A review process is considered “double blind” if neither authors nor reviewers are aware of one another’s identities. Though it is relatively rare in the life sciences, it is more prevalent in social science and the humanities. Double blind review is one approach to mitigating biases in review, which can range from a bias against female corresponding authors (noted in studies of Frontiers, eLife, and AGU journals) to a bias towards authors within reviewer’s own co-author networks. For those journals and disciplines that use double blind review, it is an important part of their peer review process.
However, double blind peer review cannot eliminate all biases since many decisions fall with editors, and because many reviewers can already successfully guess the identity of authors. As summarized by Hilda Bastian: “The rate of failure of blinding [in 8 trials] was high: average failure rates ranged from 46% to 73% (although in 1 journal within one of the trials it was only 10%).” The rates of blinding failure may be influenced by how stringently authors obfuscate their identities in the manuscript, for example in referencing previous work.
Preprints might make it more difficult to preserve blinding. Currently, no preprint servers enable anonymous posting; doing so could compromise many of the benefits of preprints, such as visibility and recognition, and raise challenges related to ethical disclosures. COPE discourages pseudonymous preprint posting for similar reasons. That said, there are also many other sources of information that could compromise blinding. For example, seminar announcements, conference abstracts or programs, or social media posts are all searchable on the web. It is expected, however, that referees participating in double blind review will not attempt to discover the identities of manuscript authors, and will recuse themselves from reviewing papers with authors known to them.
This means that any early visibility of work can either harm the integrity of the double blind review process (if reviewers ignore requests to avoid attempting to identify authors) or make it more difficult to find suitable reviewers. While exacerbated by preprints, this problem is by no means unique to them.