Thanks for the analysis Glenn and Mark!

I have played this position quite a few times. I think 5.Nc3 gives White better winning prospects than 5.h4 and is easier to play. Only Black can play 5...g4 forcing White to sacrifice a piece. This is not as bad news as might be thought since White can transpose to a line of the Rosentreter gambit with 6.Bxf4 gxf3 7.Qxf3 (i.e. 3.Nf3 g5 4.d4 g4 5.Bxf4 gxf3 6.Qxf3 d6 7.Nc3)

In the lines with ...Bg7 I prefer 5.Nc3 Bg7 6.g3 g4 7.Nh4 f3 8.Be3 over 8.h3.
I have had some games that for obvious reasons are not in the databases. I think 8...Bf6 is critical and can be met by either Burgess' (from 101 Chess Opening Surprises) 9.Nf5 Bxf5 10.exf5. Here Black can annoy us with 10...Bg5!?

So I played 9.Qd2 intending to follow Furhoff (his games are a great source of inspiration in this line)-Eriksson Helsingborg 1991 with 9...Bxh4 10.gxf4 Qxh4+ 11.Bf2 Qh6 12.Qxh6! Nxh6 13.Nd5 with compensation.
Jensen-Ferlito, email Ch League 2002 saw instead 9...Nc6 10.0-0-0 Nge7 11.Qf2 preparing e4-e5 and h2-h3. I eventually lost due to later mistakes.
After the game I analyzed 11.Nd5 as an improvement. E.g. 11...Nxd5 12.exd5 Ne7 13.Bh6 preventing Black from castling and blocking h7-h5. (13.Bc4 might be ok as well) 13...Nxd5 can be met by 14.h3

Dear Uberdeker,

the 2.a4 Anti-French is certainly not the KG, but still three remarks can be answered here.

"Could you tell me why you wish to assign me to the dogmatics' doghouse?
So who is being dogmatic ?
I do not consider myself to belong to any particular school of thought."

"the sac must be incorrect" because Your System tells you so, shows you being a dogmatic member of the anti-gambit school. See also your remarks in threads on the Morra Gambit and the BDG, which you dismissed much too easy.

Your remarks on Spielmann/Nimzo show lack of historical knowledge at best; you are just repeating the biases of Nimzo. 1)You obviously have not played through that Rubinstein game yet. 2)You can only maintain Nimzo being praising, if you equal sympathetic to belittling. Nimzo's notes on the game Nimzo-Spielmann, Carlsbad 1929, say it all. 3)Can you explain, what Nimzo's great successes consist of, besides Carlsbad 1929? You mean Nimzo's zero wins against the classical player Capablanca? His bad results against Alekhine? Note, that the primitive attacking player Spielmann beat both twice. 4)Nimzo's dogmatism was not less than Tarrasch'; this explains his failures after Carlsbad 1929.

As the Dutch proverb goes: one fool can state more, than ten wise men can refute. This withheld me from answering your last post until now.
As I don't like quarreling without giving moves, here we go:

I maintain 8...Bf6 9.Nxf3 is unclear. Just to make clear, that once again you underestimate the meaning of dynamic factors, I will give you some sample lines:
a1)10...Nc6 11.o-o-o Be6 (nice trick; 12.d5 does not work) 12.Qh5! Qe7 13.e5. Here 11...Bg5 is an improvement.
a2)10...Nc6 11.Bc4 Be6 12.Bxe6+ fxe6 13.Qh5+ Kd7 14.e5.
b)10...Bg7 (to take the sting out of e4-e5) 11.Bc4 Be6 (Qf6 12.Qg2) 12.Bxe6 fxe6 13.Rf1 (13.e5 Nge7 14.o-o Rf8) Qf6 14.Qh5+ Qg6 15.Qb5+ Nd7 16.Qxb7 Rb8 17.Qxa7 is murky; White gets quite a lot of pawns.
c)10...Be6 11.e5 dxe5 12.Qxb7 exd4 13.Qxa8 dxe3 14.Rd1 Nd7 15.Qxa7.
I do not pretend at all, that this will be the last word.

