interdisciplinary work may not suck

Shi, Adamic, Tseng, and Clarkson have a cool new article on the impact of interdisciplinary work. Using citations of academic articles and patents, they report the following:

We find that a publication’s citing across disciplines is tied to its subsequent impact. In the case of patents and natural science publications, those that are cited at least once are cited slightly more when they draw on research outside of their area. In contrast, in the social sciences, citing within one’s own field tends to be positively correlated with impact.

It’s a nice article. I recommend that you look at it.

Here’s my comment: We have two views on interdisciplinary work. The skeptics think that most interdisciplinary work fails because you have to water down both sides. The advocates believe in a “structural holes” argument. Group A may benefit from knowing what group B knows. Knowledge may be complimentary.

My personal view is that is that research is a low average/high variance business. Most of the time , we’re better off deepening our knowledge and the cost of inter-disciplinary work is too high. But progress can often be made by taking a gamble on combining ideas. Here’s more evidence for that view:

By plotting mean proximity as a function of impact, we observe that both very low and very high impact papers tend on average to cite outside of their area more often. Since very low impact publications include many publications that cited outside of their discipline but failed to attract notice, we are left with the portion of cited publications where citing outside of ones discipline is positively correlated with impact. These results suggest that citing outside one’s discipline is a gamble. While risking not being cited at all, publications that incorporate work from other disciplines tend to make more significant contributions.

Now, the question is – why are the social sciences the exception here? Knowledge is knowledge. My hunch is that social scientists have a taste for group identity, to take an Akerlof spin on it. An anthropologist cites other anthropologists to show in-group identity, as do most other social scientists (e.g.,  sociologists citing sociologists/econ/etc). Citing too many articles from other fields is interpreted as not being really in the group. If it’s true, then we’re really missing out.

Written by fabiorojas

September 18, 2009 at 12:42 am

Posted in academia, fabio, networks

4 Responses

Subscribe to comments with RSS.

  1. Ideas generated cross disciplines usually either work very well or very poorly in part because if the concept being borrowed was an easy fit it probably would have been discovered within the discipline to begin with. So this begins to appear as a high risk/reward concept where an outside idea will either provide very fertile thought or simply wither. I’ve seen this when driving innovation concepts cross industry.


    Fred H Schlegel

    September 18, 2009 at 4:11 am

  2. Take a look at this article by , forthcoming in Social Forces. The authors find that knowledge brokerage entails higher risks even when one looks at the growth of whole technology domains. Furthermore, they show that brokering knowledge across technology domains (as opposed to specializing within one domain) can be either good or bad, depending on the domain’s degree of specialization.



    September 22, 2009 at 11:44 am

  3. Sorry about that. The article I was referring to is by Carnabuci and Bruggeman and can be found here:



    September 22, 2009 at 11:47 am

  4. […] to solve real problems that business folks face. Hence, I was very interested when Fabio Rojas at Orgtheory commented on research suggesting that the natural sciences are much more accepting of crossing […]


Comments are closed.

%d bloggers like this: