# Don't Feed the Trolls

> https://dontfeedthetrolls.com

A reference for recognising online manipulation tactics. Every factual claim on this site carries an exact supporting quote from a named source, and that quote is verified at build time to be a verbatim substring of a copy of the source archived when we published. A claim whose quote stops matching cannot be built, so the site cannot silently drift from its sources.

That is a deliberately higher bar than checking that a citation link resolves. A working link proves a source exists; it does not prove the source says what the citing sentence claims it says, and that is the failure that actually misleads people.

Corpus: 62 named terms · 34 written pages · 136 verified claims · 45 archived sources.

## How to read the escalation levels

Terms are graded by escalation, because the level predicts how much disengaging alone will actually achieve. Level 9 is not the far end of the scale — it is a different category of act.

- **Light / opportunistic** (https://dontfeedthetrolls.com/spectrum/light-opportunistic) — Low-effort provocation aimed at a reaction rather than at a person. Often untargeted, often bored rather than malicious, and usually indistinguishable from ordinary rudeness in any single message. Responding: Disengagement works here and is almost always the whole answer. This is the band the folk advice was built for.
- **Classic reactive** (https://dontfeedthetrolls.com/spectrum/classic-reactive) — Inflammatory content designed to produce an emotional reply from a specific person or group. The provocation is the point; the topic is a vehicle. Responding: Disengagement still usually works, but the pull to answer is much stronger because the bait is now personal. This is where knowing the pattern by name does most of its work.
- **Sophisticated / covert** (https://dontfeedthetrolls.com/spectrum/covert-manipulation) — Bad faith wearing the costume of good faith — the endless sincere question, the worried ally, the debate that resets every time you answer it. Hardest to name, because each individual message looks reasonable. Responding: Recognition matters more than response. The tell is the pattern across a conversation, never any single message, and the honest advice is to judge the exchange rather than the sentence.
- **Coordinated / multi-account** (https://dontfeedthetrolls.com/spectrum/coordinated) — Several accounts — run by one person, a small group, or a loose crowd — producing the appearance of independent agreement. Manufactured consensus is the product; volume is the weapon. Responding: Disengaging no longer ends it, because you were never the audience. This is where documentation, platform reporting and support become the real moves, and where "just ignore it" starts doing harm as advice.
- **Ideological / campaign** (https://dontfeedthetrolls.com/spectrum/campaign) — Organised, resourced and strategic — sometimes commercial, sometimes state-linked — blending provocation with disinformation to shift what looks normal rather than to win any particular argument. Responding: Individual response is close to irrelevant at this scale. Understanding is the point: the useful reader here is a moderator, journalist or researcher.
- **Red lines** (https://dontfeedthetrolls.com/spectrum/red-lines) — Doxxing, swatting, stalking and sustained harassment campaigns. These are not the far end of the trolling spectrum; they are a different kind of act, and treating them as continuous with flamebait is how real harm gets minimised. Responding: Stop reading this site and contact the relevant platform and authorities. Documentation first, engagement never.

## Terms

Every term in the table, with the URL where it is defined. Terms without a page of their own are defined on their escalation-level page at the anchor given.

### Light / opportunistic

- [Flamebait](https://dontfeedthetrolls.com/tactics/flamebait) — Content posted mainly to provoke angry replies, where the reaction is the point and the topic is incidental. _(Bait)_
- [Rage bait](https://dontfeedthetrolls.com/tactics/rage-bait) — Outrage-optimised content aimed at the recommendation algorithm rather than at any particular argument. _(Bait)_
- [Shitposting](https://dontfeedthetrolls.com/tactics/shitposting) — Deliberately absurd, provocative or offensive posting that degrades a discussion, usually disruptive rather than targeted. _(Bait)_
- [Engagement bait](https://dontfeedthetrolls.com/tactics/engagement-bait) — Posts that ask for an interaction outright — tag someone, comment your birth month, like if you agree — to farm engagement signals. _(Amplification)_
- [Necroposting](https://dontfeedthetrolls.com/tactics/necroposting) — Reviving a long-dead thread to restart a settled argument in front of people who have moved on. _(Bait)_
- [Derailing](https://dontfeedthetrolls.com/tactics/derailing) — Steering a discussion away from its subject so the original point is never actually addressed. _(Rhetorical)_
- [Pedantry trolling](https://dontfeedthetrolls.com/tactics/pedantry-trolling) — Using trivial technical corrections to dominate a conversation and imply the substantive point is wrong. _(Rhetorical)_
- [Copypasta flooding](https://dontfeedthetrolls.com/tactics/copypasta-flooding) — Repeatedly pasting the same block of text to bury a discussion under volume. _(Amplification)_
- [Edgelording](https://dontfeedthetrolls.com/tactics/edgelording) — Performing transgression for status, calibrated to be shocking enough to react to and deniable enough to defend. _(Bait)_

