Here is an email from "Khayyam" (lightly edited):
Hello Dennis!
I would like to ask for your opinion about the following line which caught my attention about a year ago and during this year I've tested it in several practical games:
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Bc4!?
Although I'm not quite sure whether it is a "novelty" at all, I haven't found 5.Bc4 on the internet (at least in high-level games).
I've analysed the two main variations 5...Nxe4 and 5...Qa5+ and have come to the conclusion that the pawn sacrifice is quite reasonable and deserves serious consideration (So the question is that: why hasn't it been tried by grandmasters yet?) I'm aware that Black can always deviate from White's preparations by 5...e6, but then, the Scheveningen variation should be declared Black's main choice in Sicilian, shouldn't it?
I can't claim that my analysis is faultless and that's why I really need your opinion, you can imagine my curiosity about the whole matter.
Your suggestion is an interesting one, and while I've never seen it in high-level play either (according to my database, it has appeared only once in games between 2400+ players), I do recall reading a very brief article on it a quarter of a century ago! IM Mark Ginsburg had a series on opening sidelines ("Chess Openings for Heroes") in a now-defunct publication called Players Chess News. Here's his analysis, which I was able to dig up this afternoon at the Notre Dame library, from Chess Annual I (1981/2), pages 62-3:
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Bc4 Nxe4 6.Bxf7+ Kxf7 7.Qh5+ g6 8.Qd5+ e6 9.Qxe4 with a "positional sort of pull", albeit one that is "not heroic", or - his preferred idea - 6.Qh5(!) e6 7.Bb5+ Bd7 8.Nxe6 ("Sheer poetry!") Qe7 (8...Qa5? 9.Bd2! 1-0) 9.Nc7+! Kd8 10.Nd5! Ng3+ 11.Nxe7 Nxh5 12.Bxd7 "with a better endgame(!) for White."
Now I'll throw in my two cents. First, as noted by both "Khayyam" and Ginsburg, Black can play 5...e6 and transpose to the Sozin (assuming White plays 6.Nc3). "Khayyam" seems to think this is a concession on Black's part, as the Scheveningen (1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 e6) is not the most popular line of the Sicilian anymore. The last part of that sentence is correct, but not part one: Black is under pressure in the Scheveningen against the English Attack (6.Be3 followed by f3, Qd2, g4, O-O-O etc.), and to some extent the Keres Variation (6.g4) and the 6.Be2 lines, but whether one plays the Najdorf (1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6) or the Classical Sicilian (same first four and a half moves, then 5...Nc6), playing ...e6 is almost automatic against Bc4 lines. So meeting 5.Bc4 with ...e6 isn't really a concession at all. In fact, while some Sozin specialists like GM Mikhail Golubev might disagree, the 6.Bc4 variations are generally considered quite pleasant for Black these days, theoretically speaking.
Next, we should make an attempt to evaluate Ginsburg's suggestions. First, does White have a positional pull in the 5...Nxe4 6.Bxf7+ Kxf7 7.Qh5+ g6 8.Qd5+ e6 9.Qxe4 line? I have my doubts. The position is strategically complex with imbalances galore: Black has the bishop pair and a pair of unopposed central pawns; White has a lead in development, fewer pawn islands and the possibility of exerting pressure on Black's center pawns as they advance. I think Black has approximate equality after either 9...d5 or 9...e5, but if I were going to head for such a position on a regular basis, serious analysis would be required.
After Ginsburg's preferred 6.Qh5, however, accurate play leaves Black without any problems. One example of interesting but probably inaccurate play is 6...d5. Neither 7.Bb5+ nor 7.Qxd5 offer much, but 7.Bxd5 g6 8.Qe5 Nf6 9.Nb5 is good enough for an edge. So we continue with his line - 6...e6 7.Bb5+ - and now there's not only his 7...Bd7 to consider, but also 7...Nd7 (best) and the greedy 7...Ke7 to boot. Both are playable, but 7...Nd7 is simplest. The point is that after 8.Nxe6 Nef6 9.Nxd8 Nxh5, the White knight is trapped. It escapes after 10.Bxd7+ Bxd7 11.Nxb7, but after 11...Bc6 12.Na5 Bxg2 13.Rg1 Be4 it's Black who enjoys the slightly more comfortable ending.
Going back a ways, it's also worth mentioning the other 5th move possibility noted by "Khayyam", 5...Qa5+. After the natural 6.Nc3 Nxe4 7.Qf3 Nf6 8.O-O e6 (not 8...g6 9.Re1 Bg7 10.Bh6!) White has a lead in development, but I'm not sure it will yield tangible returns. For example: 9.Bf4 a6 10.Rad1 Be7 11.Rfe1 O-O and White's compensation is starting to fade.
In sum, it seems that Black should be fine with either 5...e6 or by "falling for" the trappy lines, but as White isn't in bad shape either it's a reasonable sideline for the occasional game, provided White doesn't mind the main line Sozin.
You can replay the analysis above, and a bit more besides, here.
