Why can't we make research grant allocation systems more consistent? A personal opinion.

Uncertainty is expected to enter into our grant allocation processes at many points, not limited to those directly involving assessment by peers. The selection of grants for funding is thus prodigiously low in statistical power and will remain so. The replacement of current systems with some form of lottery, as has been proposed, seriously risks weakening the quality of applications. Opportunities exist for agencies to encourage and reward greater clarity and innovation in research outcomes. Recent papers have shown that peer review of grant applications is little better than random selection. Attention here is drawn to additional sources of uncertainty and the inherent low statistical power of our adjudication systems. Suggestions are made for improved focus on outcomes and the encouragement and reward of novel proposals.