This is the first game from a new series called Smyslov's Sparklers. In the diagrammed position, Smyslov played the very pretty variation 24.Nf5! gxf5 25.gxf5 Nc7 (Smyslov: If 25...Ng5, then 26.Bxg5 fxg5 27.Nxg5+ Kg8 28.Ne6).
The attack continued logically with 26.Rg1 Ne8 27.Rg6. Now after 27...Rf7, Smyslov won with 28.Rbg1 Kg8 29.Rxh6 Kf8 30.Rh7 Ke7 31.Qh5 Kd6 (31...Rc8 32.Ng5!) 32.Bf4+. Kasparov also started his analysis from the diagrammed position, giving no more than a copy of Smyslov's notes.
[FEN "3r1r2/pp1q2bk/2n1nppp/2p5/3pP1P1/P2P1NNQ/1PPB3P/1R3R1K w - - 0 24"]
Neither Smyslov nor Kasparov mentioned the alternative 27...Rh8, which is the computer's suggestion. After 28.Rbg1 Kg8 29.Rxh6 (another possibility is 29.Bxh6 Rh7 30.Qh4 Ne7) 29...Rxh6 30.Bxh6 Qf7, White gets perpetual check with 31.Bxg7 Nxg7 32.Qh6 Rd7 33.Nh4 Ne5 34.Ng6 Nxg6 35.fxg6 Qe6 36.Qh7+ Kf8 37.Rf1 Ne8 38.Qh8+ Qg8 39.Rxf6+ Nxf6 40.Qxf6+ Ke8 41.Qe5+ Kf8 42.Qf6+. This is a long straight-line analysis and could easily be flawed, but it gives Black better chances than the 27...Rf7 played in the game.
Smyslov could have played differently with 27.Bxh6 Bxh6 28.Rg6 Qg7 29.Rxg7+ Nxg7 30.Rg1 Ne7. Black is bottled up, but does White have enough to win?
It is surprising that Smyslov did not discuss the two variations on the 27th move. It is even more surprising that Kasparov said nothing, since he subjected his notes to computer analysis. Is there an obvious refutation that I'm overlooking?
To play through the complete game see...
Vasily Smyslov vs Alexander Kotov, Moscow 1943