As for the original question, there is a Stockfish-AlphaZero game where after 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.exd5 Na5 6.Bb5+ c6 7.dxc6 bxc6 8.Bd3 Nd5 9.Nf3 Bd6 10. 0-0 Nf4 11. Re1 Nxd3 12.cxd3 0-0 13. Nxe5 c5 14. Nc3 Bb7 15. b3 Re8 16.Nc4, AlphaZero played 16...Nxc4 and won. 

Stockfish met 16...Nxc4 with 17.Rxe8+. 

Strangely the book "Game Changer" says that after 16...Nxc4, 17.dxc4(?) Rxe1+(?) 18.Qxe1 Bxh2+ 19. Kxh2 Qh4+ 20. Kg1 Bxg2 occurred in both correspondence games which reached this position and resulted in a draw by repetition a few moves later which which can't be right as 17...Qh4 simply wins for black (I'm assuming those correspondence games used the order 16...Rxe1 17.Qxe1 Nxc4 18.dxc4).

After 28...Qd6 29. Qxd6 Bxd6 30. Rd1 Nb7 31. Bxa7 I am very skeptical that Black has enough compensation for the pawn.

You are right that the position after 34. Rd6 does not seem too bad for Black, at least at first. White has the very strong threat of Nd5-Nf6+, so Black has to play 34...Ne5 to hold his pieces together. Then 35. Bxa7 Nf3+ 36. Kf2 Be8 37. Bc5 is a natural continuation. Perhaps 37...Rb7 38. Rb6 Rxb6 39. Bxb6 Nxh4 40. Kg3 Nf3 41. h3 Ne5 42. hxg4 Kf7 43. Bd4 Nd3 44. Nxd3 cxd3 gives Black drawing chances due to the opposite colored bishops?

Either way, the lines after 34. Rd6 don't seem to be fun for Black, even if there is play left in the position. I agree that it would be incredibly difficult for White to play so accurately for so long when faced with Black's early initative. (Unless, of course, the White player has studied this thread.)

On the other hand, I would like to point out that, from a purely theoretical perspective, 14...f4 seems to equalize pretty easily. To my eyes, the piece vs. 3 pawns endgame that arises is balanced. I don't see how White can set problems there, at least.

Isa I think White has a more difficult time developing after 9..Ng4, his black square bishop specifically and his q-side generally are hard to put on good squares. Black's center really has a cramping effect, and I always like space.

Oh Bologan goes 8... Ng4 9 Ne4 f5 10 Be2 h5 11 h3 fxe4 12 hxg4 Bc5 then he branches between 13 Nc3 and 13 b4, I looked a bit at b4 because it seems after Nc3 Black has some pressure, but I don't think this line is that much stronger than the Nd5 ones. I think White gets his development a bit quicker after Nd5 than Ng4 but again there is not a lot more, I think.

I looked, and Stockfish agrees about Kg1 being the only real try, but I don't like the position very much for white, and like you said I don't know how much of this would ever be seen in an OTB game. I agree about the Lokander book being a very good book. I thought maybe 18 Rf1 was a better try, but it ended up looking worse than your line. Stockfish evaluated your line with a small plus for white, but like + .3 so not really earthshaking and there is a lot of play in that position imo.  I do think from looking at this a bit White has to give at least one pawn back to get untangled, maybe both.

You are welcome Pg 467 is the scene of the crime.

Bologan's Black Weapons

@parisestmagique - You need a browser with flash enabled, or, as IsaVulpes suggested, you will need to unencode the url to get at the pgn.

Hi, all of this looks quite interesting, but when I click the Dowload game in pgn format links, I just get some numeric gibberish such as 
[Comment "auto-generated from "]
1.%20e4%20e5%202.%20Nf3%20Nc6%203.%20Bc4%20Nf6%204.%20Ng5%20d5%0A5.
Did I do something wrong, or is there a simple way to get normal pgn content ?

Pawnpusher wrote on 06/16/18 at 15:53:45:


In which publication did he say that?

So what led you to give up 8..Ng4? Not criticism, but when I looked it seemed pretty sturdy.

My initial clickings through 8. ..Ng4 didn't really inspire me, but a friend of mine pointed me towards another different path, namely 10. ..0-0!?

After 11.Re1 there are obviously loads of possible courses that the game may take here, but after some cursory analysis, I've grown quite fond of the top-scoring (..and least played; naturally 9 games make the "scoring" rather inaccurate, but still) database choice, 11. ..f5!?

I later found that Lokander had mentioned this as well (truly an excellent book!!), and we got to the same lines, but his conclusions are a bit different from mine..
Any additions / corrections / doublecheckings of the following would be more than welcome! The lines are quite wild 

A general overview:


Now for the actual "critical" lines:


Overall from what I'm seeing so far, this is a lot tougher for White to handle than anything else, where he is forced into a tightrope path that is much simpler to play practically for me.
On top of that, it's a sideline within a sideline, so it's very unlikely that White actually has some deep preparation in this line 

Feedback / a way to equalize in the very last line / anything welcome!

8....Nd5 9.h4! Qc7! by Ntirlis.

14.Nc3 is a good found and proves, there are still open minded people, who don't swallow everything.

I looked at Bologan and his recommendation is 8.. Ng4 then 9 Ne4 f5 10 Be2 h5 11h3 fxe4 12 hxg4 Bc5. Bologan writes he didn't choose 8... Nd5 because of 9 h4!

