Daniel Sarewitz argues that the pressure to publish is fuelling irreproducibility, but we disagree that the solution is to publish fewer papers (Nature 533, 147; 2016). In today's competitive arena, asking this of scientists — particularly junior ones — is to ask them to fall on their swords.

Investing more effort in fewer but 'more complete' publications could hold back early-career researchers, who already face fierce competition. To generate a first-author publication, graduate students on average take more than a year longer than they did in the 1980s (R. D. Vale Proc. Natl Acad. Sci. USA 112, 13439–13446; 2015). Introducing further delays for junior scientists is not an option as long as performance is rated by publication metrics.

In our view, publishing less is not a feasible or responsible way to improve data quality. This would be better achieved by increasing the transparency of peer review and by introducing alternative metrics as indicators of reproducibility. Science's goal is to share as much information as possible — not to withhold it.