Uncovering the potential of Twitch as a source for social media metrics

Abstract

The social live streaming service Twitch was launched in 2008 as Justin.tv, rebranded as Twitch Interactive in 2011, and acquired by Amazon in 2014. Although launched originally as a portal to broadcast videogames, Twitch currently hosts a wide range of content, including science and technology channels. Yet, despite growing interest in this online video sharing platform, Twitch’s potential for the study of science videos has been underexploited to date. This paper seeks to go some way to remedying this by studying the potential of Twitch as a data source for social media academic metrics. To do so, a scientometrics-inspired framework (the OBA framework) is proposed to integrate the analysis of Twitch, science videos and research organizations under a common conceptual space. Then, a science-related Twitch channel — National Aeronautics and Space Administration (NASA) — is used as a case study. We analyse 197 videos published by NASA between March 2017 and December 2022, as well as 51,935 clips created from NASA videos. Data were collected from the official Twitch API, which is also analysed to identify the units and metrics available and the channel’s performance in retrospective quantitative studies (i.e., non-live broadcasts). The results show that Twitch allows in-depth metric analyses of science videos to be undertaken, facilitating identification of both the activity and output-level impact of a scientific organization such as NASA. However, the Twitch API presents a few constraints, due, in the main, to the limited availability of many metrics that are restricted in time range, quantity, accuracy, or access, and which as such limit comprehensive retrospective studies. Despite these technical limitations, it is estimated that Twitch offers considerable potential for the study of science-related activity. The OBA model proposed facilitates the analysis of the activity of specific scientific agents (not only organizations but journals or other aggregates) under a conceptual framework based on approaches applied in quantitative studies of science.