Transactions growth

On a similar tip as my comments on ISIT’s size, the IT Society’s Board of Governors formed an ad-hoc committee to talk about the ballooning-out-of-control of the Transactions. They filed a report at the ISIT in Seattle. Some highlights:

  1. “… the exponential curve, which corresponds to an annual growth rate of 7.5%, is by far the best fit to the data… we may well find ourselves in the following situation within less than 10 years. The Transactions will be publishing over 10,000 pages per year…”
  2. “… it appears the growth will be sustainable as long as we charge for the cost of producing and mailing hard copies.”
  3. The average length of a regular paper was 10.3 pages in 1989 and is 15.8 pages in 2006.

Something that is distinctly missing from the plots provided in the report is the variance of paper length or a histogram of paper lengths. Although papers of average length 15.8 doesn’t seem so bad, if the distribution has a heavy tail some other solutions might present themselves.

The report describes several possible solutions, including doing nothing, splitting the transactions into two journals, limiting page lengths, going to all-electronic publishing except for libraries, etc. The recommendations they make are threefold:

  1. Go to all-electronic publishing except for libraries. This limits the financial burden.
  2. Make a hierarchical organization of the editorial board with sub-editors in chief for different areas.
  3. A 5-page limit for correspondence items

I’m not sure how I feel about all-electronic publishing. Libraries want to have hard copies of things for archival purposes, and this seems to be a neat way of passing the buck to them — will more expensive binding mean more cost on their institutional subscription? On the one hand, you save money by printing fewer copies, but on the other the hard copies may cost more. Probably this saves money all around though.

The sub-editing is a way of making the large page numbers tractable. They specifically don’t want to narrow the scope of the journal, which is good, but they also don’t want to cut more papers by making the acceptance rate lower, so the only way to keep the quality high is to add more reviewers and editors. But they simultaneously note that the number of reviewers is not increasing at the same rate as the number of pages.

The 5-page limit is somewhat odd — ISIT papers are already 5 pages and the difference would seem to be better peer-reviewing and very long publication delay. While this would help control the page numbers, they specifically do not recommend adding a page limit for regular papers. What happens in the gap between a 5-page idea and a 15.8 page regular paper?

Taken together, the proposed solutions seem to consciously avoiding taking the problem head-on. A more aggressive solution might include things like

  • Imposing page charges. At the moment the policy is:

    Page Charges: If the manuscript is accepted for publication, the author’s company or institution will be requested to cover part of the cost of publication. Page charges for this journal are not obligatory nor is their payment a prerequisite for publication. The author will receive 100 free reprints without covers if the charge is honored. Detailed instructions will accompany the proof.

    The Signal Processing Transactions charges $110/page up to 8 pages and $220/page thereafter. Making page charges mandatory for longer papers or adding a mandatory charge per page for papers over 20 pages may encourage authors to write shorter (and perhaps more readable?) papers.

  • Encouraging editors and reviewers to more aggressively promote correspondence items. They bring up the point that correpondence items suffer from “inflated introductions and similar excesses, designed by the authors in order to avoid the correspondence classification.” If there are more specific instructions to editors to… edit things down, then both regular papers and correspondence items can be trimmed.

In the end, the recommendations of the board are more carrot than stick, which may work or may not. The big message seems to be that exponential growth is good but it needs to be somehow managed. However, it may be that more active feedback is needed to control this system which is going unstable exponentially. It has been noted that there is insufficient communication between the information theory and control communities…