Revitalising rapid responses

I will go root away
The noisome weeds, which without profit suck
The soil's fertility from wholesome flowers.

Shakespeare, Richard II

When we were somewhat greener, we likened websites to gardens: both combine amazing opportunities for experiment with the option of obliterating mistakes when things go wrong.¹ But even then we warned, “Turn your back on them for a few weeks, and they're overrun with weeds.” This is the fate that has befallen rapid responses, and why we're raising our threshold for publication.

On their launch, these electronic letters to the editor were hailed as the salvation of the journal's need to provide opportunities for timely debate. Previously, we had published only about one third of the letters to the editor that we received, and those about six months late.²

Marking the publication of the 20 000th response in 2002, we judged the experiment a success, even suggesting that rapid responses might point towards new models of knowledge creation.³ By then we were publishing “just about anything that isn't libellous or doesn't breach confidentiality.” We acknowledged the downsides of such a liberal policy—bores who monopolised conversations for compelling personal reasons—but judged that the upsides more than outweighed them. In support of our policy, the previous editor quoted John Milton on freedom and truth to the sceptical readers of Nature.⁴

Having now posted our 50 000th rapid response, both the upsides and the downsides have become more obvious. The upsides are that rapid responses allow important criticisms to be made immediately after publication and that this form of peer review can continue indefinitely. Groups beyond the print journal's usual readership (such as patients and readers outside the United Kingdom) can contribute, and discussions can range beyond the original findings to suggest new avenues of research. Writing in JAMA in 2002, statistician Douglas Altman found it “remarkable and disappointing” that so few journals provided rapid publication of correspondence on their websites.⁵ Since then, other medical journals have followed our lead (most recently the Lancet, lifting most of bmj.com's submission guidelines verbatim).

The main downside of rapid responses is that the bores are threatening to take over. Some respondents feel the urge to opine on any given topic, and pile in early and often, despite having little of interest to say. Others have pet topics, which they return to obsessively, finding almost any peg to hang their views on. Some respondents don't seem to feel they're really alive until they've sparked off an angry response from someone else. Rows then continue for longer than interests anyone other than the combatants. Attacks on views can move swiftly to attacks on the holder of those views; these were often continued via abusive emails until we stopped posting email addresses with responses.⁶ Our impression was that the overall quality of responses was falling. Respondents whose views may have been worth reading told us they weren't contributing because of the conditions of engagement. The noisome weeds were sapping the wholesome flowers.

Our solution is to enforce more rigorously our original criterion for publication—that a response contributes substantially to the topic under discussion. It hardly marks a radical departure: we're merely reaffirming our editorial responsibility to readers. Raising the threshold for acceptance means that once a point has been made in one place we won't post an almost identical response in another. We will no longer offer the AIDS deniers and their opponents space to continue their shouting match of the deaf, with online content equivalent to some 20 print BMJs devoted to their row.⁷ (No material will be removed, however, and interested parties will still be able to access the archived content.) Responses directed primarily against the messenger rather than the message won't be posted; nor will responses that make reasonable points but are also gratuitously offensive.

Although space on the web is unlimited, readers' attention isn't. We therefore recommend that responses be no more than 500 words; contributions longer than 1000 words won't be published. (Our preferred limit for letters in the paper journal, which are now all selected from the rapid responses, is 250-300 words.) And, given the increasing number and complexity of responses we receive, we will not enter into further correspondence over rejections, other than directing inquirers once to our guidelines.⁸

Debate lies at the heart of this journal's mission, and nothing we intend doing should endanger that. But even the most indulgent gardener has to worry about weeds if a garden is to give benefit and delight to others.