---
bibtex: @incollection{Hellie2017-HELDLA-2,
publisher = {Oxford: Blackwell},
year = {2017},
booktitle = {Philosophy's Future: The Problem of Philosophical Progress},
author = {Benj Hellie},
title = {David Lewis and the Kangaroo: Graphing Philosophical Progress},
editor = {Russell Blackford and Damien Broderick}
}
---
Data-driven historiography of philosophy looks to objective modeling tools for illumination of the propagation of influence. While the system of David Lewis (1941–2001), the most influential philosopher of our time, raises historiographic puzzles to stymie conventional analytic methods, it proves amenable to data-driven analysis. A striking result is that Lewis only becomes the metaphysician of current legend following the midpoint of his career: his initial project is to frame a descriptive science of mind and meaning; the transition to metaphysics is a rhetorically breathtaking escape from this program’s (inevitable) collapse. Understanding this process both aids a more focused debate whether it counts as progress, and also presents novel affordances for partisans on both sides to learn from Lewis’s right and wrong steps.
Benj employs a data-driven, impartial, model-building approach to history of philosophy. (p2)
Progress ...
If there is progress in philosophy, it comes about through progression in philosophy: through development and change over time in what philosophers write. (p1)
Method ...
My approach has been to construct force-directed graphs of Lewis’s autocitations: citations of his own work (nodes are publications, edges joining them represent mutual relevance, the spatial arrangement works out on its own when nodes try to get away from each other but are constrained by edges). (p3)
My database of Lewis’s autocitations (compiled by hand) covers 129 works published by 2014, each of them assigned a date.3 Of these, 99 systemic works either cite or are cited by one another: there are 270 episodes of autocitation in all. With the assistance of David Balcarras, I have constructed force-directed visualizations of network graphs extracted from the data (Hellie 2016).
These are of two kinds: graphs of development and of subject-matter. Development graphs depict the raw data of what cites what, and when. So edges are “directed”: there is a meaningful difference between the “source” of an edge and its “target”—namely, the source is the citing work, the target the cited work.