Dear members of the SciPost editorial college:
I have been asked (by Anton Akhmerov) about my workflow as SciPost editor. What I will describe is my personal way to approach things, based on experience with about 100 manuscript assignments per year , but I do not present this as a “best practice”. It’s just an approach that works for me, in the sense that it allows me to handle a large batch of manuscripts efficiently.
I try to avoid sending out manuscripts for review that are unlikely to meet the SciPost requirement  of a “significant advance”. A speedy initial assessment is in everyone’s best interest, so if I am certain that a contribution makes only an incremental advance I choose to reject without review (= desk rejection). If am less certain I may instead send the manuscript back to the authors with a “revision” decision, informing them that they will need to make strong case for significance, and suggesting that choosing a less selective journal might be more productive.
In general I think we should have more desk rejections at SciPost, to keep the queue manageable.
Here my strategy is to cast a wide net. I typically invite 8-12 referees. Once 3 have accepted I cancel the other invitations. In this way I try to avoid having to repeatedly return to a manuscript because I don’t have enough referee reports to reach a decision. I ask the authors to suggest 3-6 names, and then add another 6 myself.
For the referees that I add myself I mainly use the reference list of the manuscript, trying to identify papers on the same topic. Sometimes the arXiv preprint has already been cited, that’s another source of referees. Google Scholar has a “related articles” functionality, I have not made much use of that.
The “wide net” strategy also means that I do not worry too much whether an invited referee is known to be an expert on the topic. In fact, I tend to have more success with junior authors on a paper, rather than senior authors. In my experience an invitation will be declined if the individual does not feel confident to write a substantial report, and in any case, if the report just says “nice paper, should be published” it will carry no weight in the editorial decision.
I switch the autoreminder off, I prefer to remind the referees myself; a referee who has promised a report will in most cases deliver; if one or two reminders have no effect, as a last resort I will write to the referee from my personal email, but this is only rarely needed.
With a minimum of two reports I will in most cases have sufficient information to proceed; if the reports do not make a convincing case for a very significant advance, I will suggest to the authors that their manuscript is transferred from SciPost Physics to SciPost Physics Core; in my experience most authors agree to this transfer; I do try to avoid an outright rejection at this advanced stage of the process.
This is about it; feedback is welcome.
- I accept about 2 assignments per week. The triage and referee selection steps take up to 1 hour per manuscript. I have a dozen manuscripts in the refereeing stage at any given time, these require work when a report comes in, when a referee needs chasing, and finally for the decision when all reports are in. This may add up to another 2 hours per week, so that I estimate I spend around half a day per week on handling SciPost manuscripts. This is not quite sustainable for a volunteer effort, but that is another topic for discussion. The main point I wish to make here, is that if you accept one single assignment each month, it should not cost you more than 2 hours per month.
- I do not find it useful to distinguish in the triage stage between SciPost Physics and SciPost Physics Core. Both require a “significant advance”, hence an incremental advance is not acceptable for either. I prefer to delay the decision between the two journals once I have the referee reports.