Our friends at Wikimedia Foundation are maintaining an incredibly rich knowledge graph of human knowledge. Whilst browsing the Wikipedia pages of famous philosophers (as I am sure everyone does), I got interested in visualising the influence map between the thinkers, to follow how philosophical thoughts have propagated across centuries and borders. My starting point was the Wikidata repository of all philosophers, enriched by Wikipedia links (I'll dedicate a separate post to the technicalities of obtaining and handling this data). This yields a set of more than 30,000 philosophers, but only ~3,200 of them having any links to the others.
As we can see from the in- and out-edge degree distributions (i.e. "how many people influenced X" and "how many people have X influenced" respectively), most of our philosophers have a low number of both relationships, with a few "super-influencers" standing out (and a "super-influenced" Nietzsche with 44 incoming edges). This distribution seems to be a power-distribution which is usual for scale-free networks (a type of network that describes most of social networks, so it wouldn't be surprising to find the influence network to follow this).Taking a closer look at who our super-influencers are brings up the familiar names. The two ways we can approach this is through direct influences, and indirect (X influenced Y who influenced Z). The latter could be more overarching in terms of capturing the importance of a thought, but also biases heavily towards thinkers who lived in earlier periods, unsurprisingly showing exclusively ancient Greeks among the top 20.
Setting the view to direct shows a different picture with five 19-20th century thinkers making it to the top 10 (Nietzsche, Hegel, Heidegger, Marx and Wittgenstein). Interestingly Aristotle had much more direct influence than indirect (not appearing on the indirect list, but topping it for direct edges). Those descriptive statistics might give us a good high level view, but are not useful for understanding fine-grained behaviour across thinkers. Visualising this as a graph however can be more enlightening.
Temporal plot of influences
Drawing a temporal graph of the influences brings it to life a bit. Click around the graph, to see the direct descendants and ancestors of a particular philosopher. To make this visualisation more managable, we're only looking at the descendants of Immanuel Kant, only spanning 3 centuries this way. The philosophers are also filtered to those with at least 7 connections in total, to further reduce the graph to a managable size. Thanks to the force simulation, the philosophers are located near their birth years along the x axis, and spaced out along the y axis. The languages they spoke and wrote in are represented by the colors within their circles ranging from 1 to an impressive 5 used by Muhammad Iqbal (the overall winner in the dataset in this respect being John Paul II with an outstanding 13 languages spoken). This gives an interesting angle for analysis, helping to answer the question whether philosophical thought is siloed by language or movement.
Language and school of thought
Plotting the distribution of common language and philosophical movements across influencer and influenced shows us which languages were mostly used to "transfer" thought, and whether language or ideological silos are created.
It immediately stands out, that by far the most common type of influence is when there is no common language between the two thinkers, i.e. one read the other through translations, which clearly addresses the language silo assumption. On the other hand, Arabic being the first non-European language with only 126 links in the dataset, as well as Chinese and Japanese being below 10 links speaks to the huge European-bias of the dataset and philosophical canon. On the movement front, however, data quality is probably the key issue - with many missing labels, there isn't anything significant we can say. Most of the thinkers in the dataset don't have any movements attached to them, and above we're only looking at those connections that had language and movement data for both philosophers involved.
Geographical distribution of links
To visualise this Europe-centeredness from another angle, plotting these connections across the globe sheds a new light on the issue:
Drag the red dot to control the visual, and drag or zoom on the globe to see what's going on in all continents. Consistent with the initial exploration of philosophers and their links, we're seeing the red connections cumulate increasingly from the 17th century onwards. Worth noting for this visual, that a connection is presented, at the time of the birth of the "influenced", and connections linger on the map for 150 years, to give us a sliding window view.
Conclusions
Looking at the history of philosophy from multiple angles I was hoping to highlight the richness of the dataset in Wikidata, and the incredible ways it could complement qualitative thinking on questions that are predominantly within the sphere of social sciences.