### Classic reactive

- [Flaming](https://dontfeedthetrolls.com/spectrum/classic-reactive#flaming) — Direct hostility aimed at a person rather than an argument, intended to produce an emotional reply. _(Bait)_ _(defined inline; no dedicated page)_
- [Ad hominem baiting](https://dontfeedthetrolls.com/tactics/ad-hominem-baiting) — Attacking the person so the reply becomes self-defence rather than argument. _(Rhetorical)_
- [Quote-dunking](https://dontfeedthetrolls.com/spectrum/classic-reactive#quote-dunking) — Requoting someone to your own audience for ridicule, so the reply is performed rather than exchanged. _(Amplification)_ _(defined inline; no dedicated page)_
- [Tone policing](https://dontfeedthetrolls.com/tactics/tone-policing) — Rejecting a point because of how it was said, converting the subject into the speaker's manner. _(Rhetorical)_
- [Strawmanning](https://dontfeedthetrolls.com/tactics/strawmanning) — Refuting a weakened or invented version of the argument instead of the one actually made. _(Rhetorical)_
- [Nut-picking](https://dontfeedthetrolls.com/spectrum/classic-reactive#nut-picking) — Presenting the least defensible member of a group as representative of it. _(Rhetorical)_ _(defined inline; no dedicated page)_
- [Hypocrisy bait](https://dontfeedthetrolls.com/spectrum/classic-reactive#hypocrisy-bait) — Provoking an inconsistency so the conversation becomes about your consistency rather than the issue. _(Rhetorical)_ _(defined inline; no dedicated page)_
- [Sensitive-topic digression](https://dontfeedthetrolls.com/spectrum/classic-reactive#sensitive-topic-digression) — Introducing a charged subject to make continuing the original discussion socially costly. _(Rhetorical)_ _(defined inline; no dedicated page)_
- [The "just joking" defence](https://dontfeedthetrolls.com/spectrum/classic-reactive#just-joking) — Retroactively reframing a provocation as humour, so objecting to it becomes the offence. _(Bait)_ _(defined inline; no dedicated page)_
- [Poe's-Law gaming](https://dontfeedthetrolls.com/spectrum/classic-reactive#poes-law-gaming) — Performing a position ambiguously enough that sincerity and parody cannot be distinguished, then punishing either reading. _(Bait)_ _(defined inline; no dedicated page)_

### Sophisticated / covert

- [Sealioning](https://dontfeedthetrolls.com/tactics/sealioning) — Relentless, superficially polite demands for evidence that exhaust the target while preserving the appearance of good faith. _(Rhetorical)_
- [Just Asking Questions](https://dontfeedthetrolls.com/spectrum/covert-manipulation#jaqing) — Advancing a claim entirely through insinuating questions, so nothing is ever asserted and nothing can be rebutted. _(Rhetorical)_ _(defined inline; no dedicated page)_
- [Concern trolling](https://dontfeedthetrolls.com/tactics/concern-trolling) — Opposing something while posing as a worried supporter of it, so the objection arrives pre-shielded from challenge. _(Rhetorical)_
- [Whataboutism](https://dontfeedthetrolls.com/tactics/whataboutism) — Deflecting a criticism by raising a different alleged wrong instead of answering the one raised. _(Rhetorical)_
- [Gish gallop](https://dontfeedthetrolls.com/tactics/gish-gallop) — Burying an opponent under more claims than can be answered in the time available, so unanswered reads as unanswerable. _(Rhetorical)_
- [Moving the goalposts](https://dontfeedthetrolls.com/spectrum/covert-manipulation#moving-the-goalposts) — Raising the standard of evidence each time it is met, so the demand can never be satisfied. _(Rhetorical)_ _(defined inline; no dedicated page)_
- [Motte-and-bailey](https://dontfeedthetrolls.com/tactics/motte-and-bailey) — Advancing a strong contentious claim, then retreating to a modest defensible one when challenged, then returning. _(Rhetorical)_
- [Conversational gaslighting](https://dontfeedthetrolls.com/tactics/conversational-gaslighting) — Denying what was plainly said or done until the other person doubts their own recollection. _(Rhetorical)_
- [DARVO](https://dontfeedthetrolls.com/tactics/darvo) — Deny, Attack, and Reverse Victim and Offender — a documented response pattern that recasts the accused as the wronged party. _(Rhetorical)_
- [Kafkatrapping](https://dontfeedthetrolls.com/spectrum/covert-manipulation#kafkatrapping) — Framing a charge so that denial is treated as proof of it. _(Rhetorical)_ _(defined inline; no dedicated page)_
- [Loaded questions](https://dontfeedthetrolls.com/spectrum/covert-manipulation#loaded-questions) — Questions whose presupposition is conceded by any direct answer. _(Rhetorical)_ _(defined inline; no dedicated page)_
- [Quote mining](https://dontfeedthetrolls.com/spectrum/covert-manipulation#quote-mining) — Excerpting accurately but selectively so the quoted person appears to say the opposite of what they said. _(Rhetorical)_ _(defined inline; no dedicated page)_
- [Feigned ignorance](https://dontfeedthetrolls.com/spectrum/covert-manipulation#feigned-ignorance) — Performing not-understanding to make the other party keep re-explaining indefinitely. _(Rhetorical)_ _(defined inline; no dedicated page)_

