Daniel Lemire's blog

, 4 min read

Let us abolish page limits in scientific publications

As scientists, we are often subjected to strict page limits. These limits made sense when articles were printed on expensive paper. They are now obsolete.

  • But we still need to print the articles on paper! At least in Computer Science, almost everyone has adopted electronic media. It is cheaper and more convenient. I carry thousands of research papers on my laptop: I would require a part-time archivist to get the same result with paper. And 99% of all references are a mouse click away. Given a research paper, I can quickly search through it for interesting terms. It is true that paper versions can sometimes be handy. However, we have this marvelous technology called the personal printer. You can get one for $100. And these printers are connected to computers smart enough to print just the pages you need. You need to review the proof of a theorem on paper? Just print out the proof, specifically. Most people who can afford access to printed journals can afford a printer and the printing costs.

  • Reviewers prefer to review short papers. It can be more difficult to review a short paper than a long paper. I speak from experience. For example, I am currently reviewing papers for ACM RecSys where we have two tracks: short and long papers. It takes me just as long to review short papers. Indeed, reading the text itself is not the bottleneck. What takes the bulk of my time?

  • Checking the literature is time consuming. I often ask myself: did they really advance the state-of-the-art? Other times, I want to check how the submitted manuscript differ from previous work from the same authors.

  • Reviewing the methodology or the mathematical proof also takes me a long time, especially when the authors have omitted details.

If the authors expand unnecessarily on uninteresting aspects of their work, or spend much time reviewing elementary facts, it does not slow me down much because I can easily skip it, as long as the work is well structured. In fact, I find that I can get the gist of an entire Ph.D. thesis, if it is well written, faster than I can understand some short research papers. To summarize: the number of pages is not the primary factor determining how long it takes to review a paper. The problem is not that papers are too long, rather it is that they are often written too poorly.

  • __We want to entice authors to be concise.__ Everything else being equal, a concise text will be better written and easier to read than its longer counterpart. However, everything else is not equal. For example, Venkatesh Rao’s [brief History of the Corporation](http://www.ribbonfarm.com/2011/06/08/a-brief-history-of-the-corporation-1600-to-2100/) is a blog post containing 7000 words. It is an order of magnitude larger than most blog posts. Aren’t Internet users supposed to suffer from attention deficit? Surely, nobody has time for such a long blog post? Yet it has become a classic. It has been extensively covered by various Internet news sites and forums, cited thousands of times. This is no excuse to use long and complicated sentences or to repeat yourself: Rao is an expert writer even though he writes long blog posts. So, while it is true that we have little tolerance for boring ramblings, what matters is less the length of the text, and more how interesting it is.

    Thankfully, page limits are going away, slowly. Adam Marcus sent me a link to the UIST call for papers where they are openly flexible regarding page limits:

    While we will review papers longer than 10 pages, the contributions must warrant the extra length.

    Similarly, John Regehr sent me link to the OOPSLA call for papers:

    The length of a submitted paper should not be a point of concern for authors. Authors should focus instead on addressing the criteria mentioned above, whether it takes 5 pages or 15 pages. It is, however, the responsibility of the authors to keep the reviewers interested and motivated to read the paper. Reviewers are under no obligation to read all or even a substantial portion of a paper if they do not find the initial part of the paper interesting.

    Further reading: Stephen King made a killing his novel The Stand. Yet it spans nearly 1500 pages. Rao wrote several posts on why he shouldn’t be expected to use few words: Seeking Density in the Gonzo Theater and Just Add Water.

    Update: According to an anonymous reader, copy editing is often charged by the number of pages. So it can cost twice as much to publish a paper twice as long, even if you only publish it electronically.