I have now completed two years as editor-in-chief of JACM. As I have done in the previous two years, I would again like to take stock of the situation.
From my (admittedly very biased) point of view, perhaps the most significant event in the past year was the introduction of CoRR, the Computing Research Repository. This central repository for the computing literature started in September 1998 and now houses over 1500 papers. I hope that you will take advantage of CoRR and submit all your papers there. This should greatly benefit the community.
As further encouragement, ACM has agreed to a two-year experiment (which I hope will be extended indefinitely) that allows authors of papers that appear in ACM journals to post the final, published version on CoRR. I would like to encourage authors submitting papers to JACM to post them on CoRR. Authors who want to submit to JACM and have posted their papers on CoRR can just send us the URL rather than hardcopy or Postscript. (For those who are concerned with anonymity, note that there is no way of telling whether a paper on CoRR has been submitted to JACM unless the authors choose to make that fact public.)
If CoRR catches on (as I hope it does), then I expect the role of journals to change significantly. Journals will act far more as certifying services--the fact that a paper has been accepted by a journal such as JACM will attest to its high quality--rather than as sources of papers. But this will not diminish their importance at all, in my view. As users of the web know, filtering will be critical as we are increasingly flooded with information.
Rather than make further predictions about the future of journals, let me now turn to issues closer to home. Last year's editorial focused on time to publication and promised more detailed data. Here it is.
Of the 132 papers that were active when I took over in May 1997, there were 24 in some stage of review last year and 20 more under revision by authors. Now (in May 1999) only 6 of these papers remain in some stage of review and 3 more are with authors (2 of which have been accepted and are awaiting final minor revisions). While I am not happy that even 6 papers have been in the system for two years (and, in some cases, much more), I am pleased with the progress.
Only 100 papers were submitted between May 1997 and May 1998. Of these, 23 were accepted, 56 were rejected, 7 are with authors, 2 were withdrawn, and 12 (i.e., 12%) are in some stage of review. The median length of time in the system (either in review or with authors) for these 100 papers was 10.5 months. (This figure includes the papers still in the system; they have all been in the system longer than the median time.) The median time in the system for papers that were rejected was 3.5 months; the median time for papers that were eventually accepted was 15 months. (This reflects the fact that accepted papers typically involve at least one round of author revisions and re-review; the time spent in the authors' hands is included in the 15 months.) Although the latter two figures do not include the papers still in the system, I do not expect these papers to change the numbers significantly.
Again, while these numbers are not quite as good as I would like--the median is hiding a few outliers--it does show that we are making significant progress. In particular, the emphasis on quick initial reviews seems to be paying off.
Next year I will report the numbers for the papers submitted between May 1998 and May 1999.
The last issue I want to discuss is coverage. One of my goals when I took over as editor-in-chief was to broaden the scope of the journal. I have not been quite as successful as I had hoped. A quick glance at the recently accepted papers shows that, while our acceptances do cover a broad spectrum, there is still a strong emphasis on theory (as opposed to systems), especially on data structures and algorithms. (Although these papers only indicate what was accepted, not what was submitted, they reflect the general trends in submissions quite accurately.) I do not plan to reduce the number of papers in areas where JACM has traditionally been strong (as long as they continue to be of the same quality as those we have been accepting); however, I do want to increase the number of papers we accept in other areas. Since we now publish six issues a year, we should have sufficiently many pages to accommodate such an increase.
It clearly takes time to broaden the scope of a journal and convince authors not used to submitting their papers to do so. Nevertheless, my goal remains that of making JACM a journal that provides coverage of the most significant work going on in all of computer science, broadly construed, not just selected portions of it. In line with this goal, I'd like to announce a new addition to the editorial board--Jeannette Wing will be handling a new area Software Engineering. In the past, papers on Software Engineering were typically submitted to the area Programming Languages and Methodology. This area has been renamed to just Programming Languages; it will continue to be handled by Robert Harper. Bob and Jeannette will coordinate on submissions at the borderline.