Editorial: A Bill of Rights and Responsibilities

This editorial was originally slated to appear in May (as my previous three have), to commemorate the end of my third year and first term as editor-in-chief of JACM, but due to what I view as a very significant development that has been in the works for the past few months, I've held off publication until now.

The issue that I've worked hardest on has been to make JACM more author friendly. Perhaps the most obvious manifestation of this has been an effort to cut down the time spent dealing with papers; my annual summary of how we are doing on that score appears below. The other work has been carried on more behind the scenes, trying to influence ACM (and thus, indirectly, other publishers) to take authors' concerns more into account. I am very pleased to report that ACM is more committed than ever to doing so.

In the process of dealing with issues, ACM has prepared a Bill of Rights and Responsibilities, outlining the rights and responisiblities of readers, authors, reviewers, and editors. The ACM headquarters staff has committed to implementing these rights within the next year. Here are four authors' rights that I view as particularly significant:

• The right to be informed of the production schedule of a paper: Perhaps the most egregious violation of this right is the experience that I suspect many of us have had of having a paper accepted, not hearing anything about it for well over a year, and then suddenly having galleys appear in the mailbox that must be returned within 48 hours. The fact that authors may not see galleys for over a year is simply a function of a journal's production schedule--other papers must be processed first. However, it seems reasonable to expect that, when a paper is accepted, authors will be given an estimate of when it will appear and what the production schedule is. Moreover, there should be a warning well before the galleys arrive, so that authors can allot time for proofreading (and inform the publisher if the timing is inconvenient). ACM intends to set up a tracking system to monitor the progress of papers in review. It has also committed to informing authors about the progress of accepted papers and warn authors well in advance as to the scheduled arrival of galleys.
• The right to have all copyediting changes clearly noted: Publishers typically send galleys to authors for them to proofread before final publication. However, the standard practice among publishers is that these galleys typically contain absolutely no indication of the changes that were made to the paper by copyeditors. That means that the paper must be proofread very carefully to find such changes. While proofreading a paper before publication is certainly a good idea, changes made by copyeditors can sometimes be inappropriate. It certainly would help authors to have all changes clearly marked. Publishers typically don't do this simply because it is more work for copyeditors to mark such changes. However, ACM has agreed that authors that request marked-up galleys will be able to have them.
• The right to have no errors introduced by the production process: Most JACM authors prepare papers in LaTeX. ACM then converts these papers to SGML (Standard Generalized Markup Language) and from there to PDF. If papers are prepared using the ACM style file, this makes the conversion to SGML go more smoothly. However, even if the ACM style file is used, some information regarding the presentation of mathematics is lost; SGML is simply not as good as LaTeX at representing mathematical information. This information is reintroduced in the process of converting from SGML to PDF. Not surprisingly, since the people doing the conversion are not mathematically trained, errors are introduced as well. There are advantages to SGML over LaTeX--SGML better represents meta-data about the paper. However, the use of SGML means that authors have to proofread a paper more carefully, to catch errors introduced by the conversion process. Personally, I would far prefer to go directly from LaTeX to PDF (perhaps also saving a version of the paper in SGML format if desired), to minimize the effort for authors. This issue is still under debate by the ACM Publications Board. I will report on the outcome of the debate as soon as it is settled. However, what is not under debate is that ACM must somehow strive to ensure that no errors are introduced in the production process.
• The right to post the paper on public repositories as well as home pages: I've discussed this before. Pretty much all publishers in computer science allow authors to post even final version of a paper on their home page (with appropriate copyright acknowledgment). The real issue is whether authors should also be allowed to post accepted journal papers on public repositories such as CoRR (http://arXiv.org/intro/cs.html). ACM allows authors to post any version of their paper up to the one copyrighted by ACM (which means it may include changes made in response to reviewers' comments). In addition, ACM has, as an experiment, allowed authors to post even the final version for the past year. The experiment will continue for at least another year (and, I hope, forever).

I have focused on the four problems above because they represent relatively standard practices of publishers that, as an author, I have found particularly annoying. I applaud the efforts of ACM to modify their policies and production processes to be more author- and reader-friendly, and hope that other publishers will follow suit. If any of you have other problems that you consider important, please do not hesitate to bring them to my attention. This is a good time to act!

Let me conclude this editorial, as usual, with a summary of time to publication. The data is from May, 2000, just to be consistent with previous reports.

Of the 132 papers that were active when I took over (in May 1997), there is only one still in the system, and that because the authors had it for 2.5 years and just sent it back to us a few months ago after a major revision.

Of the 100 papers submitted between May 1997 and May 1998, only three are still in the system. Of the remaining 97 papers, 34 were accepted, 60 were rejected, and 3 were withdrawn. The median time in the system for these papers was 11 months. However, that number is a bit misleading, because it takes far less time to reject a paper than to accept one. Of the 60 papers that were rejected, the median time to rejection was 4 months, with a minimum time of one day (it is sometimes very easy to reject a paper!) and a maximum on 23 months. I am not happy about the 17 papers that spent a year in the system before being rejected, and will try to work harder on that. Of the 34 papers that were accepted, the median time in the system was 16 months, with a minimum of 8 months and a maximum of 30 months. It is not surprising that papers that are accepted should spend longer in the system. They almost always undergo one revision and sometime more. The ``time in the system'' also includes the time spent in authors' hands being revised in line with reviews, and this can sometimes be quite considerable.

Of the 95 papers submitted between May 1998 and May 1999, 19 have been accepted, 51 have been rejected, 2 were withdrawn, and 10 are with authors, and 13 are with reviewers. These numbers are quite comparable to those of last year.

I have not discussed yet how long it takes for a paper accepted to actually appear in print. In fact, the queue is relatively short -- about 3-4 issues -- so a paper should appear within 6-8 months of being accepted. More importantly, if authors send us postscript, we post the paper on the JACM web site (with an indication that it has been accepted) within a few days of receiving the final version. As I mentioned above, I hope shortly to be able to also include the information about when it will actually appear.

Joe Halpern
(halpern@cs.cornell.edu)