About TrollOrNot

What is this?

TrollOrNot is a tool that analyzes online conversations to detect patterns commonly associated with trolling behavior. It uses a combination of rule-based pattern matching and AI analysis to identify bad faith arguments, provocation, engagement bait, and other manipulation tactics.

How does it work?

The analysis happens in two stages:

  1. Rule-based scoring: The conversation is scanned for known trolling patterns across 5 categories: Bad Faith, Provocation, Engagement Bait, Strawmanning, and Derailing.
  2. AI enhancement: When available, Claude AI reviews the conversation to understand context, identify false positives, and provide nuanced analysis.

What it detects

Bad Faith

Sealioning, moving goalposts, feigning ignorance, never actually engaging with responses.

Provocation

Intentionally inflammatory language, seeking emotional reactions over discussion.

Engagement Bait

Performative responses designed to provoke reactions rather than discuss.

Strawmanning

Deliberately misrepresenting others positions to attack a weaker argument.

Derailing

Topic hijacking, whataboutism, deflection from the actual discussion.

What it does NOT do

  • Judge opinions or political views
  • Determine who is right or wrong
  • Identify genuine disagreement as trolling
  • Flag sarcasm between friends as trolling
  • Make judgments based on single messages

Privacy

Conversations you analyze are processed on our servers and may be logged for analytics (without identifying information). We do not store the full conversation text - only aggregate statistics. No personal information is collected or shared.

Disclaimer

This tool provides analysis based on linguistic patterns and is not definitive. Context matters, and the tool may produce false positives or negatives. Use your own judgment when interpreting results. This tool is for informational purposes only.