh2 2. Kf1 h3 ({ }2... Kg3 3. Kg1 h3 4. Kh1 h2 $10 {}) 3. Kf2 Kh1 4. Kf1 h2 5. Kf2 1/2-1/2 You must activate JavaScript to enhance chess game visualization.

[Event "US Championship"] [Site "Online"] [Date "2020.10.26"] [White "L. Dominguez"] [Black "H. Nakamura"] [Result "1/2-1/2"] [BlackElo "2788"] [ECO "C67"] [WhiteElo "2795"] [FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]{[#]} 1. e4 e5 2. Nf3 Nc6 3. Bb5 Nf6 ( { The Exchange line} 3... a6 4. Bxc6 dxc6 5. d4 ({Fischer demonstrated the potential in} 5. O-O { at two Olympiads. In <a href="https://www.chessgames.com/perl/chessgame?gid=1044340" rel="noopener noreferrer" target="_blank">1970 against Unzicker</a>, he took the theoretically won endgame at move 5, and won it. In <a href="https://www.chessgames.com/perl/chessgame?gid=1044018" rel="noopener noreferrer" target="_blank">1966 against Gligoric</a>, he turned the leads in development and king safety into a crushing middlegame attack. Consequently, every club player who ever claimed to read My 60 Memorable Games took up 4. Bxc6, and experienced no change in their results. } ) 5... exd4 6. Qxd4 Qxd4 7. Nxd4 { is fine for masters who like to play endings — for instance, Capablanca, Karpov, Lasker, Smyslov; four of the best endgame technicians in history.} ) 4. O-O Nxe4 5. d4 Nd6 6. Bxc6 dxc6 7. dxe5 Nf5 8. Qxd8+ Kxd8 {[#] After Kramnik won the 2000 world championship match — partly because Kasparov couldn’t break this Berlin Defense — popularity skyrocketed. Carlsen, Anand, Karpov, Nakamura (it’s a particular favorite of the American champion), and every class C player in the world took up the Berlin when they didn’t mind entering an even endgame at move 8. This didn’t make any difference to the class C players, but grandmasters began to play 4. d3 when they didn’t want to play the Berlin endgame. The C players also picked up 4. d3, too, with no effect. Compare this endgame to that which arose from the Exchange Lopez in the previous note. What’s the difference? A hint: The Exchange Lopez with 6. Qxd4 has never been popular, while the Berlin Lopez is hugely popular and will probably remain so. The difference is that White leads the way in the Exchange line when a draw will be OK; Black defends with the Berlin when a draw will be OK. The positional bases and the drawish character are quite the same, the difference is in the player who’s steering in that direction. } 9. Rd1+ Ke8 10. Nc3 Ne7 11. a4 { New move. 11. h3 is about 60% draws.} 11... Bg4 12. a5 Ng6 13. Ra4 Bxf3 14. gxf3 { There’s some theoretical interest here. White gave up the pawn majority that confers a won pawn ending if all the pieces come off, but evened the bishop/knight balance, gained lots of queenside space, and has two fine rooks. It’s the lessened potential in the kingside pawn majority that will likely keep 11. h3. For example, in <a href="https://www.chessgames.com/perl/chessgame?gid=1816236" rel="noopener noreferrer" target="_blank">Kosteniuk-Stefanova</a>, China 2016, the e-pawn is passed at move 25, and Alexandra (first name basis; I edited the English edition of her book) was winning a few moves later. } 14... Be7 15. f4 f5 { Wow. That’s quite some judgment, handing White the passed pawn, with this caveat: the d-file alone probably won’t serve as a line of entry for the white rooks, e6 (and moreso, e7) is a good blockading square for the black king, and the f4-e5 pawns favor the black bishop over the white. This is why he’s the defending national champion; 15…f5 isn’t a call lesser players will make. Stockfish puts 15…Rd8 — the move that comes naturally — first, with 15…f5 a close second.} 16. exf6 ({ White agrees with Black’s assessment about the passed e-pawn. Stockfish prefers to keep the e-pawn by} 16. a6 {(a more easily-understandable move, right?)} b5 17. Ra5{, with} Nf8 {going to the blocking square e6 next, slight advantage for White. }) 16... Bxf6 17. Ne4 Rd8 18. Rxd8+ Kxd8 19. Be3 { Not the kind of pawn position for 19. Nxf6.} 19... b6 20. axb6 axb6 21. f5 Ne7 22. Ra8+ Nc8 23. Kg2 { This seems like the likelier time for 23. Bxf6, taking care of the threat at b2 (depending on how real you think it is), and gaining a little time, since it’ll take Black two moves to unpin the knight.} 23... Bxb2 24. c3 Re8 25. Kf3 Kd7 26. f6 ( { You can see why White holds} 26. Ra2 { in reserve:} 26... Rxe4 27. Kxe4 Bxc3 { is all kinds of unclear. 26. f6 sharpens the picture on the kingside before committing to a drastic change.} ) 26... Nd6 { By dislodging the white knight, the bind is broken on the queenside.} 27. Rxe8 Kxe8 28. fxg7 Kf7 29. Bd4 Nxe4 30. Kxe4 { Both kings are placed well for dealing with the opponent’s pawn majority.} 30... c5 31. Be5 b5 32. Kd3 b4 33. f4 Bxc3 34. Bxc3 $1 {[#] Computers can make this decision instantly, given that chess has been solved with every position of eight units or fewer on the board. A human playing White has to do some counting — perhaps while under time pressure — and calculate that when Black captures the f4-pawn, the white king either 1) captures the black pawn, or 2) reaches f2 before the black king hits g1 or g2 to stop that. } 34... bxc3 35. Kxc3 Kxg7 36. Kc4 Kf6 37. Kxc5 Kf5 38. Kc6 Kxf4 39. Kxc7 h5 ({ }39... Kg4 40. Kd6 Kh3 41. Ke5 h5 42. Kf4 h4 43. Kf3 Kxh2 44. Kf2 $10{}) 40. Kd6 Kf3 41. h4 $1 ({ }41. Ke5 h4 42. Kd4 Kg2 43. Ke3 h3 44. Ke2 Kxh2 45. Kf2 { is also a draw, but 41. h4 is easier to see, for White knows that when Black captures at h4, White needs king to f4 to hold.} ) 41... Kg4 42. Ke5 Kxh4 43. Kf4 Kh3 44. Kf3 h4 45. Kf2 Kh2 46. Kf1 h3 47. Kf2 Kh1 48. Kf1 h2 49. Kf2 {[#]} * You must activate JavaScript to enhance chess game visualization.

^{Source image for thumbnails.}