Visibility Under Constraint: TikTok and Protest Communication During Iran’s Mahsa Amini Movement

Short-video platforms have become central to protest communication, but we still know too little about how protest becomes visible when appearing on camera, which carries political risk. This paper examines how visibility is produced on TikTok during Iran’s Mahsa Amini (“Woman, Life, Freedom”) movement. I ask which multimodal tactics increase reach and engagement under repression, and what this reveals about protest communication in authoritarian settings. I argue that visibility here is shaped by a tension between state visibility, which raises the costs of identification, and platform visibility, which rewards content that is legible to format-driven norms. I develop the concept of visibility under constraint: a communicative condition in which participation must be simultaneously safer and platform-recognisable. In this context, protest publics can assemble through format-based participation, reproducible, low-exposure conventions such as symbolic gestures, modular editing, and reusable audio cues, rather than face-forward testimony or stable organisational identities. This framework links contentious politics to platformed political communication by specifying how risk and platform legibility jointly structure what becomes publicly “sayable” and “shareable.” Empirically, I analyse 145 public TikTok videos posted between 1 September and 31 October 2022, sampled from a larger pool collected via bilingual protest-related hashtags in English and Persian. Videos were coded for themes and multimodal features (visuals, audio, and hashtags). I distinguish two outcomes: reach (log views) and responsiveness (engagement rate: likes + comments + shares divided by views). Using OLS models with robust standard errors, I estimate which tactics are associated with each outcome. The results show a decoupling between what scales and what invites interaction. Symbolic protest imagery is most strongly associated with higher reach, suggesting that recognisable, low-exposure visuals function as scalable formats on TikTok. By contrast, protest audio and original/ambient sound are associated with higher engagement, indicating that collective soundscapes generate responsiveness even when visuals are distant or anonymised. Hashtag practices play a limited role: generic visibility hashtags do not predict reach, and hashtag count is not meaningfully related to views. These findings capture patterns in public content and do not directly measure recommendation mechanisms; they also cannot speak to removed, private, or suppressed content. Overall, the paper offers a framework for analysing protest communication under repression, in which platform architectures and political risk jointly structure visibility, with different multimodal cues driving diffusion versus engagement.