National research evaluation systems that use metrics for the assessment of academicinstitutions are usually regarded as exemplifying the same neo-liberal model of governance that,with minor variations, is implemented worldwide. This paper argues, however, that, despiteapparent similarities, metrics are used for different aims in different national cases. We comparethe use of figures in the UK RAE/REF, a prototypical neo-liberal framework, with various schemasof assessment used to evaluate Russian universities in recent decades. We argue that in theRAE/REF, the principal role of numbers is to solve “the lazy agent” problem by creating aprisoner’s dilemma situation for academic institutions. In the Russian case, numbers serve tosolve “the corrupt knower” problem, preventing collusion between the assessor and the assessedby leaving unambiguous and inspectable traces of decision-making processes open for futurescrutiny. While not incompatible, these usages in no way assume each other. In the UK, fewprecautions were taken against assessors’ unreliability, and in Russia the element of competitionbetween universities was suppressed. We conclude by putting forward some hypotheses on thecultural and institutional origins of different approaches to quantification.