Memes are cultural replicators that are propagated by people through social media networks. This dissertation is an examination of human-meme dynamics and an effort to perform scientific inquiry with social-computational simulations.


This is a work in progress. The target for completion is 2019.

Lecture 1: Networks of Computational Social Science

Lecture 2: Urban Legends

Lecture 3: #topoli

  • Slides - a PDF containing a slide deck.

Other work

These are not part of the dissertation.

Political Polarization

Miller, I. D. (2018). Political Polarization - Lecture. Has anybody ever unfriended you for your political beliefs? What happens when friendship ties are cut but political ties remain? Simulations suggest that in the absence of friendships, the population bifurcates along party lines. When friendships cross party boundaries, the population does not fragment. Surprising outcome when new friendships are created. This work was presented by Ian Dennis Miller on June 8, 2018 at the Political Networks 2018 conference hosted by George Mason University in Washington DC.

Election Memes

Miller, I. D. (2016). Election Memes - Working Paper. During the 2016 US Presidential election, Internet created a huge quantity of memes about the candidates. In October 2016, a meme aggregator called Sizzle shared a subset of their meme data that was filtered for political content. The data contain over 38,000 memes spanning four social networks: Facebook, Instagram, Twitter, and Imgur. This report first checks the relative quantity of candidate memes across social media platforms. We then test to see whether memes about Clinton or Trump receive more likes. Finally, we visualize the change in meme liking over time. Although liking a candidate meme is not equivalent to liking the candidate, it is safe to say that more likes are similar to more attention. The plot of candidate meme likes over time reveals the monthly competition among US presidential candidates.

Memelab: can pictures of adorable kittens explain political revolutions?

Miller, I. D. (2012). Memelab: can pictures of adorable kittens explain political revolutions? Last year, Tweets across Egypt called a political system into question, while more recently PSY’s Gangnam Style music video received over 300 million views in just two months. These viral phenomena are the product of information flowing through an interconnected population, so could the same viral mechanisms be an explanation for both events? My work uses internet memes to study the way we create and share things online. During the spring of 2012, over 100 UTSC students used Memelab, an online viral research laboratory, to create memes and share them with friends. This presentation includes a history of memes, the results of the first Memelab experiment, and the future of social network simulation.

The Social Transmission of User-Generated Memes

Miller, I. D. (2012). The Social Transmission of User-Generated Memes. (MA Thesis, University of Toronto, Toronto, Ontario, CA). doi:1807/67214. The popular concept of viral media is like the flu: once unleashed, it naturally infects all your friends. This work suggests that viral impact may not be determined by the content alone but also by the content’s creator as part of the Viral Feedback Loop. Participants interacted with a type of online meme called an image macro which has historically been shared virally. Participants made their own User-Generated Content (UGC) with Memelab, an image macro builder written for this experiment. Participants then shared their UGC online, which was longitudinally monitored to create a behavioural measure of viral impact. When sharing with a friend, participants’ predictions of how much their UGC would be liked was positively associated with viral impact. Intent to Share was modelled as a function of image macro content features and participant responses, which was then modelled with an Agent-based computer simulation.