The Word Trolling Does Not Mean What You Think

The Fires of History Newsletter, No. 2

On August 3, 2014, a CNN anchor looked into the camera and told several million viewers that “trolls” were “terrorizing women online.” The segment was about death threats, rape threats, and coordinated harassment campaigns targeting female journalists and game developers. The threats were real. The fear was real. The suffering was real.

The word they used to describe the people responsible was wrong.

Not wrong in the sense of politically incorrect. Wrong the way calling a tsunami a “wave” is wrong – technically defensible, functionally useless, and dangerous to anyone trying to understand what they are dealing with. The people sending death threats were not trolling. They were harassing. They were threatening. In many cases, they were committing federal crimes. CNN called them trolls because by 2014 the word had been stretched so far from its original meaning that it could be wrapped around anything that happened on the internet and made someone uncomfortable.

That same month, Time published “How Trolls Are Ruining the Internet.” The cover: a close-up of a crying face reflected in a laptop screen. The article treated “trolling” as a catch-all for every species of online cruelty: death threats, doxxing, revenge pornography, racist abuse, coordinated pile-ons. All of it “trolling.” All of it performed by “trolls.” The word had become a dumpster – throw anything in, slam the lid, wheel it to the curb, move on.

What follows is the autopsy of a word.

The Fish, Not the Monster

The Oxford English Dictionary traces the verb “to troll” to at least 1377, in Langland’s Piers Plowman, where it meant to move, to roll, to wander. By the fifteenth century it had acquired its fishing sense: to troll is to drag a baited line behind a slow-moving boat, drawing the hook through the water to see what bites.

This is distinct from trawling, which drags a net and catches everything indiscriminately. The distinction matters. Trolling is selective. You choose the bait. You read the water. You wait. Something bites, you set the hook. Nothing bites, you adjust and try again.

The fishing metaphor entered internet culture in the late 1980s and early 1990s on Usenet – the decentralized network of text-based discussion groups that was the internet’s first real public forum. The phrase was “trolling for newbies.” Post something subtly wrong in a newsgroup and wait for someone to take the bait. The people who responded with outrage or earnest correction identified themselves as newcomers who had not yet learned to read the room. Experienced users recognized the troll and let it pass. The line went taut, and then: release. The newbie learned something about the culture. The regulars confirmed the culture was still functioning.

In alt.folklore.urban – one of the most active Usenet groups of the early 1990s, dedicated to debunking urban legends – regulars would post well-known false claims with straight faces. “The Great Wall of China is the only man-made structure visible from space.” “Einstein failed math.” “The word ‘gullible’ isn’t in the dictionary.” The claims were bait. The people who swooped in with breathless corrections had bitten. The group’s veterans watched from the sidelines. The system worked because it was a literacy test – not of knowledge, but of attention. Could you tell when you were being tested?

The earliest well-documented Usenet usage of “troll” in this sense appears in posts from alt.folklore.urban around 1992. The phrasing was already stable: “trolling for newbies,” “that’s a troll,” “nice troll.” The fishing metaphor was understood. The fishing metaphor was the whole point. You were not attacking anyone. You were casting a line. Catch and release.

“Don’t feed the trolls” was practical advice about catch-and-release fishing. Not a moral imperative about defeating monsters.

The Monster Ate the Metaphor

There was always the other word lurking in the background. The Norse troll. The creature under the bridge. The mountain-dwelling, shape-shifting monster from the Prose Edda, the thing that turns to stone in sunlight and eats children in the dark. The Three Billy Goats Gruff gave the anglophone world its most durable image: something ugly, something malevolent, something hiding under infrastructure and demanding payment from anyone who tries to cross.

For the first decade of internet trolling culture, the fishing metaphor held. The people who used the word knew where it came from. A troll post was a test. Responding to a troll was failing the test. The correct response was to recognize it, appreciate its craft if it had any, and move on.

Sometime around the mid-2000s, the monster ate the metaphor.

As trolling migrated from Usenet to mainstream awareness – through media coverage of 4chan, Anonymous, and increasingly visible online harassment – the bridge creature replaced the fishing rod. Not accidentally. It is much easier to tell a story about monsters than about fishermen. The monster metaphor requires nothing from the audience: the troll is bad, the victim is good, the solution is to block and report. The fishing metaphor requires nuance: the troll is testing you, the test has a purpose, the correct response depends on context. Nuance does not fit in a CNN chyron. Monsters do.

The Academic Record

Judith Donath’s 1999 paper “Identity and Deception in the Virtual Community” – published by MIT Press – is the earliest serious academic treatment of trolling as a phenomenon. Donath did something that almost nobody writing about trolling in the 2010s and 2020s would do: she went to where trolling was actually happening and watched what the people doing it thought they were doing.

