TubeStats and TokStats: Research Tools for Random Samples of YouTube and TikTok

YouTube and TikTok are two of the most popular digital communications platforms in the world, playing a disproportionately large role in global communications infrastructure in general and the consumption and dissemination of information in particular. As neither platform provides adequate mechanisms to produce representative samples of the content they host, researchers largely depend on opportunistic samples of popular, recommended, or otherwise known content. In this article, we present two dashboard-based tools, TubeStats and TokStats, built upon our recent research into random sampling techniques for each platform. These tools provide platform-wide statistics such as the number of hosted videos, view count distributions, linguistic distributions, and growth over time, which researchers can use to quantify and contextualize their research. We explain the architecture and sampling pipeline of each tool as well as the unique technical and methodological affordances and constraints involved with each. We document how these related techniques and tools have been applied by our lab, other scholars, and journalists to contextualize non-representative samples, compare platform use across languages and regions, and examine quotidian uses of the platforms that attention-optimized samples may obscure, as well as the broader range of methodological possibilities that representative sampling opens for platform research. Not to be taken for granted, we also explain the many challenges we face in developing and maintaining such tools, with implications for the practical development of open research infrastructures.