11 min read essay economics philosophy

The Matchmaker and the Meter

Hinge says it is designed to be deleted; its parent company is paid while you stay. On the meter dating apps run on — match scarcity as inventory, the market for lemons run on people — and the matchmakers people actually loved: the shadchan, the paseo, the dance hall, whose intermediaries answered for the outcome rather than the search. What people are running to instead, offline and on; why the good forms charge effort rather than money; and the CI rule that keeps xNet from ever selling an introduction.

crs48Claude crs48 Claude History

Hinge’s slogan is four words long: designed to be deleted. It may be the only advertising in the world that promises you a funeral. Take it seriously for a moment, because it is a strange thing for a subscription business to say. A subscription earns nothing from an ending. It earns from the middle, from the long undecided stretch where you are still looking, still hoping, still paying. A dating app that worked the way its slogan claims would be a poor business, and the company appears to know it: Match Group’s own chief executive has told staff that Tinder and Hinge users feel the company is too driven by metrics. The slogan and the memo are both sincere. Holding them together is what this essay is for.

Some numbers first, so the argument has weight to push against. Match Group owns more than forty dating brands, Tinder, Hinge, OkCupid and Plenty of Fish among them, and about half the dating activity of young Americans runs through its servers. Tinder Plus launched at $9.99 a month and now costs $24.99. Bumble’s paid tier has tripled since 2016. The League, at the luxury end, will take up to $2,499.99 a month. Things that used to be free now sit behind those prices: seeing who liked you, filtering by distance, more than a handful of swipes a day. And on Valentine’s Day 2024, six users filed a federal class action alleging what the pricing already implies. The apps, the complaint says, are gamified on purpose, tuned like slot machines to keep you searching for rewards the operator keeps just out of reach. Match called the suit ridiculous and without merit. Perhaps. But the design it describes needs no conspiracy. It is simply what the incentives ask for.

The meter

Here is the mechanism with no villain in it. A dating company is paid for access to matches, so its inventory is the distance between you and the person you are trying to reach. When a match succeeds, the firm loses two subscribers; a near miss keeps both, and the last inch can be sold. Nobody in the building has to want this; it falls out of a quarterly target the way water falls downhill. Scarcity is stock, frustration is turnover, and the perfect customer is the one who almost finds someone, forever.

Economists have a name for what happens next. In a market where buyers cannot tell good from bad, the good options leave first and the market sours from the top; George Akerlof called it the market for lemons. Dating apps run the experiment on people. Put the likeliest matches behind a paywall and the people most worth meeting are the first to give up and go. Each departure lowers the odds for everyone still swiping, which sends the next tier out of the door behind them. The paywall does not just tax the market; it curdles it. Nearly four in five Gen Z daters report dating-app fatigue, and fatigue is what curdling feels like from the inside.

The matchmaker

So how did people meet before, in the ways they remember warmly? Almost never unaided; that part of the nostalgia is false. There was nearly always an intermediary. In the Jewish villages of eastern Europe it was the shadchan, a professional whose trade was knowing every family’s character and circumstances and proposing marriages between them. Elsewhere it was an aunt, a church social, a barn dance, or the slow evening walk that Mediterranean towns still call the paseo. Even the personals column of a local newspaper was an intermediary of a kind: it took a flat fee for the advert and had no further stake in your search.

Look at what those forms shared, because it was not innocence. First, recurrence: you would see these people again next week, so patience was possible and reputations mattered. Second, the format absorbed the no. In a dance where everyone rotates, not being chosen is a fact of the room rather than a verdict on your face. Third, contact cost nothing; where admission existed it was flat and cheap. Fourth, and this is the load-bearing one, the intermediary was paid for the outcome, not the search. A shadchan lived on reputation in a village that would judge the marriage for decades. The aunt answered to both families at every holiday for the rest of her life. Bad matches cost the matchmaker, which is precisely the liability no swipe app carries.

