Recalling a chess miracle

I happened upon the game below on chessgames.com, to which I had submitted it some time ago. It was played in 1978, at a tournament I directed at Jules Stein's Chicago Chess Center, then at 2666 N. Halsted in Chicago. It was, and remains, the biggest upset I have ever seen or even heard of - over 1,000 rating points! White was Mario Spinosa, a junior with a published rating of 1272. Black was Ove Kroll of Denmark, a strong master who was a graduate student in mathematics at the University of Chicago, rated 2315. Ove was one of the strongest players in Chicago, and became a Senior Master a few months later.

The USCF had a big problem with its ratings program at the time, and ratings were lagging several months behind reality. Mario was improving rapidly, and his actual rating was more like 1600. He was particularly strong tactically, as this game well illustrates. He became a master about five years later.

The game features a sharp line of the Nimzowitsch Defense (1.e4 Nc6) that was supposed to be good for Black, namely 2.d4 d5 (2...e5!, favored by Miles and Bisguier, actually gives Black a plus score in the databases, which is why 2.Nf3! is White's most popular move) 3.exd5 Qxd5 4.Nf3 Bg4 5.Nc3!? (5.Be2 is probably stronger, and scores well for White) Bxf3 6.Nxd5 Bxd1 7.Nxc7+. White will be up the exchange, but can he save the beast on a8? The late Hugh Myers in his book The Nimzovich Defense (1973), p. 15, wrote "I believe that 5...BxN! refutes 6.N-B3?. But it is not the simplest thing in the world; some memorizing is required." (For you young whippersnappers who don't know what these strange letters mean, those were the olden days of descriptive notation.) Theory consisted mainly of two amusing miniatures won by Black. Kiss-Barcza, Debrecen 1934 went 7...Kd7?! 8.Nxa8 Bxc2 9.Bf4? (9.d5! Nb4 10.Bb5+ or 9...Nd4 10.Be3! is correct) e5 10.dxe5 Bb4+ 11.Ke2? Nge7 12.e6+ fxe6 13.Nc7 Nd4+ 14. Ke3 Nef5#. Black mated even more quickly in Bildhauer-Janny, Sopron 1927: 7...Kd8! (avoiding a later Bb5+ by White) 8.Nxa8 Bxc2 9.Bf4 Nxd4 10.Nc7 e5! 11.Bxe5 Bb4#!

After Mario played 9.Bf4 (à la Bildhauer, though he had just stumbled into this line and unlike Kroll knew nothing of the theory), Kroll thought a long time and finally rejected 9...Nxd4, perhaps because of 10.Be5 Nc6 11.Bc7+, a line not given by Myers, nor by T. Kapitaniak in Nimzovich Defence (1982), p. 33, but which the engines say is winning for White. Instead, he tried 9...e5!?, trying to gain time to develop his pieces and trap the knight. Myers says that move "is playable," while Kapitaniak simply gives "9...e5!-+" without further elaboration. But as the further course of the game showed, things are much more complicated than that. Kroll's pieces became extremely active, and he could have given perpetual check with 22...Ne2+. But who settles for a draw against a player rated 1,000 points lower? Instead he played 22...Nxh1?!, ironically trapping his own knight in the corner. Mario finally extricated his knight with 26.Nc7 and 27.Ne6!, and wound up the game with a pretty mate of Kroll's king, marooned on a1. It's odd how important the a8, h1, and a1 corners were in this game.