Going /k/ommando: Gun Culture in Fringe Online Communities

Fatemeh Tahmasbi, Aakarsha Chug, Barry Bradlyn, Jeremy BlackburnIcwsm 2025 · DOI

TL;DR

This paper dives into the "gun culture" on 4chan's /k/ board, a fringe online community, using NLP to analyze over 4 million posts. It reveals a significant focus on gun laws and a concerning prevalence of "gun fetishism," offering crucial insights for content moderation and de-radicalization efforts.

What Problem Does This Paper Address?

The increasing frequency of mass shootings in the US has brought gun control to the forefront, but the cultural and psychological significance of guns, often termed "gun culture," remains a less understood factor. While social media's role in shaping culture is known, the intersection of gun culture and fringe online communities like 4chan, especially concerning the online radicalization of some shooters, is unclear. This research aims to fill that gap by exploring gun culture on 4chan's /k/ board, a space where users can anonymously discuss firearms, and contextualizing these discussions within existing theoretical frameworks of gun culture.

Methods

The study employs a multi-stage NLP pipeline:

  1. High-Level Topic Modeling: BERTopic is used to cluster posts and identify broad topics within the 4.4 million posts from 4chan's /k/ board.
  2. Gun Community Detection: A community detection algorithm is applied to partition topics, isolating those specifically related to guns.
  3. Low-Level Topic Modeling: Top2Vec is then used on the gun-related topics to generate more detailed and interpretable topic representations.
  4. Manual Annotation: A gun topic dictionary is created, and three authors manually annotate the derived topics, achieving substantial inter-annotator agreement (Fleiss' Kappa score of 0.75).
  5. Semantic Similarity Analysis: Word2Vec is used to explore gun attachment and identify semantic relationships between words, revealing insights into emotional motivations and the context of gun ownership discussions.
  6. Semantic Similarity Search: A fine-tuned Sentence BERT model is employed to identify posts that closely resemble baseline quotes demonstrating gun fetishism, allowing for the detection of this phenomenon.

Datasets

The primary dataset consists of over 4.4 million posts collected from 4chan's /k/ (weapons) board throughout 2022. This dataset is noted for its potentially offensive and disturbing content. No new benchmarks were introduced; rather, existing literature and theoretical frameworks were used to contextualize the findings.

Key Insights and Ideas

  • Law is a Major Focus: Surprisingly, over 17% of gun-related conversations on /k/ revolve around legal matters, indicating a strong engagement with the legal aspects of gun ownership.
  • Gun Fetishism is Prevalent: The study identifies significant signs of "gun fetishism," characterized by an extreme or obsessive connection to firearms that extends beyond practical use. This manifests in various ways, including viewing guns as magical or religious objects, and associating them with sexual desires.
  • Cultural Narratives Drive Attachment: Gun attachment is deeply intertwined with cultural narratives from both "Gun Culture 1.0" (colonization, sports shooting, masculinity) and "Gun Culture 2.0" (self-defense, Second Amendment rights). Users express attachment through lenses of masculinity, sentimentality, moral empowerment, and patriotism.
  • Religious and Patriotic Undertones: Discussions often blend gun ownership with religious beliefs and patriotism, with some users framing gun rights as a "God-given" right or a crucial aspect of being a "good and patriotic American."
  • Fringe Platform Dynamics: The anonymous nature of 4chan allows for unfiltered discussions, providing a unique window into the underlying beliefs and emotions driving gun ownership, distinct from content shaped by paid influencers or specific events.

Key Results

  • Topic Distribution: The analysis revealed that "Manufacturers" was the most prevalent label among gun-related discussions, followed by "Collection," "Marksmanship," and "Hunting" (Gun Culture 1.0), and "Law," "Self-defense," and "Religion" (Gun Culture 2.0).
  • Gun Attachment Manifestations: Word2Vec analysis identified keywords related to gun attachment such as "safe_queen" (slang for collectible firearms), "sentiment_valu," and "heirloom," indicating emotional significance beyond practical worth.
  • Fetishism Indicators: The study found 1,346 posts associating guns with fetishism or containing confessions. Semantic similarity analysis also revealed posts that clearly demonstrate religious fetishism (e.g., blessing rifles with holy water) and patriotic fetishism (e.g., viewing guns as essential for resisting "tyrannical takeovers").

Limitations and Future Work

  • Generalizability: The study's focus on a single, fringe online community (4chan's /k/) means findings may not generalize to other gun-oriented online spaces.
  • NLP Nuances: Quantitative NLP techniques might overlook subtleties like sarcasm or irony in discussions.
  • Future Directions: The authors suggest that de-radicalization efforts and content moderation strategies could be improved by specifically addressing signs of gun fetishism. Their research can also assist in creating support programs to educate individuals at risk of adopting extreme views related to gun culture. The methodologies could be adapted for studying other platforms to understand how extremist ideologies propagate across different online spaces.

Summary updated

BibTeX
@inproceedings{tahmasbi2025going,
  author = {Fatemeh Tahmasbi and Aakarsha Chug and Barry Bradlyn and Jeremy Blackburn},
  title = {{Going /k/ommando: Gun Culture in Fringe Online Communities}},
  booktitle = {Icwsm},
  year = {2025},
  doi = {10.1609/icwsm.v19i1.35905},
}
Figure