Intellectual dark web, alt-lite and alt-right: Are they really that different? a multi-perspective analysis of the textual content produced by contrarians

Contrarian groups, notably Intellectual Dark Web, Alt-lite, and Alt-right, are present across the Web, ranging from fringe websites to mainstream social media. Such massive presence raises major concerns as contrarians often engage in the spread of conspiracy theories and hate speech toward particular groups of people. Historically, there is a general sense that these groups exhibit different degrees of extremism, with Alt-right standing out as the most extremist one. In particular, prior work often takes participation in Alt-right communities as a proxy for radicalization. Yet, to which extent are these groups really different? While most previous analyses have focused on a content consumption (i.e., viewer) standpoint, no prior work analyzed these groups (i.e., contrarians) from a content production perspective. Are there significant differences in the content produced by them? Toward tackling this question, we here analyze the textual data associated with videos shared by the three aforementioned groups. Specifically, we analyze 14 years of content produced by contrarians on YouTube with data from 355,000 videos. Firstly, we assess the degree of toxicity of the content created by each contrarian group, comparing them to one another and, for control purposes, against traditional media content. The results show that all contrarian groups have a more skewed toxicity distribution than traditional media. Yet, all three groups exhibit very similar textual toxicity properties. Further analyses based on psycholinguistic properties and semantic (text) classification reinforce the observation that indeed there is great similarity among the content created by all three contrarian groups. These results suggest that, despite the different definitions, the three contrarian groups are indeed much more similar, in terms of the content produced and shared by them, than the general wisdom (and literature) seems to suggest. Moreover, we also identify a significant temporal increase in content toxicity in all three groups, corroborating prior observations regarding the escalation in the harmfulness of online speech over the years.