The current issue of Accounts has an interesting article by Dan Wang called “Is there a Canon in Economic Sociology?”. It’s a study of the contents of more than fifty Econ Soc syllabuses looking to discover which authors are most often assigned. (I don’t remember seeing the call for the data, which is odd.) There’s a lot of interesting stuff in there, including a variety of measures of “canonicity” and different ways of counting the importance of different texts and authors. Once you start thinking about it, there are all kinds of complications involved in deciding how to code and classify things. Here I just want to higlight an interesting aspect of this network of references:

According to the article, this picture presents the largest component of reference class session co-listings. “Nodes represent references, node size reflects degree centrality, and more orange nodes reflect higher degree centrality. A tie between two nodes signals that two nodes have been co-listed in the same class session on at least two separate syllabi. Tie thickness reflects the number of syllabi on which two references were co-listed in the same class session.” Note that the unit here is articles, so authors may appear in different places in the figure based on different works of theirs.

Two things struck me about this. First was that the visualization is consistent with the field characterization in Marion Fourcade’s ABS piece from a few years ago—you’ve got the structural/embeddedness people and the broadly cultural/Zelizerian work forming one large group, and then (disconnected from both) the insurgent social studies of science/finance people. Second, though, was that the network is quite fragile. But, second, the big component in the network is fragile. If you deleted Geertz (1978), Granovetter (2005), and Swedberg (2001), then you’d have four separate components which you might crudely characterize as soc of finance, culture/Zelizer, Granovetter/network embeddedness/social capital, and Polanyi/political embeddedness. Moreover, two of the bridge pieces are more reviews than research pieces: the Granovetter 2005 is his JEP piece, I think, and the Swedberg piece is his “Sociology and Game Theory” paper, I believe. The Geertz paper (the Bazaar one) is a surprisingly tenuous bridge between the structural and the cultural approaches.

Another thing I’d be interested in seeing is the list of actual works the labels refer to—most of them I know unambiguously, but there are a few that are ambiguous (because the author published more than one thing that year) and I’d be interested in seeing which one is being counted.

Update: Duh, as Omar points out, this is a network where readings are tied if they are assigned in the same class session which makes the fragility interpretation go away. This is of course mentioned in the caption I quoted but evidently did not read properly. As he says,

If this was a co-citation network, then yes, the inference to “tribalism” follows. However, here a tie indicates that two readings are classified as similar by the relevant gatekeepers. So, I think that rather than giving you a picture of the socio-intellectual structure of the subdiscipline, this network simply gives you a picture of its cognitive or classificatory structure. So it is a good thing that the network is easily fragmented, otherwise economic sociology would be a classificatorily incoherent subdiscipline.

And of course Omar is correct here.