This position is from the game Kortchnoi-Kasparov Lucerne (ol) 1982, after Black's 22nd move.
Kortchnoi here played 23.Bd2, and lost brilliantly: 23...Qb2 24.fe Be5 25.Nc4 Ng3 26.Rf8 Rf8 27.Qe1 Ne4 28.Kg2 Qc2 29.Ne5 Rf2 30.Qf2 Nf2 31.Ra2 Qf5 32.Nd7 Nd3 33.Bh6 Qd7 34.Rc8 Kf7 35.Rh8 Kf6 36.Kf3 Qh3 0-1.
Leonid Shamkovich annotated the game for issue #37 of the Players Chess News. He wrote after Kasparov's 22...Rbe8: "Now all of Black's pieces participate in the attack. White's position is hopeless."
Harry did not agree. He wrote a letter to Larry Christiansen, the editor of Players Chess News, suggesting an improvement for White - 23.Kg2!
Harry looked at 2 tries for Black:
"1). 23...Nf7. We think that White holds after
24.Qd3 even after
24...Ng5!? .
2). 23...Qd8. Here we think that White plans
24.Na3-b5. In this line White plans to freeze his pawns on f4 and
sit tight, thus 24...Qc8 25.Rh1 should Black retreat
25...Nf7 White will try either 26.Qd3 or
26.Qc4...White hopes to seek action on Ra7 with a timely sac on
f7 in some variations. We have not examined an early (or
later)...c4."
Christiansen agreed with Harry, and added in his reply that 23.Kg2 Nf7 24.Qd3 Ng5 25.fg Rf1 26.Nc4 gave White at least a draw: 26...Qd8 (26...Ref8 27.Qf1) 27.Qf1 Bc3 28.bc Re4 29.Nd6 Bh3 30.Kh3 Qd6 31.Ra8 Kg7 32.Ra7 Kh8 33.Kg2. "23.Kg2! casts doubt on Black's earlier play."
Arnold Denker weighed in. He believed an improvement for Black in Christiansen's line was taking first with 24...Bc3 (instead of 23.Kg2 Nf7 24.Qd3 Ng5), and then after 25.bc Ng5 26.fg Rf1 27.Qf1 Re4 28.g4 Qd8 29.gh Qf8 30.Kg2 Bh3 31.Kh3 Qf5 Black won.
Christiansen replied to Denker that better for White in the above line was 25.Nc4:
e.g. 25...Qb8 26.bc Ng5 (26...Qc8 27.Rh1) 27.fg Rf1 28.Qf1 Re4 29.g4 Bb5 30.Qf3 Rc4 31.gh Qe8 32.hg hg 33.Bd2.
Shamkovich by this time struck back. He agreed that White's position after 22...Rbe8 was not hopeless, but believed that Black had plenty of compensation for the pawn. He analyzed Harry's 23.Kg2 in depth:
"1). 23...Nf7 24.Qd3 (24.Nc4 Qb4 25.Rd3 Ng5) 24...Bc3 25.bc Ng5 26.fg Rf1 27.Nc4 Qb8 (27...Qb5 28.Qf1 Re4 29.Ra8; 27....Ref8 28.Be3) 28.Qf1 (28.Kf1 Re4 29.g4 Bg4 30.hg Qf8-+) 28...Re4. In spite of considerable simplification, Black's position is still better because his pieces are very active and well coordinated. Now Black threatens both 29...Bb5 pinning the Knight and 29...Qe8 with an attack. If 29.Ne3 , then 29...Qc8 30.g4 Nf4 31.Kf3 Re7 32.Nf5 gf 33.Bf4 fg and Black is much better."
"2). 23...Qd8 24.Nab5 (The Boylston Chess Club's recommendation. If 24.fe, then 24...Rf1 25.Qf1 Be5 with a strong attack e.g. 26.Qd3 Ng3 27.Bg6 Qh4). Now the Boylston Chess Club has considered 24...Qc8 25.Rh1Nf7 26.Qc4 +=, but stronger in my opinion is 25.g4 Ng4 26.Ng6 , or 25...Nf7 26.gh Bh3 27.Kh2 Bf1 28.Qf1+=. Therefore I suggest 24...Nf7 25.Qc4 (25.Qd3 Qc8 26.Rh1 c4) 25...Ne5 agreeing to a draw after 26.Qe2 Nf7 (26.fe Rf1 27.Qf1 Be5 is too dangerous for White). The last variation shows that the second line (23...Qd8) is less dangerous for White than the first one (23...Nf7). Anyway Black has in this complicated variation of the Modern Benoni - to my mind - more than sufficient compensation for the pawn, because of his great possibilities for combinative attack."
Harry's last word on this debate was:
"After Arnold Denker's (23.Kg2 Nf7 24.Qd3 Bc3) the
suggested 25.Nc4 seems to fail to 25...Qb5 26.bc Re4
followed by 27...Re8.
The Boylston Chess Club now suspects that Kasparov would be happy
to replay the position after 23.Kg2. We share your view that
Black's piece activity equals the pawn.
-Harry Lyman for the Boylston Chess Club"