Networked harassment disproportionately targets women and historically marginalized communities (Krook, 2017; Vitak et al., 2017; Blackwell et al., 2017; Data and Society, 2018; EVAW, 2020a; EVAW, 2020b; Glitch, 2023). Gamergate was dangerous back in 2014, but in an election year with at least 70 countries headed to the polls involving two to three billion citizens (O’Caroll and Milmo, 2023), the continued lack of language and legal conceptualization around networked harassment has serious consequences for platforms’ accountability and our democracies. Our work advances the discussion by looking at an as-yet-to-be understood form of networked harassment, what we term “indirect swarming”. This is characterized by the sudden rise in the volume of posts and engagement over a short time-period, catalyzed by an amplifier, a highly-networked account, who covertly signals to their followers to harass a target. Our research objectives are: What is indirect swarming and what are the methods to discern its empirical patterns? How do we distinguish between direct and indirect swarming? Research methodology: We used a critical feminist lens in our analysis to foreground the lived experiences of the targets. We analyzed case studies of networked harassment to further expand upon Alice Marwick’s 2021 model, employing a mixed-method approach using datasets from X and Facebook, collected from the target’s standpoint (Haraway, 1988; Harding, 2004). Through descriptive statistics, quantitative and qualitative analysis, we compared these cases to illustrate the similarities and differences and establish the thresholds that had led us to propose the new categories of direct and indirect swarming. The event of foci in both case studies are the targets’ resignation announcements posted on X. Focusing on the novel risks and harms linked to indirect swarming, this paper offers a solution by sketching out what we call a “protective correlate” as a method that could help platforms mitigate the risks of indirect swarming. This method focuses on signals–similar to signals utilized in recommender systems–rather than content. For social media platforms, the protective correlate gives agency to the user to report indirect swarming while protecting platforms’ editorial rights vis-à-vis the First Amendment. This work thus has important ramifications for democracy and public discourse.