A researcher has made a significant breakthrough in understanding online arguments, developing a framework called IDDS (Identity-Driven Discourse Systems) to model how these conflicts escalate. By intentionally sparking controversy on Reddit, the researcher was able to formalize the patterns of escalation into a predictable state machine. The key finding is that escalation is not random, but rather driven by identity layer activation in a predictable sequence of states: Neutral, Disagreement, Identity Activation, Personalization, Ad Hominem, and Dogpile.

The latest version of the framework, IDDS 2.1, introduces a new modifier called D_flag, which indicates whether identity activation accelerates escalation only when disagreement is already present. This means that sharing one’s identity in a friendly thread behaves differently from the same disclosure in an adversarial thread.

The researcher also identified new patterns in online escalation, including Moral Protective Framing (MPF), where individuals use ethical justifications to escalate conflicts, and Adversarial Seeding, where threads are born escalated from the start. Additionally, the researcher found that blocking or muting an opponent only terminates the local thread, but not the conflict itself, and that transient dogpile groups can form and dissipate without fully resetting.

The IDDS framework has been validated across multiple platforms, including Reddit, Threads, and WhatsApp, in both English and Portuguese. The researcher plans to build a Playwright scraper and ML classifier to further develop the framework.

A paper detailing the research is available on GitHub.

Photo by Daniel Absi on Pexels
Photos provided by Pexels