Mind you: I'd be quite encouraged by the 22%. This line ain't a 'gimme' for White!

Still, a3 seems it will cramp the better bishops(black) style or the trade off would increase your development and proceed with the minor win B for N,,,,..... perhaps at this point it is a players choice in knowing and acting on their personal chess characteristics.

7.a3 Bd6, and I don't black's bishop is extremely cramped. Just a little more than on b4, since her own pawn is on c7. Still, my personal style would say keep the bishop. Knights are better in cramped positions, bishops are better in open ones. The thing is though that cramped positions usually become open, so if the game lasts that long, bishops are better!

I dunno why it would be questioned - f7, a weak point in Black's position, especially with moves lokw Ng5 in the horizon

Relying on databases is fine, and some comment about them (specially for moves that, otherwise, may look strange) is ok. But you are just commenting each single move with it, and "only" with it.

I'm not sure that comment is fair. My comments also involve other factors, such as the post-match computer analysis.

If there are only a few games in the database, such percentages mean nothing in my opinion

Computer is right. What is the key to the black's defeat? King's exposition in the center of the board. The Qe7 plan is a good one, since white has not another pawn to support the e (unless he moves the f3 knight and plays the f pawn, but this will weaken a lot his king). So black's plan would be good, but for white having the e1 square defended by the f3 knight, so the queen is not a good piece to attack e. MOSTLY because the king is behind. So, o-o is necessary, here, before black goes on attacking.

I think on second thoughts damalfi is right, but you can hardly blame Jo for copying the mistake all the others had made. As I said, no one in this position had played 0-0 they had all played Qe7. One case where copying the databases is not a good idea!

It's not matter of database anymore. The temporary sacrifice on f7 "must" have been foresaw by black, and it wasn't. That's all the point. Black has played against the e pawn, without avoiding its advance, and should have been prepared to follow the line having thought where to send the knight (if g5, white plays Re1, then h3 and the knight has to go to to a6...)

Yes, I would think are right and black didn't see the sacrifice on f7. Having said that I think black has a slight disadvantage if she plays knight to g5 instead. The mistake was made when she didn't castle and instead followed the move played by others in this position. After Ng4, Vescovi Giovann… (2617) vs. Bachmann Axel (2505) went 10. a3 Ba5 11. h3 Ngxe5 12. Nxe5 Nxe5 13. Re1 O-O 14. Qg3 Re8 15. b4 Bb6 16. Kh1 Bxh3 17. f3 Be6 18. Rxe5 Qf8 19. Bd3 f6 20. Rh5 g6 21. Rxh7 1-0 and Thorfinnsson Bj… (2364) vs. Semcesen Daniel (2349) went 10. h3 Ngxe5 11. Nxe5 Nxe5 12. Re1 O-O 13. Rxe5 Qxe5 14. Qxb4 Qe1+ 15. Bf1 Bf5 16. Qb5 c5 17. Nb3 b6 18. Be3 Bb1 19. Bf4 Qe4 20. Bd6 Rfd8 21. Nd2 Qc2 22. Bc7 Rdc8 23. Bf4 Qf5 24. g3 g5 25. Bd6 Qg6 26. Rxb1 Qxd6 27. Bg2 a6 28. Qd3 Ra7 29. Qf5 Re8 30. Ne4 Qe5 31. Nf6+ Kf8 32. Nd7+ Rxd7 33. Qxd7 Kg7 34. b3 Re6 35. Qd5 Qf6 36. Qc4 Qf5 37. Rc1 a5 38. Bf1 h5 39. Qd3 Qf6 40. Qd2 h4 41. Bc4 Re5 42. Re1 hxg3 43. fxg3 Qf3 44. Kh2 Re3 45. Rxe3 dxe3 46. Qe2 Qe4 47. a4 f5 48. Qg2 Qe5 49. Qb7+ Kh8 50. Qf3 Kg7 51. Kg2 Qb2+ 52. Qe2 Qd4 53. Qd3 Qb2+ 54. Qe2 Qd4 55. g4 Qe4+ 56. Kg1 f4 57. Qd3 Qf3 58. Qd7+ Kh8 59. Qe6 Kg7 60. Qf7+ Kh6 61. Qh5+ Kg7 62. Qxg5+ Kh7 63. Bd3+ Kh8 64. Qf6+ Kg8 65. Bc4+ Kh7 66. Qh4+ Kg6 67. g5 Qd1+ 68. Bf1 e2 69. Qh6+ Kf5 70. Qh7+ Kxg5 71. Qe7+ Kh6 72. Qe6+ Kg7 73. Qxe2 Qxb3 74. Qg4+ Kh8 75. Qh4+ Kg7 76. Qxf4 Qc3 77. h4 Kh8 78. h5 Qg7+ 79. Kh1 Qb7+ 80. Kh2 1-0, so I would have had a pleasant choice of alternatives.