And of course, there's Austin's point about constant grantwriting --- even the PhD-level scientists who do get academic jobs end up chasing financing eternally (in the US, anyway). As far as I know, Europeans don't have to scrape up money for their salaries.
I recently read that there 's a growing group of people who've left full-time science jobs in favor of part-time jobs, who do science by squatting in friend's lab or at home.
PHDs need to "maniuplate" the system by finding a firm where they can solve the hard internal problems no one else can while using internal resources to work on their pet projects. They can then cite the firm when they publish. It is a modern twist on the old benefactor approach used prior to "big science"
If you look at Linux as a body of knowledge and not as a tool, then you can see the model for science going forward.
Take away the time writing grants and dealing with the primary and put it to work doing research!!!
