Christopher X J. Jensen
Associate Professor, Pratt Institute

In the Light of Evolution V: Cooperation (An Abbreviated and Incomplete Social Network Analysis of the Speakers)

Posted 04 Jan 2011 / 0

In a couple of days I will be attending the In the Light of Evolution V: Cooperation colloquium, and in preparation I have been reviewing the work of the featured speakers. Tomorrow, I will be posting a preview of the meeting.

The meeting promises to be broad and interdisciplinary. It also features a very elite group of speakers, certainly the largest collection of accomplished scientists who I have heard speak at a single meeting. It is clear based on their individual accomplishments that each of the speakers deserves to be presenting, but given that certain prominent evolution-of-cooperation researchers were not in the line-up, I was curious to see how interconnected the invited speakers were. Will this group of speakers provide diverse perspectives, or is this an incestuous subgroup of the overall community of scientists exploring how cooperation evolves? One of the interesting things about science is how small a world it can be. Although the number of researchers publishing their work can be daunting, often they publish collaboratively in interconnected patterns that drastically reduce the effective size of the scientific world.

To figure out how interconnected the speakers at this meeting are, I used ISI Web of Knowledge to search for the number of published papers sharing a particular pair of invited speakers as co-authors. Like any database, Web of Knowledge has its imperfections, leading to two types of errors for this sort of search. The first error is one of over-inclusion, where using too broad a search term leads to too many results. The second error is one of over-exclusion, where using too narrow a search term leads to too few results. If an author has no middle initial, avoiding one or the other of these errors becomes more difficult. For instance, searching for “Boyd R” yields 1613 publications and searching for “Boyd Robert” yields 151 publications, whereas searching for “Boyd R AND Boyd Robert” yields only 24 publications. So how many publications does prominent speaker Robert Boyd of the University of California, Los Angeles actually have? Unfortunately it is probably more than 24 and less than 151 (and certainly less than 1613). What I found was the most accurate way to determine the number of papers published by a particular author was to search for the author’s last name and first/middle initials, which produces a large number of publications which may include other authors with different first names but the same initials. I then used Web of Knowledge’sDistinct Author Sets” feature to reduce the list of publications to only those produced by colloquium speakers. Once the list of publications for each speaker was reduced to a fairly accurate set, I used Web of Knowledge’s “refine by authors” feature to count the number of shared publications with other colloquium speakers. I would not claim that this method was perfect, but I believe it achieved the best balance between over-inclusion and over-exclusion. Although this may seem like a daunting task, it actually was pretty easy to zip through thanks to the efficient search interface embedded in the Web of Knowledge database.

The results of my joint publication search are displayed in the following matrix (click on the image to enlarge it):

The list of invited speakers and organizers are displayed in alphabetical order in the first column. Each successive column displays the number of joint publications with other speakers, whose initials are displayed as column headings. The diagonal represents the number of publications each speaker has listed in Web of Knowledge (the number of “joint publications” with oneself), and the data above and below the diagonal are mirror images of each other and thus redundant. Most of the entries in the matrix contain zeros, representing two speakers who have never published together. I have color-coded those entries that are non-zero; as the key at the bottom of the matrix shows, the darker the purple color, the more publications shared by a pair of speakers.

Although I do not have time to perform formal analysis or visualization of this data, just looking at the matrix one can get some meaningful impressions of the social connection between invited speakers. First of all, it is clear that this is truly an interdisciplinary meeting, as the overall level of connection between speakers is very low. Co-organizers John C. Avise and Francisco J. Ayala share a moderate number of publications (8), most of which stem from previous In the Light of Evolution proceedings papers. The other co-organizers, Joan E. Strassmann and David C. Queller, share a whopping ninety-eight (98) co-authored papers, but if you look at their scientific histories it is clear that they have acted as a team for many years. The other “most interconnected” speakers are Kevin R. Foster and Joan B. Silk, with Robert Boyd also sharing some co-authorship with two other speakers. The vast majority of speakers have not co-authored any papers with other speakers. So what does this mean? My interpretation is that each of the invited speakers represents a distinct scientific arena and/or perspective. The overall lack of “scientific incestuousness” suggests to me that the organizers did an excellent job of lining up speakers who will bring unique ideas to the meeting. I am excited to see if this proves to be true.

Perhaps most intimidating is the overall number of publications produced by many of the speakers. A trip down the diagonal of the above matrix demonstrates just how much scientific accomplishment will be assembled at the Beckman Center. Seven of the speakers have over one hundred publications, with several others close to this mark. Given the overall lack of overlap between the work of the speakers, this should be a really challenging, varied, and interesting meeting.

Conferences, Cooperation, Social Network Analysis

Leave a Reply