Raymond Fisman, a professor of economics at Columbia business school, wrote in Slate  about research (PDF) of his colleagues from MIT.  The argument is that when you take similar young biomedical scientists and give them short-term research support (like some well-known scholarships) vs. long term support with very few strings attached (Howard Hughes Medical Institute awards), people with HHMI tend to be more creative and their work is more profound and has wider impact.  The (obvious) conclusion is that encouraging risk-taking encourages breakthrough works.

But! Because of the issues in the underlying analysis (as reported in Slate; I didn't read the original paper), the most interesting issue was missing from discussion.

The most obvious methodological issue is the level of accomplishment in the two groups being compared. People competing for awards like Searle or Pew are early career scientists (typically, no more than three to five years from getting their first independent position as assistant professors).  In contrast, a fair number of people becoming HHMI Investigators are much more established, ten years or more into their independent careers.

You could try to control for that by leaving only the young HHMI scientists.  But they belong to a more accomplished pool.  It's like comparing 15-year-olds who are doing very well in high school with 15-year-olds who happen to be sophomores at Harvard. {Having said that, many of the same people had Searle or Pew, in addition to HHMI award.  Those Harvard students have been good h.s. students, too}.

But the most striking difference is in the INCENTIVE STRUCTURE.  Basically, to become an HHMI Investigator, you have to forswear becoming rich.  As a condition of HHMI funding, you agree not to own more than 5% of any commercial entity.  So if you are thinking of starting a company to profit from your discovery of the magic anti-cancer drug: the answer is no.  You do get a lot of freedom to pursue your interests; but these are non-commercial interests.  So another conclusion seems to be that the fundamental beakthroughs are more lokely to be accomplished by people who don't have personal riches as their primary goal.

Hmmm...