The win takes at least 39 moves according to the Nalimov tablebases. A reasonable explanation would be difficult to post here.
K+2N vs. K+P

How is this a win? 2 knights can't force a mate on a lone king, and the pawn has no hope of being promoted.

How is this a win? 2 knights can't force a mate on a lone king, and the pawn has no hope of being promoted.
2 knights can't force a mate against a lone king, BUT this can be done against a king and a pawn. Just a fun thing worth remembering sometimes.

Other fun projects we can help the OP with:
squaring the circle
invading Russia
figuring out Fermat's last theorem
finishing Schubert's 8th symphony for him
concocting a unified field theory
finishing Schubert's 8th symphony for him
...
Actually that has been done. Several people have made a completion, so there are many versions: http://en.wikipedia.org/wiki/Symphony_No._8_%28Schubert%29

Yes, I know that it is possible (Not April Fools Day!). However, I would like to know the general strategy behind the mating attack, not the specific moves involved.
Well, you need to chase the king in a way so it is forced into a corner and then trap it there with the king and a knight. Then, because black has a pawn, he can not be stalemated, so white has time to move his other knight closer, while black makes pawn moves, until white can make a check mate with the other knight.

finishing Schubert's 8th symphony for him
...
Actually that has been done.
So has invading Russia. Don't get me sidetracked.

@chessisgood - why not play the mating line through and see if you can figure out how it's done? you'd be surprised how much you'd learn from it.
BTW, i didn't realise white could force a mate with NNvP, thinking they could only win by helpmate. Playing this through i've realised the subtleties of the position

You might want to check out this article - The Troitzky line. Very instructive, actually.

Very cool. Thanks Natalia. I'll admit, I didn't know that it was impossible to checkmate a king with just 2 knights. I always thought I just wasn't good enough yet.
Karsten Mueller wrote a couple of articles on the 2N v. P endgame. It is a very rare endgame though.
http://www.chesscafe.com/text/mueller35.pdf
According to most sources, this endgame is a win for the stronger side. Of course, if the pawn is gone, a draw is forced. Can someone please explain how to win and show some variations as well?