ABSTRACT This article draws on ethnographic and historical research at two Australian libraries to explore how the platformization of the web has altered the content, character, and potential future utility of web archives. I argue that, in attempting to collect social media, these libraries face a double bind: while web crawling undertaken at the National Library of Australia allows for immediate but often incomplete or inconsistent access, the API-based approach taken by the State Library of New South Wales constrains both the data collected and how it can be made accessible due to a shifting set of rules that are established and enforced by platforms. By examining the constraints of current strategies to collect, preserve, and make available social media content, I illustrate how changes to platform design and policies significantly influence what is included in web archives and how they are made available. As the ruptures and inconsistencies of collections of social media in web archives are often opaque to both creators of web archives and those using them, I argue that web archives can be read ‘along the archival grain’ for evidence of the platformization of the web. This approach, which draws on anthropologist Ann Stoler’s critical readings on the form and placement of colonial archives (rather than just their contents), allows an assessment of how the gaps, silences, densities, and distributions of web archives are shaped by the shifting power dynamics between different actors involved in the production, circulation, distribution, and use of information on the web.