When you are running a journal, no matter how where it sits in the prestige rankings, all sorts of things wash up in the submission box. One might think of these as roughly falling into four categories:
- Clearly plausible and on-topic papers
- Papers whose focus or quality is dubious, but are still worth sending to review
- Papers that are clearly off-topic
- Papers that are clearly below the journal's quality threshold
Until you actually see such a stream, you might not realize just how bizarre the material can be that shows up in the third and fourth categories. I'm talking about likely-schizophrenic ramblings where the text changes color and font multiple times per paragraph, blatant plagiarism, or papers that don't even bother to connect themselves to the research field of the journal. Heck, even as a graduate student I was once sent a videotape by a man who had decorated his trash can to look like a robot and explained in great length how the "robot" can would react to All in the Family episodes so that it could save humanity by terraforming the Sahara desert.
In a highly interconnected world with people working at all sorts of levels of quality, a lot of this type of junk shows up. It's entirely appropriate for an editor to reject category #3 and #4 material, so as not to waste everybody's time. I really don't want to ask people to review plagiarized or insane material. I also don't want to waste time asking them to review things that are probably science of some sort, but still absolutely certain to be rejected. It is thus the case that nearly every editor will be asked and find it justified to perform some degree of pre-evaluation and desk rejection.
Where it becomes difficult, however, is that different editors and different journals have different standards for determining where to draw the line. When authors disagree with the journal, it can get rancorous, especially for the "glamour" journals that can be so high value for authors to publish in. Thus, we hear primarily about the mistakes, the borderline cases, and the places where politics and prejudice hurt the scientific process.
The key questions are this:
- What percentage of the time is the system going wrong?
- What can be done to improve it?
I'm almost certain that #1 is a pretty small percentage, simply because there's a lot of detritus that nobody is likely to object to desk rejection on. Despite that, the number may well be too high: study of this topic would be both useful and likely very difficult, given the subjective nature of the judgements. There clearly are places where the system goes wrong, and reducing that fraction will likely be good for all of us.