I've really seen a game of such unusual sacs and uneven exchanges.

Both sides promote-and black-who achieved rook,knight,bishop for his original lady proceeds to win with a new one.

[White "azaris"]
[Black "gothicgirl"]

After 3. ♗f4 (I think Trompowsky played 3. ♗h4!?) 3...d5 is very strong and almost deseves the exclam but White still has methods for keeping the initiative. 4. ♘d2 ♗f5 5. e3 e6 6. ♗d3 as played in Adams vs Stohl, 1992 (Adams hasn't always played 1. e4) looks promising for White.

I want to share some of the thoughts on the Tromp from Joe Gallagher in his <Beating the Anti-King's Indians> though, where he does NOT recommend letting White capture on f6. The whole book is of course written from the perspective of KID players. Anyway, so after the starting moves 1.d4 Nf6 2.Bg5 Ne4

3.Bh4:

<This retreat was once quite popular but is now rarely seen. This is no doubt due to a number of heavy defeats which White suffered in the latter half of the 1980s. The drawback of Bh4, as opposed to Bf4, is that Black is able to force the exchange of knight for bishop and consequently obtains counterplay on the dark squares. This may sound contradictory to those of you who have just read the introduction to this chapter where I stated that 3.Bxf6 should be avoided, but the difference there is that the pawn structure makes it hard for Black to acheive active play.> He gives the main continuation as 3..g5 (3..c5 right away is equally good, transposing) 4.f3 gxh4 5.fxe4 c5 6.e3 Bh6 7.Kf2 cxd4 8.exd4 Qb6 9.Nc3 e6! <Taking the b-pawn is a risky business..> and <Anyway the main purpose behind ..Qb6 was to create pressure along the important g1-a7 diagonal which happens to number the white king amongst its occupants. The text, of course, maintains the position of the queen by preventing Nd5.>

3.h4:
<The 'h4 Tromp', introduced as a stopgap between the fading 3.Bh4 and the rising 3.Bf4, has been the subject of much ridicule and laughter over the years. From a purely chess point of view it is hard to believe in this line but in practice White has actually scored quite well. Perhaps this is due to the psychological impact that a move such as 3.h4 can have on the opponent. On seeing 3.h4 Black probably begins to feel very confident (how can he play such rubbish?) but he may also experience some difficulty in taking the game seriously; not a good combination. I can recall one of the first games in this line, Hodgson-Gufeld, Hastings 1986/7. Black played the opening quite well and by move 15 was a pawn up for nothing. But soon afterwards things started to go wrong; he castled queenside rather riskily and White was able to sacrifice a piece for unclear complications after which 'Big Eddie' proved to be no match for a Hodgson in his element, eventually overstepping the time limit in a lost position. What made this game so memorable though is the reaction of Gufeld to his defeat. For the next hour he remained alone on the stage, a tragic figure with his head clasped in his hands and for the rest of the tournament he could be heard explaining to anyone who would listen that this game was not chess and how disgraceful it was that someone could conduct a game in the way Hodgson had just done.> Gallagher gives 3..c5 as his main recommendation but feels that 3..d5 probably is the simplest way to equalize. After 3..d5 4.Nd2 <4..Bf5 is the most solid, no-nonsense approach to the 'h4 Tromp'.>

a) 6.Nd2
b) 6.d5 (most usual)

a) 6.Nd2 cxd4 7.Nb3 Qb6
a1) 8.Qxd4 <8..Qxd4 9.cxd4 d5 is probably a little better for White; practice suggests he can transform his slight lead in development into a space advantage or something more concrete>, so Gallagher gives 8..Nc6! 9.Qxb6 axb6 which <leads to a more dynamic position where Black's central superiority should compensate for his weakened queenside> and it goes on, White's most important choices are 10.Nd4 and 10.Be3.