Sylvie Noël submits to us the following fallacy—which appears in the editorial of a peer-reviewed journal. The editor-in-chief makes the following observation:

(…) only a small fraction of the top 100 papers ranked by the number of citations (17 of 100) were published by single authors.

From this, he makes the following conclusion:

(…) a published paper resulting from collaborative work has a higher chance of attracting more citations.

I leave it as an exercise to figure out why the logic is wrong.

But the implication is that solo authors are less interesting. Instead, I believe that solo authors probably work on different problems. (Hint: This could be the subject of a study of its own!)

Why?

Because of something I call problem inertia. For collaboration to occur, several people must come together and agree to a joint project. Sometimes money is required to pay the assistants or the students. All of these factor means that small problems or risky problems will be ignored in favor of safe bets. To put it bluntly, Microsoft will not sell PHP plugins! Hence, statistically, teams must be deliberate and careful. Also, fewer problems can be visited: even if the selected problem is a bad one, changing the topic in mid-course might be too expensive.

An autonomous author can afford to take more risks. Even more so if he has a permanent position. This may explain why Peter Turney seems to believe that researchers lack ambition. They may simply be rational: if it takes you three weeks to even get started on a project, you cannot afford many false starts!

Consider Seth Robertsself-experimentation strategy and Roberts’ take regarding funding:

One reason my self-experimentation was effective was it didn’t depend on grants. No matter what I found, no matter how strange or upsetting or impossible or weird the results might be, I could publish them and continue to investigate them.

This view supports my theory that solo authors will work on different problems.

Note: I do hold a couple of research grants currently. I write almost all my research papers with others.

3 Comments

  1. > I leave it as an exercise to figure out why the logic is wrong.

    Because you have to compare it to the proportion of single-author papers in the population of papers as a whole, which may be even lower than 17 percent?

    (I’m not a mathematician, so when you ‘leave things as a problem for the readers’ you leave things that may genuinely perplex me).

    Comment by Stephen Downes — 27/2/2009 @ 17:00

  2. @downes

    Thanks.

    “If you can’t spot this fallacy right away, what about the other fallacies around you?”

    (That’s what I am going to tell my students.)

    But I am more interested by the tension between “solo” work (I call it “autonomous”) and the touted “collaboration” (sometimes called “managed research”).

    Not that collaborating is wrong… but *requiring* collaboration means that some stones will not get turned, I think.

    Comment by Daniel Lemire — 27/2/2009 @ 17:45

  3. I wonder what fraction of the population–even of the population trained in statistics–can reason correctly about base rates. The research I’ve read on the subject is pretty depressing.

    http://en.wikipedia.org/wiki/Base_rate_fallacy

    Comment by Daniel Tunkelang — 27/2/2009 @ 20:06

Sorry, the comment form is closed at this time.

« Blog's main page

Powered by WordPress