The three factors it would seek to improve
- Increased Usability - Increasing Security - Reduced maintenance time/costs
Some research requires starting from scratch see Coyotos (http://www.coyotos.org/) for example. This is pretty difficult. (http://www.uni-koblenz.de/~vladimir/breviary/os-research.html)
Microsoft does do research into OSes but their goal is to increase profits so there are certain types of research they will not pursue. Reducing the need for the upgrade cycle, for example.
The operation of the charity would be mainly funding postgrad research places and PhDs studentships. If it generated a promising cocktail of technologies that would be worth the cost of switching from established technologies, then it would fund a group to produce an open source OS.
Jonathan Shapiro's decision to join MSFT (http://www.osnews.com/story/21262/Jonathan_Shapiro_of_Coyotos_BitC_Joins_Microsoft) made me think that something like this is necessary.
So what say you hacker news? Is there space for something like this?