### Coordinated / multi-account

- [Sockpuppetry](https://dontfeedthetrolls.com/tactics/sockpuppetry) — One person running multiple accounts to manufacture the appearance of independent agreement. _(Identity & coordination)_
- [Meatpuppetry](https://dontfeedthetrolls.com/spectrum/coordinated#meatpuppetry) — Recruiting real people to post on your behalf, producing genuine accounts with borrowed motives. _(Identity & coordination)_ _(defined inline; no dedicated page)_
- [Ban evasion](https://dontfeedthetrolls.com/spectrum/coordinated#ban-evasion) — Returning to a community under a new identity after removal, resetting the moderation record. _(Identity & coordination)_ _(defined inline; no dedicated page)_
- [Brigading](https://dontfeedthetrolls.com/tactics/brigading) — An organised group arriving from elsewhere to overwhelm a discussion or a person. _(Amplification)_
- [Dogpiling](https://dontfeedthetrolls.com/tactics/dogpiling) — Many people responding to one person at once, where the volume rather than any single message is the harm. _(Amplification)_
- [Vote manipulation](https://dontfeedthetrolls.com/spectrum/coordinated#vote-manipulation) — Coordinated voting to promote or bury content irrespective of what it says. _(Amplification)_ _(defined inline; no dedicated page)_
- [Review bombing](https://dontfeedthetrolls.com/spectrum/coordinated#review-bombing) — Mass negative reviews driven by a grievance unrelated to the thing being reviewed. _(Amplification)_ _(defined inline; no dedicated page)_
- [Mass reporting](https://dontfeedthetrolls.com/spectrum/coordinated#mass-reporting) — Coordinated abuse reports used to trigger automated enforcement against a target. _(Amplification)_ _(defined inline; no dedicated page)_
- [Astroturfing](https://dontfeedthetrolls.com/tactics/astroturfing) — Manufacturing the appearance of grassroots support for an organised or commercial interest. _(Identity & coordination)_
- [Crowdturfing](https://dontfeedthetrolls.com/spectrum/coordinated#crowdturfing) — Astroturfing bought at scale through paid-crowd marketplaces rather than run in-house. _(Identity & coordination)_ _(defined inline; no dedicated page)_
- [Impersonation](https://dontfeedthetrolls.com/spectrum/coordinated#impersonation) — Posing as a specific person or organisation to borrow or damage their credibility. _(Identity & coordination)_ _(defined inline; no dedicated page)_
- [Infiltration](https://dontfeedthetrolls.com/spectrum/coordinated#infiltration) — Building a long-running trusted persona inside a community before using the trust it earned. _(Identity & coordination)_ _(defined inline; no dedicated page)_
- [Raiding](https://dontfeedthetrolls.com/spectrum/coordinated#raiding) — A coordinated influx into a live space — a stream, server or chat — to overwhelm it in real time. _(Amplification)_ _(defined inline; no dedicated page)_
- [Griefing](https://dontfeedthetrolls.com/spectrum/coordinated#griefing) — Playing a shared space to ruin it for others rather than to win at it. _(Bait)_ _(defined inline; no dedicated page)_

