Here is an alternative sequence. The kernel is similar: the fianchettoed bishop with a knight in front of it on f6. Again the target is a White knight on d4; again that knight is protected once and attacked once; again Black has to decide whether to execute the discovery now or do a preliminary exchange first. One difference is that White’s king is not available as a target. Another is that in the previous problem White was defending the targeted knight with a very valuable piece (his queen) and Black was attacking it with a piece that was less valuable (his own knight); so it made sense to make the preliminary exchange first, turn the queen into a target, and then play the discovered attack against it. Here the situation is reversed. White guards his knight with another knight, and Black attacks the targeted knight already with his queen. If Black plays the preliminary exchange QxN+ first, he loses his queen and doesn’t improve the target. So here he skips the preliminary exchange and just plays the discovery Nxe4. If White replies f3xN, Black has BxN+ and nets a pawn. If White then takes the bishop with his e2 knight, Black recaptures with his queen.
Notice that with the sequence Black has spent his dark-squared bishop, which was doing important defensive work (look at all the vulnerable dark squares near Black’s king; Black no longer has a bishop to patrol them). From Black's standpoint the wisdom of giving up that bishop to gain a pawn is not clear-cut; some strong players might therefore conclude that the sequence just described wasn’t Black’s best move after all—though it was what Bobby Fischer played in Surgies-Fischer (1957).