Depends on what interests you. I think massive distributed systems processing huge amounts of data using exotic network technologies and super tight code are i

Depends on what interests you. I think massive distributed systems processing huge amounts of data using exotic network technologies and super tight code are interesting, but it's not everyone's bag. C++ is still used a lot because it's fast and allows you to efficiently communicate with the hardware and OS. I'll admit it's ugly, but it's a good choice for certain systems. On the pure analytics side, most quants seem to be moving to more modern and flexible languages, where prototyping ideas quickly is more important than latency. Jane St Capital uses OCaml extensively (http://video.google.com/videoplay?docid=-2336889538700185341# http://video.google.com/videoplay?docid=-2336889538700185341...). I'm curious to see if Google's Go makes any inroads. It seems well-suited to building these types of systems and does away with some of C++'s baggage.