3.3 mil is a lot of money for uncertain outcome. We already have microreboots in some toy systems, that should solve crashes and ensure that system continues to operate (though it will probably go through the crash-reboot-work-crash cycle endlessly. We already have in-memory kernel patching from SUN and partially from linux. Not to mention SELinux and Hurd and the rest of security ideas. I think that the real purpose of the project is to suck funding. Let me quote from Tannenbaum's project proposal:

• "..but I should start out by pointing out how ambitious and risky this research is." In layman's terms it means: "I want to experiment, but no promises or even deliverables".
• "... nearly all experience with actual security incidents shows that security problems almost always stem from actions that the design and rules forbid but which bugs in the code allow to happen anyway." and "The most serious reliability and security problems are those relating to the operating system" I am a security consultant and most security incidents stem from misunderstanding the basics, like password management, not buffer overflows. Seriously, has he ever consulted a security practitioner?
• Some classic proof by (broken) analogy: "Banks lock their front doors at night and have strong safes even though there are laws forbidding bank robbery" What does this prove exactly?
• "What I am proposing is a fundamental redesign of the operating system." Dude, seriously, You mention known concepts and offer fundamental redesign? Is this just a rewrite project for minix?

I could go on and on like this. This is how funds are spent without any real gain, not even new concept evolution. Andy, give the EU taxpayers money back!