What she found was not harassment. It was a game about identity and community boundaries.

Donath’s definition: trolling is “a game about identity deception, albeit one that is played without the consent of most of the players.” The troll posts a message designed to appear sincere, but the message is bait. Its purpose is not to communicate but to provoke a response that reveals something about the responder. The troll is not trying to hurt anyone. The troll is running a penetration test on the social infrastructure. Can this community detect deception? Can its members tell the difference between someone who belongs and someone who is pretending?

Community catches the troll: community passes. Community misses the troll: the troll has identified a vulnerability – a place where the immune system is weak.

Whitney Phillips, in This Is Why We Can’t Have Nice Things (MIT Press, 2015), extended the framework. Phillips, who spent years conducting ethnographic research with self-identified trolls, documented the spectrum of behavior that the word “trolling” covered before the media collapsed it into a synonym for “bad person on the internet.” She later published “The Oxygen of Amplification” (Data & Society, 2018), which identified the precise mechanism by which media coverage of trolling feeds the behavior it claims to oppose: the amplification-condemnation loop. Cover the troll, condemn the troll, drive engagement with the condemnation, attract more trolls who want the same attention. The coverage is the oxygen.

Neither Donath nor Phillips – the two scholars who have done the most rigorous empirical work on the phenomenon – define trolling as harassment. They define it as a specific mode of interaction with specific social functions. The fact that the word was later grafted onto every form of online cruelty is not a natural evolution of language. It is a category error that serves institutional convenience.

The Three-Part Damage

When you call everything “trolling,” you produce three specific harms, and all three are still compounding.

First: real harassment becomes harder to identify and address. If a Ken M comment on Yahoo Answers and a death threat in someone’s inbox are both “trolling,” then the word communicates nothing about severity, intent, or appropriate response. Law enforcement, platform moderation teams, and the targets themselves are left without vocabulary to distinguish between someone casting a fishing line and someone committing a federal crime. The word does not illuminate the problem. It obscures it.

Second: productive provocation is erased from public discourse. Socrates was a troll. Fermat was a troll. Diogenes was a troll. The Rosicrucian manifestos were a troll. The tradition of deliberate, high-effort provocation designed to expose contradictions, test assumptions, and force the audience to think – a tradition that runs unbroken from ancient Athens to the Usenet newsgroups – disappears when “troll” becomes a synonym for “monster.” You cannot defend a practice that has no name. When the word “trolling” becomes the word “harassment,” the entire mode of discourse it used to describe becomes invisible.

Third: institutions get cover to dismiss uncomfortable speech. This is the most dangerous consequence and the least discussed. When “troll” means “bad person who says bad things online,” any institution facing uncomfortable questions can dismiss the questioner as a troll. The mechanism is the same one Athens used on Socrates: do not engage the argument, attack the character of the person making it. The corruption of the word “trolling” gave every institution, every public figure, every corporation facing legitimate criticism a single-word dismissal. “They’re just trolls.” Conversation over. No engagement required. No argument addressed. The word became a shield, and the people it was shielding were not the ones being trolled.

What We Lose When the Word Dies

Imagine losing the word “satire.” Not banning it – just letting it drift until it meant any form of meanness that used humor. Jonathan Swift’s A Modest Proposal? Satire. A comedian mocking a politician? Satire. A schoolyard bully making fun of a kid’s clothes? Also satire. Once you cannot distinguish Swift from the bully, you cannot defend Swift without appearing to defend the bully. So you stop defending Swift. And then you stop reading Swift. And then the entire tradition of using humor to expose power – from Aristophanes to Voltaire to The Daily Show – becomes culturally indefensible, because the word that distinguished it from cruelty no longer functions.

That is what happened to “trolling.” Not metaphorically. Literally. A word that once described a specific, identifiable, culturally productive practice was expanded until it described everything and therefore nothing. The expansion was not organic. It was driven by media organizations that needed a single villain for a complicated story, and “troll” – with its built-in monster imagery – was too convenient to fact-check.

The fishing metaphor is recoverable. The word is recoverable. But only if someone explains what it meant before CNN got hold of it. Consider this that explanation.

This essay draws from Lurk More, coming fall 2026.

Next week: In 1614, a group of German graduate students published an anonymous pamphlet claiming the existence of a secret brotherhood. Nobody could find them. Four hundred years later, real organizations still trace their lineage to the hoax. The Rosicrucian manifestos were the original shitpost.

Source URLs