There have been a lot of discussions of refereeing over at “Brian Leiter’s blog”:http://leiterreports.typepad.com over the last few weeks. I think many of these discussions suffer from a misapprehension of how refereeing works. In particular, many people are equating the time it takes for them to get a verdict on their paper with the amount of time it takes for a referee to write a report. This equation would work iff the following steps took zero time.

1. Processing a submission (i.e. entering it into the system, getting it ready for editors etc).
2. Deciding whether the paper was worth refereeing.
3. Choosing who should referee it.
4. Contacting that referee, and getting them to either agree to referee it, or decline to referee it, or alternatively waiting/re-emailing them enough to assume they’ve tacitly declined.
5. Repeating steps 3-4 ad nauseum until one has sufficient referees.
6. If a referee agrees to referee a paper, repeating steps 3-5 until one has a referee who will, eventually, referee the paper.
7. Reading the referee report and making a decision on the paper.
8. Contacting the author to inform them of that decision.

In most cases where I’ve been familiar with long long times from submission to decision, many of those steps have taken a very very long time. I’m hopeful that in the future the use of journal management software can improve steps 1, 4 and 8. But there will still be a lot of ways for things to fail other than for the person who referees the paper taking too long. Indeed, typically the person who writes the review the author gets is part of the solution to the long delays, not the central problem.

So while everyone else is talking about speeding up review times, I think the following three steps would make just as big a difference, if not more of a difference.

1. When someone asks you to referee a paper, reply more or less immediately.
2. If you decline, suggest alternative referees who can referee it, and who the editors will likely not have thought of. Junior faculty, or even trustworthy grad students, are excellent suggestions. Suggesting that Tim Williamson referee the paper on luminosity that you’re too busy to referee isn’t so helpful. If the paper is on a relatively specialised topic, this step is more or less essential, else the editors literally run out of people they know and trust who are experts on that topic.
3. Never, under any circumstances, fail to review a paper that you’ve agreed to review.

Journals which send copies of the paper along with requests to referee the paper make it much easier for potential reviewers to make an informed decision about whether to review, and hence help with point 3. I think this practice (which isn’t I think universal) should be much encouraged.

There’s another issue lurking around here that I think deserves discussion. Some journals have a blind initial review/selection of referees. Many, I believe, do not. Making this stage blind is *very* time-consuming. It requires that there be a staff member who handles all interactions between the author and the editor, and who can tell the editor whether the suggested referee is the author and/or too close to the author. Since most staff members do not work 24/7/365, and some have even be known to do things like get sick or go on annual vacation, this can introduce large delays into the system.

(It’s an important point here that many journals have precisely one staffer. A move to a system like Philosophy Compass has, where the administrative work is done at a publisher’s office, and there are people to spread the work around when one is on leave, helps remove these delays a lot. I think something important would be lost if the administration of all journals moved ‘off-campus’ like this, but it would smooth out some of the administrative bumps that I’ve noticed on-campus journals are suspectible to.)

If you don’t have blind initial screening by editors, and blind assignment of editors, these steps can be cut out, and staff time can be spent on other activities. (Such as dealing with complaints from subscribers, and fun stuff like that.) So I’m interested in knowing how important people think it is that papers be blinded from editors, and whether this is worth introducing delays (sometimes weeks long delays) into the system?