Funding Research Randomly

This paper compares three different procedures to select research applications: the current peer review procedure, pure random election, and a modified lottery involving some level of peer review as a first step and random selection as a second step. This paper considers four important desiderata: (1) the ability of the procedure to select the best applications, (2) the minimisation of the costs of selecting applications, (3) the impartiality of the selection procedure, and (4) the fairness of the selection procedure. The paper shows that there are trade-offs between these desiderata and that it cannot be concluded that a single procedure will necessarily satisfy all of them. Yet, the article shows that the Modified Lottery provides the best balance between the ability to keep costs low, impartiality, fairness, and, most importantly, the ability to select the best applications.