Talking about this position:



This is obviously a rather critical line, Black now being down two pawns, but common literature on it is somehow *very* quiet.
- Sverre Johnsen gives 13. ..Re8 and follows this game for a bit https://lichess.org/AIMpRsAt/black#26 , calling it equal, which it may or may not be; but this looks/feels like a very poor Marshall endgame to me, where one holds sometimes and sometimes doesn't. Certainly I see very little chance of winning this pawn-down endgame, even against a significantly weaker player.
- Sverre Johnsen also offers https://lichess.org/451lHsWS#26 this line, calling it equal after move 21, but omitting that White's 19th (Be3) is seen as a rather large mistake by our silicone friend, which instead gives eg 19.Re8:+ Re8: 20.f4 c5 21.Nc3 cd4: 22.Ne4 Bb8 23.Ng3 Nc6 24.Bd2, and while we got one pawn back and obviously White's structure isn't great, we're still a pawn down and White managed to unravel his pieces (+0.8 at depth 22).
- Lokander notes that the whole line is critical, but still only mentions it in a sidenote (rather than a game of its own / at least having it as the mainline), offers 13. ..c5, and follows this game https://lichess.org/2vgLLhnu/black#26 for a bit, until he finds an improvement where Black sacrifices two Bishops and goes in for a perpetual check. Now that is theoretically fine of course, but in practical terms I cannot play this against someone lower rated than me, and it's also not something I will ever find over the board, so I'd have to memorize the entire variation, else I will just end up losing (similar to Mr.Urkedal.. actually, even the position that is 'a perpetual' I got kinda lost in, when I looked at it without a computer)
- Ntirlis makes a very good point *against* 13. ..Re8, gives 13. ..c5 like Lokander does, and goes on a bit about how computers tend to misevaluate this position badly, leading to wins for White even in correspondence. Now that to me sounds like something that should be covered in some detail.. but again he just gives it as a sidenote, and doesn't even cover 14.Nc3 for White at all (the clearly most natural move in the position & also Database top choice).

So now I sit here, three 1.e4 e5 books in hand, and not a single of them offers a really satisfactory solution to playing the 2 Knights in an Open Tournament. 
Johnsen steers me straight into a bleh endgame, Lokander tells me to learn 20 moves by heart and then give a perpetual (in my last tournament a 1450 played this line against me; I can't give that guy a perpetual!), and Ntirlis doesn't mention the most common variation at all.

I'm perfectly fine playing a pawn down middlegame with sufficient compensation (as in other lines of the 2N); I'm also quite happy with an equal symmetrical position that still leaves some room to outplay a weaker player (like arises out of eg some forcing lines in my Qc2 Nimzo rep). But a perpetual obviously offers very little room for that  , so I'd need something that actually keeps the game going..
An endgame that is "just a draw", but where White had to navigate some significant complications on the path (such as in the Marshall), is also acceptable to me - there I then just have to hope that a lower rated White player would be unable to handle the complications; but again, at the point of the Lokander-Perpetual, White just played the most natural moves in the position, so he may very well find them over the board.

The most logical first attempt for me was 13. ..c5 14.Nc3 Re8, attempting to go back into 13. ..Re8 lines, now that its "critical counter" (14.d4 15.d3) isn't possible anymore, as the Knight on c3 would obviously be attacked, but SF quickly spits out an almost +1 advantage after 15.Nc4 Nc4: 16.dc4: Bg4 17.Re8:+ Qe8: 18.Qf1 (in the same line after 13. ..Re8, Black would now play ..Be2, winning, but here the insertion of Nc3+..c5 clearly favours White)

The next most logical try after 13. ..c5 14.Nc3 (and only real one that I could even think of) would be 14. ..Bb7, but then not to follow up with ..Re8 after 15.b3.
Compi first line is the apparent novelty 15. ..Qf6, with the idea being that after 16.Nf3 Bf3: 17.Qf3: Qf3: 18.gf3: Nc6, White is still 2 pawns up and we have exchanged a bunch of pieces, but this time around his pawn structure is truly wrecked and he is still a bit away from unraveling; so Black can pressure the pawns early on, at least win one of them back, and then still hold a sizeable amount of initiative. Of course the problem here is we gave up the Bishop pair, so if that initiative ever dissipates, the endgames will again be a huge pain..
A sample way of how a game might continue (obviously no guarantee this is perfect play, I just quickly threw out some natural moves that didn't obviously blunder anything) here would be:

Where White retained his extra pawn, but Black's pieces are so active that the pawn doesn't do much, and at least White had to answer some questions on moves 19-25.
This appears quite playable, but obviously I am still not perfectly happy with it. 

15. ..f5, with the idea of just starting a quick attack via Rf6-h6, might make some sense as well and appears like the most normal 'practical try', but the usual stone cold machine just calculates to an advantage with 


So basically that leaves me with the question, what do/would you play against this 13.Ne5:? 
Just one of the "incorrect" lines, in the hopes that White won't the correct path? This 13. ..c5 14.Nc3 Qf6 thing? Something entirely else, that I missed in my own analysis?
Or do you just say "Ok this variation is a draw, that's theoretically sufficient, and against a weaker player I don't go into the 2N at all but rather let him hang himself in the comparably quiet waters of 3. ..Bc5"? That at least is my current state, but I do kinda love the 2N aside from this single variation, so I would prefer to be able to keep playing it..

Any comments would be welcome