<p>Scientific production is growing at an unprecedented rate, but how many of the published articles are really useful to society? Using a model of asymmetric information, I identify the conditions under which scientific journals might publish more articles than would be socially efficient. I also show that publishing all the science would certainly be inefficient. These results apply to both submission fee and free submission publication systems. To avoid this social inefficiency, policymakers could keep inefficient researchers out of the market by offering, for example, a subsidy or alternative employment in the public sector. In this sense, the division of academic labor between research and teaching would be helpful in streamlining the research market. The paper also examines the role of meritocracy, highlighting the negative consequences of decoupling researchers’ revenues from the scientific impact of their work.</p>