This is the title of the Back Page editorial of the March 2008 issue of APS News. Authors are Nobel laureates Thomas R. Cech and Steven Chu, together with physicist Neal Lane; they are members of the American Academy of Arts & Sciences committee charged with revisiting the mechanism and procedures underlying federal funding of research.
In general, I agree with the tone of the editorial. Here are some excerpts, and my comments.
Just as securing tomorrow’s talent is imperative for American scientific competitiveness, so too is supporting high-risk, high-reward research. When funding becomes tight, there is a natural tendency for reviewers and program officers to give highest priority to those projects that are most likely to produce “useful” results. Much research that might be described as incremental is important and worthy of funding. But our nation’s research portfolio needs to be balanced with some projects that set out to transform our understanding of the world or develop radically new technology, while accepting the risk that they might fail completely.
Proposals should be idea-based, focusing on goals and strategies, and should articulate the potential impact of the work, not be encumbered with excessive methodology. In many cases, more emphasis should be placed on the track record of the investigator or, in the case of early career scientists, on the creativity they showed during their training period.
Here, I am a bit more skeptical. While I agree in principle, I am afraid that placing emphasis on “creativity” of early career scientists and “track record” of senior ones may amount to nothing more than re-assessing established hierarchies of institutions and investigators. In times of meager funding, this may well mean the death of many thriving research programs at second-tier schools, and I do not see that as good for science as a whole.
In addition, transformative research can be stimulated by seed money dedicated to projects that are truly high risk, high reward. Such grants could be non-renewable, but should be of sufficient size and duration to permit proof-of-concept.
Here too, I agree in principle, I am not so sure about the practical implementation.