Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
I prove a convenient reformulation of Kreps and Wilson (1982, Lemma A1), whose proof has a nontrivial gap. Essentially, the support of a consistent assessment is characterized by the additive representability of the infinite-relative-likelihood relation that the support implies. My proof is unexpectedly elementary, for it relies solely on a classic result about additive representation, which in turn relies solely on Farkas’ Lemma.