Matters are not so clear either after 8...Bf6 9.Nf5 Bxf5 10.exf5 d5 11.h3 h5 12.hxg4 hxg4 13.Rxh8 Bxh8 14.Qd2 and 15.o-o-o. You call that passed pawn on f3 a monster. Well, I have learned, that the safety of the king is most important of all. And it is still a long way to castling queenside, isn't it? Eg Nc6 15.o-o-o Qd7 16.Bg5 f6 17.Bf4 o-o-o 18.Nb5.

8...Nc6 9.Qd2 Bf6 10.o-o-o as Bxh4 11.gxh4 Bd7 12.Qf2 intention 13.h3 leaves Black with the question how to finish his development; or Nge7 11.Qf2 Bd7 12.h3 h5 13.Re1. White also has 10.Nd5 Bxh4 11.gxh4 Be6 12.o-o-o Bxd5 (Qd7 13.Nb4!?) 13.exd5 Nce7 14.Bb5+.
Alas for Black it is not so rosy as you insist to think. Let me remind you of your earlier remark about that Shirazi game. You were wrong and that should have teached you a little modesty. Dogmatism and modesty hardly go along though.

No, I will not advocate the Allgaier Gambit proper - I already have explained why I have removed 3.Nf3 from my repertoire. You already have admitted, that Black, to play for a win, must allow White to play an improved version after 3...d6 4.Nc3 g5 5.h4 g4 6.Ng5 h6 7.Nxf7 Kxf7. Two corr games you might not know:

Terry - Barlow,B [C39]
corr GBR, 1988
1.e4 e5 2.f4 exf4 3.Nf3 g5 4.h4 g4 5.Ng5 h6 6.Nxf7 Kxf7 7.Nc3 d6 8.Bc4+ Kg7 9.d4 Be7 10.Bxf4 Bxh4+ 11.g3 Bg5 12.Qd2 Bxf4 13.gxf4 Bd7 14.0-0-0 a5 15.f5 Be8 16.Nd5 Nd7 17.e5 Ra6 18.Nf4 1-0

Doplmayr,F - Tarmak,M [C39]
corr 1995
1.e4 e5 2.f4 exf4 3.Nf3 g5 4.h4 g4 5.Ng5 h6 6.Nxf7 Kxf7 7.Nc3 d6 8.Bc4+ Kg7 9.d4 Be7 10.Bxf4 Bxh4+ 11.g3 Bf6 12.Qd2 Nc6 13.Be3 Bd7 14.0-0-0 Bg5 15.Rdf1 Be8 16.Nd5 Bg6 17.Nf4 Bxf4 18.Rxf4 h5 19.Rhf1 Qe7 20.Qf2 a6 21.e5 Rd8 22.Rf6 Nh6 23.Bg5 Qe8 24.Rxg6+ Qxg6 25.Bf6+ Kh7 26.Bd3 Qxd3 27.cxd3 Rdf8 28.Qf4 Rhg8 29.Qe4+ Rg6 30.Rh1 1-0

Of course I am aware of the game Bosboom-Dautov, Lippstadt 1991. It might be an entertaining exercise for you to find out how White can improve. Hint: distrust backward development, even when it overprotects. Maybe that will cure you of your systematical underestimation of attacking chances, which already happened three times now.

OK, good fellow, show up with some decent analysis, more than 1½ ply deep and a superficial conclusion like "it seems reasonable ....". Then next time I will give you a straightforward refutation of Hanstein Gambit main line 8.g3. It is not in the books and I am quite sure you will like it.

Ch.gr.

Have you already looked at 8.Be3 Bf6 9.Nxf3 ? Don't dismiss it too early - f7 remains weak, Bf6 is a bit shaky, castling queenside is a long way to go and White only needs two moves to play e4-e5. I have not got further than the stereotypical "unclear" to me. For this piece sac it is mandatory not to play 8.h3.

