Understanding the Diffusion of YouTube Videos

In this paper we tackle several questions arising in the context of online content diffusion. In particular, we analyse the reason why some videos become viral, how popularity of a tagged video evolves over time and if there exist recurrent patterns in the dynamics of content popularity. Indeed, while the ultimate question is if it is even possible to predict the popularity dynamics of a newly published video, several interwoven factors impact the process of diffusion of online contents. In this paper we propose a framework able to put all the previous questions into a complex system science perspective. We first analyse the mechanisms that affect the popularity growth of a tagged video. We then illustrate why a multi-scale multi-level model appears the most appropriate to capture the effect of such phenomena. We finally present an open dataset of YouTube videos’ popularity, which has been released with the aim to let researchers in the field validate their findings against real-world data.