It was another good day for the favorites, as top seeds Hikaru Nakamura and Gata Kamsky both won with White and share the lead in the 2012 U.S. Championship with 2.5/3.
Nakamura used the trendy 9.g4 Yugoslav Attack against Ray Robson's Dragon Sicilian. Robson was the first to innovate with 16...Nb6 (16...Rad8 had been usual in the few games to reach that point). That more or less committed Black to an exchange sac three moves later, but although Robson had a bishop and two pawns for the rook, his tripled e-pawns negated some of that compensation. White was always better, and Nakamura finished things off with a well-calculated mating attack. He allowed Black to promote a pawn, with check, no less, but Black was helpless against the "blind pigs" (a pair of rooks on the 7th rank).
Alexander Onischuk is the U.S.'s reliable third seed, so Kamsky's win was significant. Kamsky's usual anti-theory approach proved effective once again as he used the London System not in a bid for an advantage (at least not an overt bid) but to create a position where the best player (him) could fight for a win. Typical Kamsky, he made lots of prophylactic moves (6.h3, 9.a3, 12.Bh2) and still managed to have some pull after 16 moves. Once Kamsky played 20.Na5 several imbalances were about to favor him; in particular, the bishop pair and a superior pawn structure. It isn't always better to have the two bishops, but when the opponent has ragged pawns it can be a huge advantage.
The crisis came after Onischuk's 30th move, 30...h5. Kamsky had been massaging his advantage, trying to make something of his bishops and Black's weak d-pawns, but 30...h5 was the proverbial red cape to a bull. Kamsky sacrificed a pawn for a kingside attack, and although it might not have been objectively best it put Black under a lot of pressure near the end of the time control. 38...Ne6 may have maintained equality, but Onischuk's 38...Qd5 was inaccurate and his next move a blunder. After 39.Rb8 g6? 40.Re8! Rc7 41.Bxg6! Black's position was collapsing, and 43...Rb7? was the final straw. Maybe Onischuk initially thought White was forced to go for a perpetual, but after 48.Qe8+! he resigned, seeing that after 48...Kg7 49.Qd7+ Qxd7 50.exd7 Black would be unable to prevent White from making (and keeping) a new queen.
In other games...
Lenderman-Kaidanov was an unusual Tarrasch Defense where Black was better most of the way but couldn't figure out how to convert it. One possibility is 26...d4, which isn't winning but gives Black an important trump. After missing or neglecting that chance, the game was balanced and soon finished in a draw.
Hess seemed to have the advantage on the white side of a trendy Kan line against Ramirez, but with some neat tactics Ramirez reached a drawish, and eventually drawn, ending with rooks and opposite-colored bishops.
Stripunsky got on the scoreboard, defeating Akobian with White in a Tarrasch French. Akobian's 7...Nfd7 was highly unusual, moving the knight before it was kicked with e5, and after 9...a4 they reached a position that had only been seen in five previous games. In four of those games White played the pedestrian 10.Nbd4, but Stripunsky followed the fifth with 10.h4!?, daring Black to take the piece on b3. The idea, of course, is that White has 11.Bxh7+, when accepting the Greek Gift would be fatal. 11...Kh8 might be playable, but Akobian nixed the whole thing with 10...f5. This reduced White's attacking options somewhat, but White enjoyed an edge thanks to his dark-squared control. Black sacrificed a pawn for some draw-compensation, but that left once he sac/blundered the e-pawn on moves 25-28. After 30.Nxa4 White was three pawns up, and that was more than enough for Stripunsky to win.
Unfortunately for Seirawan, his miseries continued, and now he's 0-3. He played very quietly with Black against Shulman, looking for or at least not resisting a draw, and after 22 moves it seemed likely (though not guaranteed) that he would achieve it. The knight swap on move 35 was a bit of an error though; had Black played 35...Bd5 or 35...g6 (36.Nxd6 Kxd6 37.Ke4 Bd5+ pushes the king back) he'd have enjoyed equality and the draw would have been in sight. Still, it seemed that a draw was most likely, but after 37...Bg8 that ceased to be the case. Instead 37...hxg5 38.fxg5 g6 39.Kf4 Bf7 holds down the fort on the kingside, and now Black can push on the queenside to keep things balanced. (40.Kg4 Ke5 41.h5 gxh5+ 42.Kh4 c4 43.Bxh5 Bd5 44.Be8 Bf3 should hold.) Black's last two moves make a bizarre impression, but the problem is that he's in zugzwang. Most of his pawn moves are terrible (and he'll soon run out in any case), his bishop can't move and he can't afford to let White's king penetrate to the kingside. So he retreated to f8 in hopes of playing ...Bg8, but Shulman's 42.Kf6! stops that: 42...Bg8 43.Bxg8 Kxg8 44.Ke7 followed by f6-f7-f8Q. So Seirawan resigned.
Round 4 Pairings:
Seirawan (0) - Ramirez (1)
Robson (.5) - Hess (1.5)
Onischuk (1.5) - Nakamura (2.5)
Kaidanov (2) - Kamsky (2.5)
Akobian (1.5) - Lenderman (2)
Shulman (2) - Stripunsky (1)