Somewhere in the past few years, millions of people independently developed the same small tic. They open Google, type what they want, and then add one more word: reddit. "Best headphones reddit." "Is Lisbon worth it reddit." "Why does my sourdough hate me reddit." Nobody taught anyone to do this. There was no campaign, no onboarding flow. The habit emerged the way a footpath emerges across a lawn — the collective discovery that the paved route no longer goes anywhere you want to be.
Look closer at what the tic actually means, though, because it's stranger than it first appears. These searchers aren't looking for Reddit, exactly. They're looking for a person. The first page of results has become a hall of mirrors — listicles reverse-engineered from other listicles, buying guides assembled by affiliate commission, how-tos padded for an algorithm that rewards length over knowledge. Appending "reddit" is a filter for the unoptimized: threads written by strangers with no funnel, no newsletter, no call to action. It may be the most telling feature request in the history of the web, filed one keystroke at a time: show me someone who isn't being paid to talk to me.
The interesting thing about the modern internet isn't that it's annoying. It's that every single thing that makes it annoying is, by the standards of the people who built it, working. The recipe site that buries the ingredients beneath 1,800 words about a grandmother in Tuscany — dwell time, ad impressions. The autoplay video — view count. The pop-up demanding your email before you've read a word — capture rate. The infinite scroll — time on site. None of these are failures of taste or accidents of design. Each irritation has a number behind it, and the number went up. The dashboards are all green. Everyone hates it here. Both things are true at once, and holding them together explains the last fifteen years of the web.
Economists have a name for the mechanism: Goodhart's law — once a measure becomes a target, it stops measuring the thing you cared about. "Engagement" was always a proxy, a stand-in for the hope that people were getting something out of all this. Then the industry optimized the stand-in. It's the digital version of judging a restaurant purely on table turnover: feed them fast, move them along, and one day notice that nobody lingers, the room has no buzz, and the regulars are gone — but look how many covers we did.
Measurement was never neutral, either. It selects. The things software can count — clicks, seconds, shares, scroll depth — multiplied until they colonized everything. The things it can't count — pleasure, trust, the satisfaction of getting pleasantly lost — were classified, in effect, as waste. The web didn't lose its joy by accident. Joy was an externality, and "externality" is the spreadsheet's word for everything it can't see.
What got optimized out first was slack. The old web was mostly slack: webrings, fan shrines, GeoCities neighborhoods, a retired man's exhaustively maintained page about lighthouses. StumbleUpon had a button whose entire function was serendipity — press it and be taken somewhere gloriously unnecessary. None of it converted, ranked or scaled, which was precisely the charm, though nobody would have used that word. It was the internet's front porch: space that existed for its own sake. Optimization reads slack the way a cost-cutter reads an empty park bench — dead square footage. But slack is where pleasure lives, in cities, in games, in afternoons. A park is a page with no call to action.
What went next was generosity. The early web ran on unmeasured gifts: the forum stranger who wrote nine hundred words diagnosing your printer problem in 2006, the fans who built episode guides for free, the bloggers posting for twelve readers because they couldn't help themselves. Then the gifts got counted, and once counted they got professionalized — generosity became content, and content became content marketing. Nonna's Tuscany story isn't a gift. It's dwell time with mozzarella in it.
And here is the part that should be carved above the door of every analytics company: the numbers were often wrong. In the mid-2010s, Facebook's video statistics convinced an entire generation of publishers to "pivot to video." Newsrooms laid off real writers to chase view counts. Then Facebook acknowledged it had been overestimating how long people watched; advertisers sued, alleging the inflation was far worse than admitted. The case settled. The writers did not get their jobs back. The internet tore out part of its own load-bearing culture to please arithmetic that wasn't even accurate. We didn't sell the web for good data. We sold it for bad data, faithfully reported.
None of this is an argument that measurement itself is wicked — analytics killed genuine scourges, and the pre-metrics web had its own cruelties and cliques. Nor is the machine wholly to blame, because the metrics are, in one sense, honest: they're mirrors of what we actually do at 1 a.m., aggregated and fed back to us as design. The infinite scroll doesn't ask what you want; it asks what you'll do, and those are different questions with different answers. Aza Raskin, who invented infinite scroll in 2006, has said he regrets it — by his own estimate the thing consumes the equivalent of entire human lifetimes of attention per day, which is a grim legacy, delivered in roughly the tone of a man apologizing for leaving the stove on in several million homes. The trap isn't that the numbers lie about us. It's that they tell the truth about our worst five percent and then build the whole house around it.
The pleasure, meanwhile, didn't die. It fled. It lives in group chats, invite-only Discords, newsletters with two hundred subscribers, RSS readers, bookmark folders, forums that still make you wait a week for an account. People are rebuilding the unoptimized web in its margins — in spaces that are, not coincidentally, hard to search, hard to rank, and hard to sell advertising against. There is now a fairly reliable inverse relationship between how good a corner of the internet is and how legible it is to a dashboard. The moment a space gets discovered and measured, the surveyors arrive with their KPIs, and the fun quietly packs a bag.
Which reframes that little search tic. When someone types "reddit" after a query, they're asking to be routed around the entire apparatus — past the funnels and the rankings and the content — to a place where people speak without being metered. The answer they find might be wrong. But it will be an answer, not an asset. Every one of those searches is a small vote for the unmeasured web, and no dashboard will ever register it, because the one thing the metrics can't count is the thing people were actually looking for. The best parts of the internet are now, almost by definition, the parts that don't count. So if you want to know whether a corner of the web is still alive, don't check its engagement. Check whether anyone is measuring it at all.