TN wrote on 07/16/10 at 13:52:49:
After 3...a6, White should keep an edge with 4.Nd2 when 4...b5 5.e4 bc4 6.Bc4 is comfortably better for White (usually Black has played ...d6 instead of the worse than useless ...a6), whilst 4...e6 5.e4 d6 6.Bd3 ed5 7.ed5 also feels a bit better for White. 4.Qc2 intending 4...e6 5.e4 ed5 6.cd5 b5?! 7.a4! could be even better.
4.Nd2 looks clumsy to me. It's been tried once in corr though when black changed system again:
[Event "?"]
[Date "1989.??.??"]
[White "Hoedl-Schlehhofer Prof Josef "]
[Black "Sommerbauer Dr Norbert (AUT)"]
[WhiteElo "2324"]
[BlackElo "2397"]
[EventDate "1989.??.??"]
1. d4 Nf6 2. c4 c5 3. d5 a6 4. Nd2 e5 5. e4 d6 6. Bd3 Be7 7. Ngf3 O-O 8. O-O
Nh5 9. Re1 g6 10. Nf1 Ng7 11. Bd2 Nd7 12. g4 Nf6 13. h3 h5 14. N3h2 hxg4 15.
hxg4 Nh7 16. Ng3 Bg5 17. Kg2 Bxd2 18. Qxd2 Qg5 19. Qxg5 Nxg5 (1/2-1/2, 68)
4.Qc2 looks more normal, but haven't been tried in corr (only 77 games with 3...a6). Assuming Black is content with any Benoni, why not play 4...e6 5.e4 exd5 6.cxd5 d6 7.Nc3 g6 - ? I'm no expert on the Benonis, but hasn't white's early Qc2 limited his options here?