I had asked a related question on Quora some months ago. There is a need in many scientific fields, especially life sciences, for experts in scalable infrastructure, statistical learning, visualization, graph analysis, and good software engineering, but since hard science doesn't pay well enough, the majority of these people are going to the Facebook's of the world. Nothing wrong with that, but we have some challenges in figuring out what "value" means, how we measure it, and perhaps most importantly, how we reward it.

There has to be some way of getting some fraction of the really smart computational people into other fields to figure out very hard problems. Unfortunately, the future of the internet is not the only future out there.