Nimzo on Spielmann in the Carlsbad tournament is about the most unreliable source you can think of. Nimzo just sought a cheap revenge for his painful losses against S in Semmering 1926 and Berlin 1928. This influenced to a great extent Nimzo's comments on Nimzo-Spielmann, Carlsbad 1929. That is a pity, as it is one of the best games played between 1918 and 1930.
First of all in the gambit tournaments of 1912 and 1914 S was as successfull defending against them as applying them. Second S was not particularly successfull with the KG. Third S did not end 3rd at Carlsbad, but shared 2nd with Capa, which he defeated. Fourth S played many fine archaic attacking games in Carlsbad, beginning in the first round against Grünfeld - a very non hypermodern Queen's Gambit Accepted. Fifth hypermodern chess did not show up as suddenly as many people think: see Rubinstein-Spielmann, Bad Pistyan 1912.
S revising his playing style from archaic gambit play to hypermodern chess is just nonsense. The important change in Carlsbad was from 1.e4 to 1.d4 - but in general he employed his usual forcing chess. S was just very late recognizing the attacking possibilities of 1.d4 - and yes, that was one of his flaws.

So we agree, that 8.Be3 Nge7 and 8.h3 h5 9.Be3 Nge7 give White sufficient compensation or more, I suppose.
I will take a look on 8...Bf6 and 9...Bf6 the next couple of days.
Have you already studied The art of Sacrifice (Richtig Opfern)?

Ayayay, an overview by me. 15...d5? 16.Bg5 f6 17.Bf4 Qd7 18.hxg4 hxg4 19.Rxh8+ Bxh8 20.Qh2 o-o-o 21.Qh7 with a big advantage.
Maybe it is time for 3...d6?!  

Dear MNb,

Thank you for your lines. I had missed the transposition to the Becker, and despite Needtoknow's enthusiasm, that is probably not the place to look for an advantage, although we should note that the 4. b3 variation has been avoided.   
It should also be pointed out that in the 4. Ktc3 line that you had first put forward [4. ...g5 ; 5. h4 g4 ; 6. Ktg5 f6 
(6. ...h6 perhaps, with an improved Allgaier for White. The original form is very poor, so this may well still be good for Black) ; 7. Kth3 gh ; 8. Qh5+ Kd7 ; 9. Qf5+] Black may avoid the repetition with 9. ...Ke7 ; 10. Ktd5+ Kf7 ; 11. Qh5+ Kg7. But with the g-file about to be opened, White might actually have sufficient compensation.

As regards 4. d4 g5 ; 5. Ktc3 Bg7 ; 6. g3 g4 ; 7. Kth4 f3 , whether or not h3 and ...h5 are inserted, I think Black should follow the game Shulman-Kamberi that you quoted. It seems to me that he should then play 13. ...c6 with ...Qd7 or ...Qc7 and ...0-0-0 to follow. White will find it difficult to arrange a d4-d5 push or other breaks. Black is a pawn up, but what a pawn! Protected and ensconced on -f3, it's power can only grow. It doesn't seem unreasonable for me to claim an advantage here.

                                                                           Regards,
                                                                                Hubert Longtoe
                      

P.S. Did you quote two different Furhoff-Sandor games or did you reproduce the names incorrectly?

2.f4?! violates every opening principle there is.  Fischer's formula is just one of many that are objectively better for black.   The h6 lines are what I used to play.  White has almost no chance.  I say "almost" because the board gets very scrambled and if black does not stay composed, he could lose.  Then again, so couldn't white.  I gave up on 1...e5 because of the Ruy.  If more people would have played 2.f4?! I would still be playing it and picking up the easy points.  That statement about easy points will probably upset the King's Gambiteers, but my results were overwhelmingly in favor of the black side...no losses, 5 or 6 draws, and at least a dozen wins.  I even had a 2400 player try 3.Qf3 against me, and I let him escape with a draw.  He kindly pointed out my win after the game.