About a year ago, the life science journal eLife changed its approach to scientific publishing (see details here). The goal was to focus the activities of the journal on peer review. Perhaps the most consequential change was to abandon the accept/reject decision. Instead, eLife reviews get attached to the author’s preprint, along with a brief editorial assessment that rates the manuscript along the dimensions of significance and strength of evidence. 

In the eyes of some scientists at least, this change diminished the reputation of the journal. There was a sense that by leaving the decision to publish up to the authors, the journal would open itself up to all sorts of low-quality content. A few days ago, eLife published a self-study about the first year of experience with this model, and it is well worth a look. Here I want to add a few personal comments, based on my experience as an author and as a reviewing editor: 

• There’s a widespread belief that eLife’s new model is to “publish everything”. That’s just not the case. In fact eLife’s new model is just as selective as the old model: about 1/3 of submissions get sent for review. 
• The decision whether to review a manuscript is typically made by a group of 4 researchers: a senior editor and 3 members of the board of reviewing editors. In my practice, I read the manuscript with enough care so I could start writing a review. If I take the manuscript on as reviewing editor, I will be one of the reviewers. 
• One of the questions I ask myself during triage is: Would I recommend this manuscript to my students for a journal club? Because all eLife submissions are available as public preprints, this is not a theoretical question. A journal club can then be a good start to a review.
• By the time a manuscript gets reviewed, it has typically been seen by 6 researchers: the above 4 and 2 additional reviewers. The reviews and a short editorial assessment get attached to the published preprint. The authors can choose to (a) do nothing further with eLife; (b) publish the current version under eLife’s banner; (c) revise the manuscript and ask for a second round of reviews.

Here are some things I like about this process, compared to many conventional journals:

• The decision to review is made entirely by fellow academics, 3 of whom are chosen from a panel of hundreds for their expertise in the author’s field.
• The reviews are published and, optionally, signed. This leads to a healthy degree of civility, enforced where needed by the reviewing editor.
• As a rule, eLife reviews don’t tell the authors what additional experiments they should do (and what the outcome should be) in order to get published. The authors design their research report, and reviewers assess it for what it is. 
• The reviews are published alongside the preprint regardless of the assessment of the manuscript. Most conventional journals don’t publish reviews at all, or only if the manuscript is “accepted”. In all other cases the reviewers’ work ends up in the trash, never to be seen again.
• The assessment of a manuscript is nuanced, not binary (accept/reject). It is expressed in the editorial assessment, with details in the reviews. Users can scan eLife’s web site for title/abstract/assessment and decide what is worth reading.

And a thing I would like to change:

• The intent of the “new model” is to focus the journal’s product on peer review. I think we can take this one step further: There is no need to typeset the “version of record” using eLife’s fonts and page layout. In my experience that publishing step adds many weeks of delay and several rounds of proofing to correct errors introduced by the typesetters. What is the point? The paper is already published in perpetuity on the preprint server; that version is the one we reviewed; the reviews are attached in perpetuity to that version. Why duplicate the effort of typesetting and formatting? It seems like a pointless waste of time and money. Instead, we should simply list the title/abstract/assessment on the eLife web page along with a link to the preprint. That can further reduce the fee we charge to authors.

Finally a plea to you authors:

• When you post your preprint, please make it a live document! It is shocking how many manuscripts on biorXiv are still built like the dead stacks of paper we mailed to the publisher in the 1980s: Pure text in the front, followed by pages of figures, followed by pages of captions, then references, and no hyperlinks between them. Really people, get with the times! It is trivial now to produce a live PDF document where figures are placed where they belong in the text; captions are below the figure; hyperlinks allow you to instantly pop to a figure, or reference, or equation; a table of contents organizes the manuscript by section. If you have no clue how to do this, please try Overleaf. Format your preprint as though it was the only version of your report that people will ever read. Increasingly that’s the truth anyway.

In summary, I continue to think that eLife offers a great service to discerning authors and readers, and that its focus on open-access peer review points the way to a better future for scientific publishing.