Sir

Progress in science depends on public debate and criticism of ideas. Unfortunately, debate is largely restricted to four domains: conferences, private conversations, journal clubs and peer-reviewed publications. The first three channels tend to be too private and ephemeral to help the community at large. Publication in journals runs on a timescale of months and, unless people bring new data, they are lucky to be allowed into the debate at all. This slow communication hampers progress. Most people do not know other scientists' views on published manuscripts. When a new paper appears, readers often spot logical flaws, experimental weaknesses, questionable assumptions or alternative interpretations. Yet individual criticisms may not be considered important enough to warrant publication. Even major criticisms are unlikely to appear until months or years later, and are often overlooked in the haystacks of the literature. Scientists even miss official retractions, continuing to cite withdrawn data accidentally.

We propose a simple solution. Each record in publication databases (such as the US National Library of Medicine's PubMed) should have a link for adding scientific commentary: essentially the electronic version of a Post-it note. The submitter of a comment would be required to provide his or her name, e-mail address, a short communication and potentially a password (for later editing or deletion). Insights, criticism, replications and non-replications could be posted.

The benefits of this system would include: immediate free and open debate of scientific ideas and results; easy dissemination of non-replications and negative results; a reduction in wild-goose chases by researchers unaware of others' insights into a published paper; and posting of links to subsequent papers that provide important verifications or contradictions.

The attachment of commentary to databases has been successful in other areas. Users can add 'Post-em' links to a public-domain genealogy database, the Social Security Death Index (http://ssdi.genealogy.rootsweb.com/cgi-bin/ssdi.cgi). Readers attach reviews to books in Amazon.com's database.

Since we desire our proposal to be practical and inexpensive, there would be no human gatekeeper. To avoid abuse, we suggest that unhelpful or libellous notes could be reported to a removal moderator, who would decide whether to eliminate them. To prevent milder forms of abuse, Amazon.com asks: “Was this review helpful to you?” Answers are tallied to give a sense of how the community values the comment. Automatic advertisement posting can be prevented by a requirement to key in a text code readable only by humans.

Given the importance of peer-reviewed publications to their careers (see P. Lawrence Nature 422, 259–261; 2003), will researchers freely provide their insights without any career benefit? They already do. During the peer-review process, most scientists spend a lot of time providing insights to a very few people. Posting a note would benefit the whole community and take very little time.

A handful of journals have begun experiments in 'open' peer review online (see T. Gura Nature 416, 258; 2002). This operates on a journal-by-journal basis, whereas our proposed post-publication commentary process should be centralized and independent of the journal of publication. For example, the commentaries on biological and medical papers should be stored in a single database under the free-access, disinterested aegis of the US National Library of Medicine.

We believe that this simple, practical idea could substantially speed the practice and progress of science. We address further issues and welcome feedback at http://www.cnl.salk.edu/~eagleman/postit.