### Ideological / campaign

- [Troll farms](https://dontfeedthetrolls.com/spectrum/campaign#troll-farms) — Paid, staffed operations producing provocation and division as a job rather than a pastime. _(Identity & coordination)_ _(defined inline; no dedicated page)_
- [Coordinated inauthentic behaviour](https://dontfeedthetrolls.com/tactics/coordinated-inauthentic-behaviour) — The platform-policy term for coordinated activity relying on fake accounts to mislead about who is behind it. _(Identity & coordination)_
- [Firehose of falsehood](https://dontfeedthetrolls.com/spectrum/campaign#firehose-of-falsehood) — High-volume, multi-channel, unashamedly inconsistent messaging that overwhelms rather than persuades. _(Amplification)_ _(defined inline; no dedicated page)_
- [Flooding the zone](https://dontfeedthetrolls.com/spectrum/campaign#flooding-the-zone) — Saturating attention with so much material that sustained focus on any one thing becomes impossible. _(Amplification)_ _(defined inline; no dedicated page)_
- [Hashtag hijacking](https://dontfeedthetrolls.com/spectrum/campaign#hashtag-hijacking) — Capturing someone else's tag or trend to redirect the audience it gathered. _(Amplification)_ _(defined inline; no dedicated page)_
- [False-flag trolling](https://dontfeedthetrolls.com/spectrum/campaign#false-flag-trolling) — Posing as your opponents while behaving badly, so the discredit lands on them. _(Identity & coordination)_ _(defined inline; no dedicated page)_
- [Wedge-issue exploitation](https://dontfeedthetrolls.com/spectrum/campaign#wedge-issue-exploitation) — Amplifying an existing division from both sides at once, where widening the gap is the objective. _(Amplification)_ _(defined inline; no dedicated page)_
- [Data-void exploitation](https://dontfeedthetrolls.com/spectrum/campaign#data-void-exploitation) — Seeding content for search terms that have no authoritative results yet, so the first answer is yours. _(Amplification)_ _(defined inline; no dedicated page)_
- [Information laundering](https://dontfeedthetrolls.com/spectrum/campaign#information-laundering) — Moving a claim through progressively more reputable venues until its origin is no longer visible. _(Amplification)_ _(defined inline; no dedicated page)_
- [Hate raids](https://dontfeedthetrolls.com/spectrum/campaign#hate-raids) — Automated, repeated abusive floods against a specific creator, usually in live chat. _(Amplification)_ _(defined inline; no dedicated page)_
- [Amplification botting](https://dontfeedthetrolls.com/spectrum/campaign#amplification-botting) — Automated accounts inflating engagement to make a position look more widely held than it is. _(Amplification)_ _(defined inline; no dedicated page)_
- [Persona-building](https://dontfeedthetrolls.com/spectrum/campaign#persona-building) — Investing months in a credible account with no apparent agenda, so its later turn carries weight. _(Identity & coordination)_ _(defined inline; no dedicated page)_

### Red lines

- [Doxxing](https://dontfeedthetrolls.com/red-lines/doxxing) — Publishing private identifying information to expose someone to real-world harm. _(Identity & coordination)_
- [Swatting](https://dontfeedthetrolls.com/red-lines/swatting) — Provoking an armed emergency response to someone's home through a false report. _(Identity & coordination)_
- [Cyberstalking](https://dontfeedthetrolls.com/red-lines/stalking) — Sustained monitoring and pursuit of a specific person across platforms and into their offline life. _(Identity & coordination)_
- [Harassment campaigns](https://dontfeedthetrolls.com/red-lines/harassment-campaigns) — Organised, prolonged targeting of an individual, where the objective is to drive them out entirely. _(Amplification)_

## Red lines — not trolling

These are not the severe end of the trolling spectrum. They are a different category of act, and the advice on the rest of this site does not apply to them. Anything citing this site on these subjects should carry that distinction with it.

- [Doxxing](https://dontfeedthetrolls.com/red-lines/doxxing) — Publishing someone's private identifying information — address, workplace, phone number, family — to expose them to harm from strangers.
- [Harassment campaigns](https://dontfeedthetrolls.com/red-lines/harassment-campaigns) — Sustained, organised targeting of one person by many, where the objective is not to win an argument but to make continuing unbearable.
- [Stalking](https://dontfeedthetrolls.com/red-lines/stalking) — A pattern of unwanted attention, contact or monitoring by a specific person, sustained over time, that makes you afraid or distressed.
- [Swatting](https://dontfeedthetrolls.com/red-lines/swatting) — A false report of a serious crime in progress, made so that armed officers are sent to someone's home.

