I was invited to give a talk at NetStructure, a satellite of the Networks 2021 conference. on our work studying the platform Gab using temporal interaction graphs.

Abstract

Modern online social networks foster different social bubbles, with some platforms hosting opinions that would not be acceptable on mainstream venues. Gab is one such platform, know as a hub of alt-right politics and users barred from other networks. Gab presents an interesting opportunity for social networks research because near-complete data is available from day one of its creation. In this talk I present our investigation of the evolving user interaction graph, which is viewed both at different times and timescales via a sliding window approach. Varying these timescales, I show that the Gab network is slowly growing on a monthly basis, but smaller windows reveal that this is owing to bursts of arrivals which often follow polarising political events such as suspension of far right users from Twitter or the Charlottesville rally. The network is characterised by interactions between strangers rather than reinforcing links between friends, with around half of the interactions on a daily basis being between users who have never previously interacted. Gab’s connectivity follows the diurnal cycle of the predominantly US and Europe based users, which in the off-peak hours fragments into sub-networks with absolutely no interaction between them. Finally, Gab has a small core of influential users who hold a huge portion of the network’s attention across all timescales, with a much larger pool of users who gain influence only fleetingly.