All of them have gotten valuable work done, therefore, in some sense the system works. Whether it is the best of all possible systems for basic research is open for debate. As to whether universities take a cut, that seems like a minor point that can be easily corrected with some adjustment of the rules. Either you're accussing the universities of corruption, or what they do is allowed by the grant-givers, and if it is allowed, it is because the grant givers have accepted the reasonable argument that the overhead of running the university should also be covered by the grant. If the cut is excessive or if the taxpayers feel strongly about this the rule could always be changed. If you feel strongly about this you could always start a petition; no one ever accussed the American taxpayer of being too genorous, and it seems easy to rile people up to fight against a perceived excess of spending.
Hothouse design shops need radically different structures. Pick up any business magazine that's doing a portrait of some hot design place. I think Wired had a special issue on design not that long ago, and FastCompany has them on a regular basis. The emphasis at a lot of these design shops is on brainstorming and creativity. Whereas, in academic research, a major effort is put into training people during their 20s not to make assertions as fact that they can not prove or at least cite a reference to justify. The whole environment in academia is anti-hypothesis and anti-speculation and pro-testing and pro-publishing the results of those tests. There is social pressure not to talk about hypothesises except with your advisor - don't talk about them in public, don't appear to be a media hound. Product development, I think, needs a sexier workplace environment than this, and something more tolerant of rentless idea generation.
>From what I've seen my friends go through, academic life is almost
mindlessly cruel and hard. All of them now get paid more and work less than
when they were in school. My friend who gave up on her post doc went from
$28,000 a year to $70,000 a year and from 50 hours a week to a more normal
40. She now does statistical analysis for a marketing firm.