Untangling the furball: A practice mapping approach to the analysis of multimodal interactions in social networks

Summary

This article introduces practice mapping, a methodological framework for analyzing multimodal social media interactions through vector embeddings of user actions. The authors argue that conventional social network analysis and its visualizations — frequently degenerating into illegible “hairball” or “furball” diagrams — are poorly suited to platforms where users engage through many distinct interaction modes (likes, replies, shares, mentions, follows, etc.). By embedding network actions as vectors, practice mapping locates users in a shared analytical space according to the similarity of their interactional repertoires, allowing multiple modes to be integrated into a single, interpretable representation. The piece is primarily methodological, outlining the framework and its conceptual motivation rather than reporting empirical findings.

Key Contributions

• Proposes practice mapping as a novel analytical approach for social media research.
• Reframes the unit of analysis from ties between accounts to similarities between users’ interactional practices.
• Provides a way to integrate multimodal interaction data (across action types) into a unified visual analysis.
• Offers a methodological alternative to conventional network visualizations whose density renders them analytically opaque.

Methods

• Conceptual/methodological introduction of the practice mapping framework.
• Use of vector embeddings to represent the actions and interactions performed by users on social media.
• Projection of users into a shared embedding space where proximity reflects commonalities in practice and distance reflects divergence.
• Comparison with, and critique of, standard network-graph visualization techniques.

Findings

• The paper is methodological; no specific empirical findings are reported in the available material.
• The principal argumentative result is the claim that practice mapping yields more legible and analytically tractable representations of multimodal interaction than conventional network graphs.

Connections

This paper sits alongside other methodological interventions seeking to make large-scale platform data analytically tractable, including work on platform data access and measurement infrastructures such as Rieder2025-ju and Ulloa2024-jm, and engages debates about how researchers conceptualize online “behavior” and engagement, as in Bak-Coleman2026-mk and Freelon2024-sc. Its emphasis on multimodal interaction also resonates with research that moves beyond single-signal analyses of social media activity, such as Balluff2026-if.