Those F2L algorithms you talk about are quite beneficial. I know of two benefits: reduced cube rotations and multiple slot insertions (multi-slotting). These two ideas focus on getting the cube solved as fast as possible by reducing the bulk of intuitive f2l.
Reduced cube rotation ,from what I know, is done by filling slots in the face opposite from you(B Face). These algorithms are categorized in two groups: "Pieces stuck" and "Empty Slots". The algorithms in the two groups don't require you turn you cube for three f2l slots. Feliks Zemdegs, in his f2l stage shown in this slow-mo video
http://www.youtube.com/watch?v=waAFfGK_r_s, turns his cube only twice!
Multislotting is filling in two slots at once in the time in time would've fill one. Even though this reduces your time, multislotting is hard to learn simply because it's too difficult recognize cases where you can insert two edges at the same time.
Certainly, they worth effort if you want to solve faster. You might not find it such a pain, when you realized you just performed your first multi slot or rotated your cube only twice, and the suddenly slam down a
sub-13 time.
On a somehow related note, some people cope with the agony of speedsolving practice simply by practicing to music. Here's Rowe Hessler,U.S. 3x3 average rank one,cubing while listening to music through headphones
http://www.youtube.com/watch?v=6wU9tJFx_Gc&feature=plcp.
All in all, learning the new algorithms will drop your times. If you want to be forever sub 17 ,then it's probably not worth learning. But, if you aspire to be up there with the elites like Hessler and Zemdegs, all I can say is "work hard and buckle up ."
