I trust Rybka's evaluations, but sometimes they are based upon the dynamic play that is possible in a position – often including moves played at the master level – as compensation for sacrificed material.
In games like the following, my opponent could be content with his extra material throughout the game. I struggled, successfully, as it turned out, to find all of the compensation that Rybka, afterward, assured me was there.
perrypawnpusher - irak
blitz 6 10, FICS, 2011
1.e4 e5 2.Nf3 Nc6 3.Bc4 h6
The Semi-Italian Opening.
4.0-0 Bc5 5.Bxf7+
The Semi-Italian Jerome Gambit.
5...Kxf7 6.Nxe5+ Nxe5 7.Qh5+ Kf8
Stronger, but more complicated with more risk, is 7...Ke6. The text move keeps Black's advantage.
8.Qxe5 Bd6
I have faced this move, without ...h6 and 0-0, against LeiCar and dogofthesouth.
9.Qc3
Possibly a bit stronger than 9.Qd4 in perrypawnpusher - LethHansen, blitz, FICS, 2009 (1-0, 26).
9...Qf6
Black would not mind exchanging some pieces, to remove any danger to his King. White should be able to take advantage of this mind set and get an equal game.
10.d4 Bf4 11.Nd2
After the game Rybka 3 preferred 11.e5.
11...c6
AlonzoJerome - MarleysGhost, blitz, ICC, 2011, also continued with an overt "exchange" theme: 11...Bxd2 12.Bxd2 Qc6 13.Qf3+ Qf6 14.Qe3 d6 (1-0, 17)
12.Nf3 Bxc1 13.Raxc1 Ne7
Black is rightly pleased with his solid game, although a skeptic would point out that he moved his dark-squared Bishop four times in exchanging it for White's un-moved counterpart. These missing tempi can be seen in comparing the two home ranks: White's Rooks are linked, while it will take Black three moves to do so.
The question, as always, is: can White make something of this, or will Black's extra piece prevail in the end?
14.e5 Qe6 15.Qd3 Qxa2
I knew that Black could not afford this further loss of time, even with his solid position. This is the kind of pawn-grabbing that Chess Challenger 7 used to do against me all the time, many, many years ago.
How to punish it??
16.b3 Qa6 17.Qe4
Understandable, although more consistent (i.e. "Jerome pawns") was 17.c4. Nothing is going to happen quickly, but White's clamp-down on Black's position can grow tighter and tighter.
17...Ke8
Far-sighted: he plans to make f5 a strong point for defense. It will cost him more tempi to get his Queen back into play, but if he does not create further weaknesses in his position it will remain White's responsibility to show compensation for his sacrifices.
18.Nh4 Qb5 19.f4 Qd5 20.Qe3 Rf8
As planned, but Black gets his moves out of order.
After 20...g6 White can still go for the breakthrough with 21.f5 but after 21...Nxf5 22.Nxf5 gxf5 23.Rxf5 c5 Black is still holding his own, according to Rybka 3. White would still have to play accurately and actively to show his compensation.
21.f5 d6
Understandable: Black hits White's center, and prepares to finally develop his Queen Bishop and Queen Rook. Unfortunately, it opens things up for White, who can now better get at his King.
As painful as it might have felt, Black needed to continue to set development aside and play something like 21...c5.
After the game Rybka 3 suggested 22.Rcd1 cxd4 23.Rxd4 Qc5 24.f6 Nd5 25.Qf2 gxf6 26.Nf5 Kd8 27.Nd6 b6 28.Ne4 with an edge for White. Again, that is a rather sophisticated evaluation: White's compensation is largely in dynamic play, and if he lags, he loses.
22.f6
Good, and thematic, but second best: how could I have overlooked the "Jerome pawn"-themed 22.c4 (followed by 23.exd6)?
22...gxf6 23.exf6
23...Qf7
After some excellent cold-blooded defense, Black slips again, and his game rolls downhill...
Rybka 3 reduces the game to a Rook and pawn endgame that is better for White with 23...Rf7 24.Rce1 Qg5 25.Qxg5 hxg5 26.Ng6 Bg4 27.Rxe7+ Rxe7 28.fxe7 Kd7 29.Rf7 Rc8 30.c4 a5 31.Kf2 Bh5 32.Nf8+ Ke8 33.Rg7 Bf7 34.Nh7 Kxe7 35.Nxg5 Kf6 36.Rxf7+ Kxg5 37.Rxb7 Kf4 38.Re7 Rb8.
24.fxe7 Qxe7 25.Qxh6
Again, this is enough, but the puckish 25.Rxf8+ was better, as 25...Kxf8 would have been followed by the Knight fork 26.Ng6+, winning the Black Queen.
25...Rxf1+ 26.Rxf1 Be6 27.Re1 Kd7 28.c4 Re8
Finally, the pieces are developed. If Black could move his Queen out of the pin along the e-file, he could envision further resistance, being only a pawn down, with a Bishop against a Knight.
29.d5 Qf8 30.dxe6+ Ke7 31.Ng6+ Kf6 32.Nxf8+ Ke7 33.Qg7+
Here Black's time ran out. His clock was still running over 2 1/2 minutes later, so I courtesy adjourned the game. He resigned the next day.
5...Kxf7 6.Nxe5+ Nxe5 7.Qh5+ Ke6 
Or 7...Kf6 8.Qf5+ Ke7 9.Qxe5+ Kf7 10.Qd5+ Kf6 11.e5+ Kg6 12.Qxc5 d6 13.exd6 Qxd6 14.Qxd6+ cxd6 15.Nc3 Bf5 16.d3 Rc8 17.Be3 a6 18.Bd4 Nf6 19.Rae1 Rhe8 20.f3 Kh7 21.Bxf6 gxf6 22.Rxe8 Rxe8 23.Kf2 Kg6 24.Re1 Rc8 25.Re7 Rb8 26.Ne4 Bxe4 27.fxe4 b5 28.Rd7 Rb6 29.c3 b4 30.c4 b3 31.a3 f5 Black resigned, 
Black does better with 10...Qf6 11.Rxd4+ Ke7; but, make no mistake, even with the text he has the advantage.
Or 12.Qh3 Qb6 13.cxd4 cxd4 14.Qa3+ Ke6 15.f4 N5c6 16.f5+ Kf7 17.Nd2 Re8 18.Nc4 Qc7 19.Nd6+ Kf8 20.Nxe8 Kxe8 21.Qf3 Ne5 22.Qg3 Kf8 23.Bf4 d6 24.Rac1 Qa5 25.f6 N7g6 26.Bxe5 Qxe5 27.Qxg6 Be6 28.Qxg7+ Ke8 29.Qe7 checkmate, joev6225 - liverpool, GameKnot, 2006.
White is now better, and Black's game goes to pieces.
5...Kxf7 6.Nxe5+ Nxe5 7.Qh5+ Kf8 
New to me. I have seen 4 other moves: 7...Kf6 (
