Zwaig Variation III

The Norwegian Chess Federation (Norges Sjakkforbund) has opened a weekly column at their website , called 'Fra arkivet' (From the Archive) which I look forward to following. This week they offered a game which was new to me (although I believe I have the scanned magazine somewhere in my shelves):
S.Heim - A.Zwaig, Cht Norway 1975
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 b5 5.Bb3 Na5
The Norwegian variation.
6.0–0
Some day I will return to 6.d4!? which I feel may be underrated.
6...d6 7.d4 f6 8.Bxg8 Rxg8 9.b4
This is given as a novelty so I assume the game was played before Helmers-O.Moen from the same year. 9.a4 b4 10.c3 c5 11.cxb4 cxb4 12.Nbd2.
9...Nc4
9...Nc6 10.d5 Ne7 (10...Nxb4?? 11.a3+-) 11.a4 +=.
10.Nbd2 Nb6!? (Dia)
For 10...Nxd2 see 'Zwaig Variation II'
11.c4!
In 'Zwaig Variation II' I only gave Oim's 11.Qe2 which was the only example I could find in the databases. But even with more than 3 millons games the electronic sources cannot yet compete with paper when it comes to historical games (for ChessBase the critical point seems to be around 1985).
11...bxc4
Curiously Rybka for quite a long time like 11...Kf7?!. Should I re-install the program?
12.a4 Be6 13.a5?
Zwaig's suggestion 13.Qc2 indeed seems to offer White a plus.
13...Nc8! 14.Qc2 Rb8 15.Rb1 Na7 16.Nxc4 Qc8! 17.Be3?!
Zwaig mentions 17.Ne3 Qb7 and it seems Black is holding.
17...Nb5 18.Rfc1 Be7 19.Qd3 Qb7 20.Bd2?! (Dia)
I like this position! With the possible exception of his light-squared bishop, none of Black's pieces seems to be on a 'normal' square. Yet Black suddenly seems better. Zwaig suggests that 20.d5! is best (probably the position is then still '+=').
20...f5! 21.exf5 e4 22.Qe3
22.Qe2 Bxc4 23.Qxc4 d5 -/+.
22...Bxf5!
22...Bxc4 23.Rxc4 Qd5 (23...exf3 24.Re1 c6 25.Rxc6 unclear) 24.Re1 Qxc4 25.Qxe4 Qf7 26.Ng5 unclear.
23.Re1 Qd5! 24.Qf4! Bg6!
Actually 24...g5! 25.Ne3 gxf4 seems even stronger as 26.Nxd5 is fairly harmless.
25.Ne3 Qh5!
Zwaig discards 25...Qf7 26.Qxf7+ Kxf7 27.Nd5! e3 28.Rxe3 Bxb1 29.Rxe7+ Kf8 30.Ng5 Nxd4 31.Bc3 h6 32.Rf7+ with at least a perpetual for White but 32...Ke8 33.Nxc7+ Kd8 34.Bxd4 must be close to winning for White.
26.g4 Qh3 (Dia)

Somewhat surprisingly White is lost thanks to his trapped knight.
27.Nf5
27.Ng5? Bxg5 28.Dxg5 Sxd4 -+.
27...exf3! 28.Rxe7+ Kd8 29.Nh4?
29.Qg3 Qxg3+ 30.hxg3 Bxf5 31.Bg5 Bxb1 32.Rxg7+ is just as hopeless. Black escapes a piece up.
29...Kxe7 30.Qg5+ Kd7 0–1
Notes are based on Zwaig's for Norsk Sjakkblad.