zydeco: Really impressive game by Geller -- he gets the initiative early as black and maintains it all the way into the endgame. According to the CG database, 6.....Be6 is Geller's idea. 7.Ne4 seems like a really bad decision -- moving a piece for the second time in the opening, reducing pressure on d5, and blocking a path for both the e-pawn and the g2-bishop. 7.e3, 7.e4, and 7.Nf3 all seem more logical.
11.Nc3 seems to make more sense than 11.Nh4 -- white ends up with his pieces very disharmonized at about move 15.
17....Qb5 seems like a very strong, subtle move, sliding the queen onto the light squares to begin controlling d3 and stepping out of a pin after 18.e4 N moves 19.Be3. 17.....Nb3 looks tempting, but white can get all sorts of complications after 18.Bxb3 (if 18.e4 Nd4) Qxb3 19.e4 Qc2 20.exd5 Qxb1 21.Bh6
I like 19.Be3 better than 19.Bh6 -- d4 becomes a problem for white.
21.Bb3 seems like a bizarre decision -- why weaken d3? 21.a4 and 21.Be3 make more sense.
Geller is impressively centralized after 26.....NBC.
White loses a pawn no matter what after 30.....Rb8. The bishop can't move, so black plays 31.....a5 if the rook stays on b1. After 31.Ra1 Rxb4 32.Rxa7 Rb1 (or Rc1) black threatens both 33.....Rcc1 and 33.....Nxf3+
From his horrible-looking position on move 45, Korchnoi whips up a counterattack that almost succeeds with 46.f5.
48....Rd3+ is the key finesse, forcing the king to a square where the knight can gain a tempo giving check and get back to g7 to cover the black king. If 48.....Rxd1 49.Raa8 I don't think black has anything better than a perpetual check with 49.....Re1+. If instead 49.....f6 white's rooks can chase the black king to the queenside (maybe all the way to c3) and then, after exf6, white has very good winning chances.