"One state just quietly changed a rule that affects every renter — and nobody noticed." * You know it's a trick. You know the state will be named in paragraph two, that the rule is probably small, that you will feel mildly cheated. And your hand moves anyway, because the itch is real and the itch is the product. * That is clickbait, and it is worth being precise about the word, because it has drifted into meaning "any headline that annoyed me." It doesn't mean that. * ### The actual definition * **Clickbait is a headline engineered so that the click, not the reading, is the goal — usually by withholding the key fact or by inflating it beyond what the story supports.** * Meta's own 2017 write-up on the subject split it into exactly those two moves: headlines that **withhold** information to force the click, and headlines that **exaggerate** to oversell what's there. Their example of the genre was the immortal shape "When She Looked Under Her Couch Cushions And Saw THIS…" — a sentence with a subject, a verb, and a deliberate hole where the news should be. * By that definition, a headline can be dramatic, funny, rude, opinionated, even unfair, and still not be clickbait. And a mild, professional-looking headline can be clickbait if the thing it promised isn't in the article. * ### Why the hole works: a psychology paper from 1994 * The mechanism was described before the web had headlines to sell. In 1994 the behavioural economist George Loewenstein published "The Psychology of Curiosity" in *Psychological Bulletin*, laying out what became known as **information-gap theory**: curiosity isn't a pleasant hunger for knowledge, it's the discomfort of noticing a gap between what you know and what you now realize you *could* know. Learning about the gap creates the itch. Closing it is relief, not pleasure. * So the optimal headline, commercially, is not the most interesting one. It is the one that makes you aware of a gap **narrow enough to feel closable in one click**. Too vague and you shrug. Too complete and you don't need to click at all. * Upworthy built a company on this in the early 2010s, and its co-founder Peter Koechley said the quiet part out loud, crediting the curiosity gap as the formula. It worked so well that Facebook rewrote its News Feed ranking twice to stop it. * ### What the platforms did about it * **August 2014.** Facebook announced it would down-rank click-baiting. Two details from that announcement are still the most useful things anyone has published about the genre. First, it said that when it surveyed people, **80% of the time they preferred headlines that helped them decide whether to read the article before clicking.** Second, it started using **time spent after the click** as a signal: click, read for four minutes, good; click, bounce straight back, bad. Bouncing back is the physical signature of a broken promise. * **May 2017.** Facebook went again, this time with a phrase-based classifier — reviewers tagged thousands of headlines as withholding or exaggerating, and the system learned the recurring constructions, the way a spam filter learns "limited time offer." * Two things follow from that history. One: clickbait is a measurable thing, not a matter of taste — a machine can be trained on it. Two: publishers have been playing against these filters for a decade, which is why the modern version is subtler than the couch-cushion era. It grew up. It got a tie. * ### The genre's current disguises * **The withheld subject.** "A major airline just changed its carry-on rules." Which airline is the entire news. The headline is the same sentence with the news removed. * **The undated "just."** "Just" doing the work of a date, on a story that broke on Tuesday, or in 2023. * **The question mark.** "Is your bank about to freeze your account?" A question in a headline is almost always a claim the outlet cannot stand behind as a statement. * **The vague scale word.** "Prices are soaring." "Support is collapsing." A real number would settle it, and the absence of a number in the headline usually means the number in paragraph nine is less impressive than the verb. * **The transplanted verb.** Slams, blasts, destroys, sparks fury, erupts. Somebody replied to something. Nobody destroyed anybody. * **"Here's what we know."** Honest when there genuinely is a fast-moving event. A confession, when it's used to package a story with nothing new in it. * ### The part that should actually worry you * In 2016 researchers from Columbia University's Data Science Institute and the French institute Inria published a study of roughly **2.8 million shares on Twitter**, matching shared news links against click data from the shortener Bitly. The finding that got quoted everywhere: **59% of shared links were never clicked by anyone** — often not even by the person sharing them. They also found the traffic was wildly lopsided: around **9% of shared links took about 90% of the clicks**. * Read that against everything above. If most links are shared unread, then for most people the headline *is* the article. It's not the advertisement for the news — it is the news, the whole thing, the only text that enters their head. Which means a headline that overstates by 20% doesn't overstate a story by 20%. It **replaces** it. * And remember who wrote it: almost never the reporter. Headlines are written by editors and increasingly A/B tested — two readers can be shown two different headlines on the identical article, and the one that gets clicked more wins, whether or not it's the one that's more true. * ### The two-second test * Before you click, and especially before you share: * **Does the headline contain a subject, a verb, and a number or a name?** If any of the three is missing on purpose, that's the gap, deliberately dug. * **After reading — was the withheld thing worth withholding?** This is the real test, and it's retrospective. If you finish the piece and the answer is "oh, that's it," you weren't informed. You were harvested. * **Check the URL slug.** The address bar often carries the *original* headline the outlet typed before someone sharpened it. It is the least edited part of any news page and it remembers things the page has forgotten. * **Don't share on the strength of a headline.** Fifty-nine percent, remember. That number isn't a fact about the internet. It's a fact about a specific decision that a specific person makes, dozens of times a week, and you are one of them. * ### The line worth defending * None of this means a headline has to be dull. A good one can absolutely tease — "The town that pays people to leave" makes you curious and tells you the truth at the same time. The difference isn't drama. It's whether the sentence would still look honest **after** you've read the story. * That's the whole test, and it costs nothing: read the headline, read the piece, then go back and read the headline again. Clickbait is the stuff that doesn't survive the second reading.