Samuel Arbesman and Nicholas Christakis of Harvard's Institute for Quantitative Social Science have a short and interesting article in the current issue of PLoS Computational Biology on the science of scientific discoveries. It's more a position piece observing that quantitative analysis of scientific discoveries is an emerging field, but it does include one lovely set of figures :

(A) Mean diameter (kilometers) of minor planets discovered, 1802–2008. (B) Mean physical size (g) of mammalian species discovered, 1760–2003. (C) Mean inverse of atomic weight of chemical elements discovered, 1669–2006.

The authors are open about the obvious problem with "eurekometrics," i.e., that scientific discoveries aren't easily systematically defined. That said, that sort of difficulty is pretty common in bibliometrics and scientometrics research, and trying to come up with quantifiable proxies for relevant concepts is part of the fun (and sometimes leads to fruitful metrics, e.g., the H index). All in all, the piece is short and worthwhile and ties in with a bunch of other interesting recent bibliometrics papers (e.g., Wuchty et al.'s Science paper on teams) for those who are inclined. Check it out.


  1. This is great. Although I'm not sure I buy that the atomic weight graph is the same story, since we've been making a lot of the heavier elements in the lab.

    For a while I've wanted to write a paper asking if the speed that pollutants are regulated reflects how discernable they are to the human senses. Eg. pollutants that are smelly or very visible probably are regulated faster than equally harmful pollutants that have no smell and are invisible to the human eye.

  2. I think if you couple that with a voting model and heterogeneous pollution from point sources you could easily get an optimal time to lobby for pollution reduction. As in, let's wait til we can scrub both the SO2 and the CO2 before lobbying to scrub the SO2, since the marginal visible benefit of abatement is higher for the former.

  3. I really like Lav Varshney's follow up post here.