If you're interested in reliability and security of code, here's a nice little bit about some of the work going on at NICTA, Australia's national IT research center: http://www.technologyreview.com/article/37206/ Efforts to develop a crash-proof kernel will pay huge dividends in the long term, IMO. The article discusses security issues in high-risk applications, like automobile computers. Avionics is a logical extension. I'm interested in the consumer end: as computers move into consumer devices like fridges, stoves, air conditioners, water heaters, laundry machines, etc., there are two implications in my mind: First, that reliable kernels are needed in my fridge to keep my food frozen; and second, that we will all be surrounding ourselves with enough processing power to build useful, personal, IaaS clouds. Crash-proof kernels should go mainstream due to the need for the former, but they would be a gateway to actually realizing the latter: I don't think I need to get into the advantages of a personal IaaS cloud made of an array of ARM-based processors just waiting to be used for something more than boiling water. It's putting wasted cycles to use, and gleaning value from infrastructure you've already got. Imagine a situation where you misconfigure your email server. I know, it's NEVER happened, right? (lol) So you run a couple of tests, figure it's working, release it to the whispy expanses of your cloud, and… your toaster won't shut off. If you think that sounds far fetched, ask yourself how comfortable you'd feel having a "smart" toaster in a world where malware can do the same thing. Did you ever look at your toaster as high risk? I haven't felt I had to since I learned to stop poking at its contents with a knife. Stuxnet changed that as soon as a processor and a network were involved. Cool stuff. Sent from a locked proprietary device. I'm working on my freedom and I'll be there one day.