i was watching a video of jean pons solving the 3x3 in about 12.x seconds and from the looks of it he uses the fridrich method!

here i am thinking that the fridrich method's limit was about 15-16 seconds.

does anyone else (that is consistantly sub-15) use the fridrich method?

yes, most if not all of them do. Some use COLL too, which will solve corners if the edges are oriented already upon reaching the LL. I have gotten a best average of 16.02, with a personal best of 11.23 and I use a fridrich F2L, with a CLL ELL last layer. BTW, if you want advice from strictly a sub 15 cuber, you're gonna have to look real hard. Sub 25 cubers have a lot of knowledge too, don't discount them.

I'd like to see other methods explored to see their limits. Particularly roux, his speed appears to be amazing..

~Thom

the roux method is insanely fast, Gilles himself is sub 15 with it. some other methods with potention in my mind are: the guimond method, the heise method, petrus (can be really fast, with good recognition), and last but not least, columns first, developed independently by myself and Masayki Akimoto.

Do you have any more details around your columns first? I'm currently converting to Corners First and Columns First seems to be an interesting alternative.

As far as I know, you're asking for the empty set:
http://www.speedcubing.com/records/recs_misc_av100.html

And yes, Jean does use the Fridrich method.