Tweets Are Not Created Equal: investigating Twitter's client ecosystem

This article offers an investigation into the developer ecosystem of platforms drawing on the specific case of Twitter and explores how third-party clients enable different “ways of being” on Twitter. It suggests that researchers need to consider digital data as traces of distributed accomplishments between platforms, users, interfaces, and developers. The argument follows three main steps: We discuss how Twitter’s bounded openness enables and structures distributed data production through grammatization of action. We then suggest ways to explore and qualify sources by drawing on a weeklong data set of nearly 32 million tweets, retrieved from Twitter’s 1% random sample. We explore how clients show considerable differences in tweet characteristics and degrees of automation, and outline methodological steps to deploy the source variable to further investigate the heterogeneous practices common metrics risk flattening into singular counts. We conclude by returning to the question about the measures of the medium, suggesting how they might be revisited in the context of increasingly distributed platform ecosystems, and how platform data challenge key ideas of digital methods research