This edition of CCR bears a dubious distinction of having no technical articles, only editorial content. This is not because no technical articles were submitted: in fact, there were 13 technical submissions. However, all of them were rejected by the Area Editors on the advice of the reviewers, a decision that I did express concern with, but could not, in good conscience, overturn.
One could ask: were all the papers so terrible? Certainly some papers were unacceptably bad and some were simply out of scope. However, the fate of most papers was to be judged to be not good enough to publish. Some submissions were too broad, others too narrow, many were too incremental, some too radical, and some were just not interesting enough. The opposite of a Procrustean bed, CCR has become a bed that no paper seems to fit!
This, by itself, would normally not cause me too much concern. However, I feel that this attitude has permeated our community at large. A similar spirit of harsh criticism is used to judge papers at SIGCOMM, MOBICOM, CoNEXT, and probably every other top-tier computer science conference. Reviewers seem only to want to find fault with papers, rather than appreciate insights despite inevitable errors and a lack of technical completeness.
I think that a few all-too-human foibles lie at the bottom of this hyper-critical attitude of paper reviewers. First, a subconscious desire to get one’s back: if my paper has been rejected from a venue due to sharp criticism, why not pay this back with sharp criticism of my own? Second, a desire to prove one’s expertise: if I can show that a paper is not perfect, that shows how clever I am. Third, a biased view of what papers in a particular area should look like: I’m the expert in my field, so I think I know what every paper in my field should look like! Finally, unrealistic expectations: I may not write perfect papers but I expect to read only perfect ones. I think I have a good understanding of the psychological basis of reviewer nitpicking because I too am guilty of these charges.
These subconscious attitudes are exacerbated by two other factors: a ballooning of reviewer workloads, and, with journals in computer science languishing in their roles, conference papers being held to archival standard. These factors force reviewers into looking for excuses to reject papers, adding momentum to the push towards perfection. As the quote from Voltaire shows, this has negative consequences.
One negative consequence is the stifling of innovation. Young researchers learn that to be successful in publishing in top-tier venues, it pays to stick to well-established areas of research, where reviewers cannot fault them in their assumptions, because these already appear in the published literature. Then, they scale the walls by adding epsilon to delta until the incrementality threshold is breached. This has an opportunity cost in that well-studied areas are further overstudied to the detriment of others.
A second negative consequence is that it turns some researchers off. They simply do not want to take part in a game where they cannot respect the winners or the system. This has an even greater opportunity cost.
How can we address this problem? As PC chairs and Area Editors, we need to set the right expectations with reviewers. No paper will be perfect: that is a given. We have to change our mental attitude from finding reasons to reject a paper to finding reasons to accept a paper. We will certainly be trying to do this from now on at CCR.
We can also remove the notion of a publication bar altogether. An online version of CCR, which will be coming some day, could easily accept all articles submitted to it. Editors and reviewers could rank papers and do public reviews and readers can judge whether or not to read a paper. This is already common practice in physics, using the Arxiv system.
Finally, I would urge readers to look within. As a reviewer of a paper, it is your duty to critique a paper and point out its flaws. But can you overlook minor flaws and find the greater good? In some cases, I hope your answer will be yes. And with this small change, the system will also change. One review at a time.