In the comments is where the magic happens: visual and argumentative strategies of the far right on social media

Abstract

This article presents a methodological model for analyzing the role of comments on multimodal social media posts. Posts with ambiguous meaning and strong emotional appeal — consisting of images or videos without captions or accompanied by vague captions — often conceal messages that could compromise the author, disinformation, hate speech, xenophobic content, and attacks on opponents. It is, however, in the comments section that the implicit meanings of the original post emerge, engagement intensifies, and processes of radicalization consolidate.

The methodological proposal brings together argumentation theory, data science, and digital sociology in a hybrid model that integrates the Social Argumentative Analysis Framework (SAAF) and the Model Context Protocol (MCP). The MCP provides a standardized and auditable infrastructure for integrating heterogeneous data flows, while SAAF operationalizes the detection of argumentation schemes, fallacies, emotive language, and info-arguments, comparing them with contextual libraries that include scientific, religious, partisan, ideological, and other repertoires deemed relevant for analysis.

In its broader version, this study’s empirical scope covers Brazil and Portugal, focusing on posts by far-right party leaders on X, Instagram, and TikTok. The model demonstrates how ambiguous audiovisual content shifts the interpretative burden onto comments, a space where coordinated or automated profiles seed preferred readings reinforced by interpretative and reception communities.

The contributions are centered on four key points: (i) valuing social media comments as the core unit of multimodal analysis; (ii) integrating argumentation schemes and repertoires of legitimation (libraries) for comparative analysis; (iii) combining methodologies through MCP/SAAF integration; and (iv) introducing the concept of meta-identity, which describes the dynamics involved in interactions between users, digital platforms, and their processes of classifying digital identities through policies, parameters, and algorithms.

Portuguese version: https://doi.org/10.1590/SciELOPreprints.13311