Jonathan Adams' analysis cannot distinguish whether scientists collaborate internationally because they are successful, or whether they are successful because they collaborate internationally (Nature 497, 557–560; 2013).

If it is the former, then any government initiative to promote collaboration for its own sake risks simply degrading the correlation between collaboration and success, rather than improving the quality of scientific output. Impact factors and publication rates are arguably imperfect estimators of past success, and insecure guides to future success.