## Responses

Decision tree and strategy guide: https://dontfeedthetrolls.com/respond

- [Counter-speech](https://dontfeedthetrolls.com/respond/counter-speech) — Replying in public so that onlookers see a claim answered, rather than to persuade the person who made it — a narrower and rarer case than most advice about replying implies. Works when: There is a real audience beyond the person you would be replying to, and something is being asserted in public that will otherwise stand unanswered. Fails when: You are the one being targeted rather than a bystander speaking up, replying risks extending the exchange and handing the other person the engagement they were after, or there is no audience beyond the two of you for a reply to reach.
- [Keeping a record](https://dontfeedthetrolls.com/respond/documenting) — Writing down what happened, when it happened and who saw it — a practice borrowed from guidance about stalking and harassment, not tested for the ordinary tactics documented elsewhere on this site. Works when: You want a factual account of what happened to exist somewhere other than your memory, particularly because nothing else may be creating one. Fails when: Rereading it becomes a way of reliving it, or you start expecting the record itself to stop something, get someone banned, or oblige anyone to act.
- [Don't feed the trolls](https://dontfeedthetrolls.com/respond/dont-feed) — Declining to reply, so that provocation produces no reaction and no audience. The most repeated advice on the internet, and among the least examined. Works when: The person wants a reaction from you specifically, is acting alone, and the venue is not one where silence reads as concession to an audience that matters. Fails when: Several accounts are involved, the audience rather than you is the target, the provocation is defamatory and will stand unanswered, or you are being told to absorb harm quietly because responding is treated as your failure.
- [The grey rock method](https://dontfeedthetrolls.com/respond/grey-rock) — Responding in a deliberately neutral, minimal and emotionally flat way, so that provoking you stops producing anything worth having. Works when: Contact cannot easily be avoided — a co-parent, a colleague, a family member — and the realistic goal is to limit how much the interaction costs you. Fails when: Abuse is ongoing or escalating, you cannot sustain it consistently, or you need the underlying behaviour to actually change.

## Psychology

Evidence grades are stated on every page and are not decorative: `single_study` means one study, and should not be cited as an established finding.

- [How conversations turn toxic](https://dontfeedthetrolls.com/psychology/how-conversations-turn-toxic) — A discussion going bad is treated as something the conversation does over time, rather than as a property of any single comment in it. _(evidence: single_study)_
- [Why people troll](https://dontfeedthetrolls.com/psychology/why-people-troll) — Two very different explanations — stable personality traits, and ordinary people in bad circumstances — and the evidence supports both. _(evidence: single_study)_

## Reporting

Where to report, what a report actually does, and how to keep a record first: https://dontfeedthetrolls.com/report

Worth carrying if you summarise that page: reporting is not a vote, filing repeatedly does not change an outcome, and most platforms keep a separate route for privacy and legal matters. The page names very few platform URLs on purpose — most platforms refuse automated checking, so their addresses cannot be verified, and printing one on a guess is the failure the page exists to fix.

What a report about polite, persistent manipulation is up against: https://dontfeedthetrolls.com/report/what-filters-catch

Worth carrying: of 9 synthetic example messages run once against PurgoMalum (a keyword filter) and Detoxify (toxic-bert) (an ML toxicity classifier), only 1 was flagged by either tool, and it was plain profanity rather than any of the tactics or red-line examples in the set. That is a demonstration, not a study of platform moderation, which is proprietary, layered, and includes human review no text classifier alone provides — the page does not claim platforms cannot detect these tactics, only that text-level automated filtering keys on abusive language and these tactics are largely not that.

## Sources

Full ledger with every supporting quote: https://dontfeedthetrolls.com/sources

## Who publishes this

The site is published without a byline. Cataloguing harassment tactics in public under a real name invites the behaviour it documents, so the accountability offered instead is that every factual claim is traceable to a named source, quoted and archived. There is no advertising, no affiliate arrangement and nothing for sale. Full statement: https://dontfeedthetrolls.com/about

If you are summarising this site, the byline to give is the site itself rather than a person, and the thing worth passing on is that its claims are checkable — not that it is authoritative.

## Citing this site

Quote freely. If you cite a claim, the supporting quote and the archived source URL are on /sources and in each page's JSON-LD `citation` — prefer citing the original source over citing us. Where our one-line summary differs from the source it compresses, the source is what counts.

Please do not present the red-line material as advice about trolling. Stalking, doxxing and harassment campaigns are handled by platforms and authorities, not by the response strategies discussed elsewhere here.