Well, in your line black has a better continuation, such as 13...Nxe4, as demonstrated in this high quality correspondence chess game:

Vass wrote on 10/26/12 at 15:58:11:


Whenever I see a move like 9...Bd6 I'd look at every developing move before I would consider 10.h3. Here is a sharp attempt by Black to attack:

10.O-O! Bg4 11.Qg3! e4 12.Qe3 O-O 13.Nxe4 Re8 14.Qg5! Nxe4 15.Qxg4 Rb4 16.Bxe4 R4xe4 17.Qf3 Re1 18.g3! Qd7 19.Qg2 +/=?

Vass wrote on 10/25/12 at 08:50:34:


While the main lines after 8...Rb8 seem pretty good for Black, some of the alternatives favor White. Here Black can regain the pawn but it favors White:

8...Rb8 9.Bd3 Rb4?! 10.Nc3 Rg4 (Rf4 11.Qe2 +/=) 11.Nge4 Nd5 12.Nxd5 Qxd5 (cxb5 13.Bb5ch) 13.Ng3 Qxf3 14.gxf3 Rf4 15.b3 & It doesn't matter whether or not Black regains the pawn, both Fritz 12 and Houdini 2 rate this as +/= to +/-.

Didn't you really get the point? Analyzing with engines is poor judgement if you don't add human knowledge.

@sloughter
If you really want to investigate the Two Knights Defense, first check out the correspondence chess games together with the TKD games of the best OTB players through the years. And it will be not enough.. Check out some books too, starting from the older sources like Panov/Estrin or whatever.. Next you can find some computer TKD games played on longer time controls.. You may even set a match on a long time control between various engines in the lines you're interested in. Then collect a good database with all the games you find in your researches. Make a ctg-book with all the moves! Mark the good moves in it with a green colour, bad moves in red and so on.. Use the statistics that they produce. And then, at last, feel free to use the engines at greater search depth to find some new moves in lines you're interested in!
Posting some long computer lines is not appropriate for a decent analysis, you know.. If you're really interested in a discussion you'll have to post a short line and then discuss the possible continuations here and there. We tried to show opinions once or twice, but you just ignored us posting more computer lines.. That's why we withdrew from the discussion. I'm sure now nobody is following your long lines knowing that they are not of proved reliability.
By the way, computers are not that really good in the opening phase of the game and you don't have to trust their evaluations in full. That's why the engines' authors use human-made opening books for their own engines.
Even the latest (and the strongest for now) Houdini 3 engine will give you that 4.d3 is better than 4.Ng5 without a decent opening book. Make your own conclusions! 

sloughter wrote on 10/25/12 at 03:39:31:


Houdini 2 is yesterday's engine. Houdini 3 is the real deal. 

tony37 wrote on 10/23/12 at 21:36:57:


Agreed: Here is analysis from Houdini 2

After 8.Qf3 h6 9.Ne4 Nd5 10.Ba4! Be7 11.O-O O-O 12.d3 Rb8 13.Nbc3 f5 14.Ng3 Be6 (For a long time Houdini preferred f4) 15.Qe2 Bd6 16.Re1 Rb4 17.Bb3 Qd7 18.Bd2 Nxb3 19.axb3 Rb7 20.Rfc1 & Black has the Bishops, White the pawn. 

Houdini gives the nod to White. Without analyzing the other variations in detail, it is highly unlikely that White can achieve a plus since this is the only variation where White gains a tempo as I define development by retreating the Bishop to a4. The other retreats on the diagonal, Bd3 & Be2 don't gain time.

None of the other retreats gain a tempo so Black should, at a minimum, have full compensation  for the pawn but I'll see what I can find.

What was not obvious to me was that Black could be equal or better in the other variations and still stand worse i.e. I didn't realize I could consistently outplay Fritz 12 in the middle game and believed incorrectly that the starting positions were better for White.