Christopher Exley, in Correspondence (“Funding should recognize outcome, not income” Nature 440, 1112; 200610.1038/4401112c), criticizes the use of research income as a metric of science somewhat unjustly. As someone involved in the review of research grants in the United Kingdom, I have found that research income — at least in the case of government research councils, and probably also for charities and industry — is highly dependent on previous and proposed research outcomes. Funding proposals require a brief biography listing previous outcomes as well as a list of expected outcomes and dissemination strategies, both of which reviewers are asked to assess.
Perhaps this is why previous runnings of the Research Assessment Exercise (RAE) have shown a strong correlation between research income and RAE score, at least in disciplines such as science, which require substantial grant income. Although this correlation is not perfect, the time cost to the UK research community of conducting the RAE is not justified by the difference. Further, interdisciplinary researchers who contribute to more than one literature will no longer be disadvantaged.
The main concern is that comparison must be done on a sufficiently narrow scope so that different branches of disciplines do not lose political clout if their research is relatively inexpensive. For example, if comparisons were made at the full departmental level, biology departments would be rewarded for dumping their field researchers in favour of molecular or cell biologists, and computer science departments for dumping theory of computation in favour of robotics.