Definition:
A troll is a person or account that deliberately enters a digital conversation to provoke, divert the subject, generate conflict or obtain reactions from other participants.
The behaviour may appear in an internet forum, community, chat, the comments section of a blog or a social network. A critical, unpopular or strongly expressed contribution does not by itself make its author a troll: the apparent intention, pattern of behaviour and context need to be considered.
Table of contents
How to identify a troll
Trolling commonly seeks a reaction that moves the conversation away from its purpose. Indicators that may help identify a pattern include:
- Repeated provocation: The account posts messages designed to anger, ridicule or set users against one another and continues after receiving reasoned responses.
- Derailment and bad faith: It repeatedly changes the subject, misrepresents what was said or makes extreme claims without showing an interest in a useful conversation.
- Persistent conduct: It repeats the behaviour across several contributions, accounts or spaces. An anonymous or generic username is not sufficient evidence because many people legitimately use pseudonyms.
Offensive language may accompany trolling, but it is not required. Some accounts use apparently innocent questions, misleading information or out-of-context replies to provoke without clearly violating a particular rule.
Troll moderation strategies
Moderation should apply known rules consistently and distinguish disagreement, criticism, humour, abuse and manipulation. The history, context and harm should be reviewed before action is taken, without attributing intent solely from a single post.
Not responding to a provocation may prevent its amplification, but ignoring it is not always sufficient. Depending on the severity and the community rules, it may be appropriate to warn a user, apply a temporary restriction, hide or remove content, block an account, or escalate threats and harassment to the platform’s safety team.
Decisions should be based on conduct rather than on whether an opinion is held by a minority. Documenting interventions, providing an appeal route and explaining the rules helps reduce arbitrary decisions. Brand communities and social networks should also have a protocol for crises, disinformation, impersonation and risks to participants.