This past week, during my self-imposed exile, I came across the work of Derek J. De Solla Price, the originator of scientometrics, the study and analysis of science. Much of his work is concerned with which scientific papers get referenced and how often. Now, theoretically, science and its literature should be a rich area to study network effects. Because so much research builds upon previous research, the addition of a new work to the corpus of literature is a meaningful event. Furthermore, the formalized practice of bibliographic citations acts like a proto-hyperlink to show the relationship between papers.
As a means of valuing new internet ventures, like Skype or Flickr, many look to Reed's and Metcalfe's laws which optimistically posit that as any given network grows steadily larger, the value of it grows exponentially. I voiced my skepticism to this a few months ago.
I find it telling that in all the reading I've done of various internet pundits, venture capitalists, and prognosticators, I've never come across Derek J. de Solla Price. One of his most important discoveries, Price's Law states that among a group of scientists, half of the output will come from the square-root of the total group. So, at a university with, say, 100 physics researchers, 10 will be responsible for half of the output, leaving the remaining 90 to publish the other half. This simple idea should be pondered by the whole Reed's Law/Metcalfe's Law/Long-Tail Crowd:
A striking implication of Price’s Law is that the larger the total number of producers in a field, the smaller the 50%-producers group becomes as a percentage of the whole. If it is a group of a hundred poets or chemists, half of the worthwhile creative work will be produced by ten (10% of the whole); if there are 10,000 in the pool then half the work will be produced by a group of mere 100 (1% of the whole). This is presumably because the larger the group the greater the likelihood of it possessing a few highly productive talents, and it explains, why, for example, the nineteenth-century music we listen to is dominated by a tiny minority of familiar composers; try as we might to resuscitate lesser lights (Cherubini, Clara Schumann, John Field), there remains an overwhelming dominance of listener preference in a relatively small handful of composers.
This stands in direct contrast to Reed's and Metcalfe's optimism. As a field expands, value doesn't necessarily grow exponentially, instead a greater percentage of the whole is simply less productive. His study of bibliographic citations also led him to believe that just a few papers get cited frequently, while the majority get cited rarely, and eventually not at all. You can download the paper here. Here is the paper's final paragraph:
Journal citations provide the most readily available data for a test of such methods. From a preliminary and very rough analysis of these data I am tempted to conclude that a large fraction of the alleged 35,000 journals now current must be reckoned as merely a distant background noise, and as very far from central or strategic in any of the knitted strips from which the cloth of science is woven.
A lot of businesses are hoping to profit from the "Long Tail" but perhaps they're just serving as an amplifier to a lot of noise. I think de Solla Price needs to be heard along with the Metcalfes and the Reeds.
Comments