Scientific administrators discuss ways to make research assessment more robust