Musings
February 26, 2002

The case against blind refereeing


.....  


Many journals nowadays send manuscripts out to referees without showing the name(s) of the author(s). This is called blind refereeing. I believe that this is a childish practice that may actually harm the authors, impede the referees, and impact the quality of the journal. Authors should refuse to submit to these journals, referees should refuse to referee, and journal editors should put pressure on the managers to change their policies.


Plain decency  


When I write a letter of recommendation, the most important part of that letter is my name and signature. It is a personal guarantee. When that letter is received by a promotion or hiring committee, I do not want my name to be removed, because it will devalue the letter. The committee won't be served either because it won't be able to put the letter in the proper context. In the same vein, I want my name on my papers, because I am proud of them---otherwise, I would not submit them. I don't want my papers to be accepted under a pretense, or through a backdoor. It's just professional decency. If the world operated using aliases, assumed names, and hidden identities, we would be in big trouble.


Avoiding absurdities  


The referee receives a blind referee report request from Ed, the editor, with the name of the author, Arthur, removed. Here are some absurd situations that may arise:

  • The referee rejects the paper based on overlap with the work of Tim at the University of Timbouctou. He does not realize that Tim=Arthur, and draws the wrong conclusion. In his zeal, Ed, the editor, disregards that recommendation, and accepts the paper, not realizing that the referee has not bothered checking the paper in detail because of the obvious overlap.
  • The referee hates the paper, and writes a scathing report. As it turns out, Arthur is Ed's colleague down the hall, and his drinking buddy. Oops! If the referee had seen the name, he could (and should) have refused to referee that paper.
  • The referee, an expert in his/her area, has received another very similar paper to referee, and immediately rejects the paper. It is unacceptable to submit similar, almost identical material, to two different journals. Unfortunately, by accident, the other paper is from a competing research team, and a terrible injustice has just taken place.
  • The referee spends three full days combing through the paper, and lists all errors diligently in his/her seven-page report. The only problem is that Arthur=Ed, and the referee has just done the dirty work for Ed. You will say "This is impossible, editors do not process their own papers." Well, this has happened to me with an ex-editor in chief of the Journal of Multivariate Analysis (which was not even blinded!). Granted, it only happened once in my career, but it did happen.


Detective work  


The argument that takes the cake is this one: it is soooooooo simple to learn the identity of the author thanks to Google, the Research Index, and other places on the web. Just type a few key words taken from the title or the body of the text, and see the matches appear on your screen. With the latest generation of search engines opening postscript and PDF files, it is virtually impossible to escape scrutiny. So, why make it hard? Why play games?


Fraud and sabotage  


An evil-minded person might grab a technical report from a web site, change the name of the author(s) to his/her own name, and submit it to a "blind" journal. If this is a good paper, it will probably be accepted. Imagine the embarrassment for the editor and the journal, and the anger of the ripped-off author(s). Well, far-fetched? I don't think so. Such a scandal could not occur with a "regular" journal with a serious refereeing process. In fact, then, if you receive a paper to referee from a "blind" journal, there is little else you can do but recommend rejection "based on the plagiarization of the work of (fill in the names of the real authors)", if you happen to be able to deduce who the real authors are. By not doing so, you might play in the hands of the cheats.


Respect  


The blind refereeing process shows total disrespect for the referees. The journals could not live without referees--they are the most important part of the publishing process. Yet, they are often treated like the scum of the earth. They are not paid (commercial publishers should pay, societies should not), they are often listed in an annual list of referees (thus divulging their identities!!!), they are continuously harassed by associate editors ("we need your report by Monday, or else"), and now, they have to work without even knowing whose work they are refereeing. I have esteemed colleagues who never make errors. They spend all their time checking and rechecking proofs and manicuring their work. If I see their names on papers, I know that the work is correct, so I will concentrate on other things, such as motivation, relevance, and so forth. It saves me a lot of time, but oh no, I am scum.


A proposal  


Here is my proposal for referees:

  • If you are a commercial publisher, pay the referees. If your name is Kluwer or Elsevier, pay double.
  • If you are a professional society, give the referees a one-year susbscription to your journal or send them a free book from your book series. Or waive their fees for the next annual conference.
  • Good referees should be flagged, and in the end, they should make good candidates for associate editorships.
  • Do not remove names from manuscripts. Actually, give referees more information on the author by pointing to a web site and a list of publications, so that the referees can put the work in the right context.
  • Do not send the paper. Place it on a web site, and email the URL of that site. No password fiddling, please. A secret URL should suffice.
  • Accept referee reports in any possible format, whatever is most convenient for the referee.
  • Do not publish lists of referees in your December issues.
  • Do not send your referees thank you letters unless you are offering them some of the goodies mentioned above.


  



Copyright © 2001-2002 Luc Devroye
School of Computer Science
McGill University
Montreal, Canada H3A 2K6
luc@cs.mcgill.ca
http://luc.devroye.org/index.html