8.g3 is a blunder losing the rook. I think 8.Kd2 Qxe4 9.Nc3 is much better - White has a piece for two pawns in the middlegame. Black's development isn't enough to threaten the king.
Sicilian-Prins Variation
Sort:
8. Kd2 Bg4 9. white resigns
9. Be2
Aaaaaaaaaagh! I totally missed that. Funny thing is, I typed Kd2, when I was thinking Ke2. And now that I think of it, Kd2 leaves white fine. I was thinking of Bg4+ and winning the queen. 
Well... Who knows what I was thinking???
Yikes. Anyways, I guess that the sac wouldn't quite make it.
Well the f3 variation has a name and everything.
And yet, Short, Svidler, Rublevsky, Kramnik and Ivanchuk have all played this. Mostly in blitz or against weaker players, but also for instance Ivanchuk-Dominguez in last year's Corus A.
5...e5 6.Nb3 d5 was played only once in the 10 games where White was 2700+; but White won.
It doesn't work because the rook isn't actually hanging in that line. After Qxe4, Kf2, if Qxh1, then Bb5+ wins the queen.
[COMMENT DELETED]
Estragon wrote:
White isn't going to play 5 f3 anyway, is he? 5 Nc3 is normal, and good.
If you only want/need a draw with white, this is a solid line. For instance, Carlsen chose it vs Karjakin in the last game of the 2016 WC match, as he had won the previous one and only needed a draw
6...Nxe4? loses. Black should continue 6...d5.
I'm trying to figure out why black doesn't play the following:
Now, I have to assume that white would respond differently to Nxe4, otherwise, I see no way out. Can I get some help here seeing whatever I'm missing?