Ari Patrinos and Dan Drell (Nature 417, 589–590; 2002) recommend that journals embrace the commercial realities facing science, suggesting that arrangements such as Science allowing some data restriction are necessary for progress. Although they accept that open access to research data has been crucial to scientific success, their core argument is that too much data will remain locked in the private sector unless journals become more accommodating. Patrinos and Drell's suggestion is that publicly funded information should be unrestrictedly available, whereas privately funded findings can be legitimately restricted for an agreed time.

Such reasoning may be the start of a slippery slope leading to different standards and treatment for privately funded, profit-making science. The growing trend towards public–private partnerships will grease the slope and stud it with inconsistencies. Will the public's share of data be embargoed with the private? Will the length of the embargo be determined by the ratio of public and private money? Will the time of data release be negotiated by authors with publishers on an individual basis? After all, offsetting the credit reward from early disclosure against the loss of privileged access to one's own data is as important to the career of a publicly funded scientist as to the profit of a corporation.

Even more general than these considerations is the role of data disclosure in scientific success. The tension between openness and protection of novel findings is a key feature in the dynamics of collective scientific practice. The way scientists have historically balanced competition and cooperation has led to generations of scientific achievement. More data may indeed be gained for all from special publication arrangements, but will this be sufficient compensation for the loss of a mechanism that has been crucial in the extraordinary accomplishments of many disciplines?

Both supporters of early, full data release and advocates of partial commercial protection argue that their strategy leads to more scientific progress. We agree with Patrinos and Drell that norms of 'free access' and 'total disclosure' are unrealistic: indeed, pre-publication secrecy has always been part of successful science. Nevertheless, accepting a trend of increasing market orientation as a policy guide is unscientific. How science works is itself a scientific question, not an ethical or pragmatic one. Surely, scientific inquiry should rule on the advisability of overthrowing conventions associated so strongly with scientific success. It is remarkable that the effects of radical changes in institutional settings and reward structures for scientists have not been more extensively researched.