Thomas Schelling was a famous economist. He won a Nobel Prize, published agenda setting books, and influence Cold War policy. He also wrote one of my favorite papers, on radically time-inconsistent preferences (specifically, the curious case of asking your friends not to give you a cigarette when you later ask for one, and related phenomena). One the things Schelling is most famous for is a simple agent-based model of residential segregation known as the “Schelling Model.” The model was first published in a 1971 article in the second issue of The Journal of Mathematical Sociology, and was discussed at length in Schelling’s very widely-read Micromotives and Macrobehavior. Many readers of this blog have likely come across the model; the original article now has over 3600 citations on Google Scholar and the book has over 6600.
James Sakoda was a relatively unknown computational sociologist. His main lasting fame seems to be among origami enthusiasts, e.g as the author of Modern Origami. But, as detailed by a fantastic (and lengthy) new article by Rainer Hegselmann, Sakoda may well have the better claim to having first invented checkerbaord models of discrimination. In his 1949 dissertation, Sakoda lays out a class of checkerboard models. And in a 1971 article in the first issue of The Journal of Mathematical Sociology, Sakoda published “The checkerboard model of social interaction.” As Hegselmann shows, Schelling’s model is very nearly a special case of Sakoda’s more general treatment published one issue earlier. Yet Sakoda’s articles boasts just a bit more than 200 citations, and no one speaks of a “Sakoda model.” What happened?
Likely Schelling had not read Sakoda’s work and vice versa, so the case seems to be one of simultaneous discovery rather than anything nefarious. Nonetheless, the historical process by which Schelling rather than Sakoda became synonymous with checkerboard models is a fascinating one. Hegselmann identifies several factors, including Schelling’s greater prominence (the traditional Matthew effect story), the tractability of the models (Schelling’s model could be “run” on a literal checkerboard; Sakoda’s required a computer and programming knowledge beyond most social scientists in the 1970s), and Schelling’s promotional work (Schelling re-described the model in his book and elsewhere; Sakoda moved on to promoting computational social science more broadly including his own programming language developed for social scientists). By the time personal computers became widespread, social scientists identified checkerboard models with Schelling even as they gained the capability to implement Sakoda’s more general version.
Beyond the interesting and well-told intellectual history, Hegselmann also describes the details of Sakoda’s fascinating life. Sakoda did his dissertation research as an assistant for Dorothy Thomas, who he met when he was a student at Berkeley and she was a new faculty member. During World War II, Sakoda conducted participant-observation as part of a larger team studying Japanese internment camps. That is, Sakoda was actually involuntarily interned from 1942 to 1945, and used the opportunity to conduct research on behalf of Thomas. Thomas published two books based on his (and other) research; Sakoda himself later published a short version of what would have been the third volume.
Hegelsmann ends the piece on a somewhat ironic note. He writes a recipe for a sort of reverse Matthew effect – how not to get credit for a brilliant idea:
If asked for a recipe for how to become an unknown pioneer, we might recommend the following:
1. Be brilliant, early, and exclusive: Have a good idea that almost all others can’t pick up because they do not have the technical equipment or do not command it.
2. Be modest: Do not promote your model—even if obviously your time has come. Do not care about publishing, no strategic considerations, focus on the technical side, no publication campaign!
3. Trust the Matthew-Effect: There will be well reputed others that will reinvent or pick up your idea—and then it will spread. (126)
If you are at all interested in the history of sociology, economics, or computational social science, read the whole article!