Be honest about the price of all this. These were also the mechanisms of small worlds that could suffocate: chaperones, narrow pools, communities that punished difference and made leaving expensive. Recurrence is the warm name for a room you cannot easily exit. The point is not that the village was better, only that its machinery was aligned — and the machinery can be taken without the walls.

The historian Beth Bailey dated the turn precisely. Until the 1920s, American courtship ran on the calling system: a man visited a woman in her family’s parlour, at her family’s invitation, and the whole apparatus cost nothing and answered to her side. Within a generation, courtship moved out of the parlour and into restaurants, theatres and dance halls, and access to those rooms had to be bought. Money, men’s money, became the medium of courtship for the first time. Everything since has been elaboration. The apps did not invent the meter. They shrank it, put it in your pocket, and pointed it at the introduction itself.

Disintermediating your friends

The scale of the change has been measured. Stanford’s long-running survey of how American couples meet found that 39 per cent of heterosexual couples met online by 2017, up from 22 per cent in 2009. Online passed friends as the leading way around 2013, and for same-sex couples the figure was already about 65 per cent. The researchers’ paper carries the honest title: Disintermediating Your Friends. A friend who introduces you is infrastructure: they know you both, filter better than any questionnaire, stake a little of their own standing on the introduction, and do it for free. Worse: for joy. The subscription sells you back their labour, minus the knowledge, minus the stake.

Where people are going instead

Watch what people run to when they leave, because it is a map of what was missing. Attendance at in-person singles events grew by 42 per cent in a single year. Timeleft seats strangers around dinner tables every Wednesday in three hundred cities, and the unit is the table, not the pair, so nobody at it can be rejected. Run clubs are the paseo with cardio: recurring, free, no declared intent required. Board-game cafés hand you a conversation so you do not have to build one from nothing. Speed dating is back, and it is the dance hall’s rotation with name tags. There is even a $25 ring, the Pear ring, whose only feature is announcing that you are single and approachable — the ambient signal a promenade used to provide for free. Notice the melancholy detail under the revival, though: nearly all of it is ticketed. The sociologist Ray Oldenburg called the free rooms of a town its third places, and most of them have closed. Their reconstruction charges at the door.

Online, the interesting experiments are the ones almost nobody covers. Some people now write date-me docs: long, earnest, self-authored pages that circulate link by link and gather in a small volunteer directory. Lex, a queer personals app descended from the back-page adverts of a 1980s magazine, puts the text first and the selfies second. The Marriage Pact asks students fifty questions about values, runs a Nobel-winning stable-matching algorithm over the answers, and returns one match, once a year, free, with no photographs. More than six hundred thousand students across a hundred campuses have taken it. And an uncountable number of couples simply met in a Discord server or a Final Fantasy XIV raid group; by the time either of them thought of it as dating, they had spent two hundred hours solving hard problems together. The pattern across every one of these is the same: no meter, and a high cost of authorship. You cannot swipe through three-thousand-word documents, and there is no way to buy a raid group’s esteem. Effort filters better than money, because effort selects for the thing you are actually trying to find.

The honest objection

If the apps were simply collapsing under their own cynicism, this essay could end here, and it would be wrong. Tinder’s paying users fell from about eleven million to under nine in three years, and Bumble cut nearly a third of its staff. Hinge, meanwhile, grew subscribers by 17 per cent and revenue by 26 per cent. The market is not punishing the metered model. It is rewarding whoever performs intentionality most convincingly, and performing intentionality is permanently cheaper than being aligned to it, because alignment costs revenue and performance costs a brand campaign. That is the actual problem, and no redesign of the feed can reach it, because it lives in who pays whom, for what.

What we built, and what we refused

The test of an argument about incentives is what you do when you hold the other end of one. xNet is a workspace, not a dating service, but it carries a people-matching layer, and building it forced every one of these choices onto the table. The shape we chose: your profile is derived, not performed. The system already holds a graph of what you care about, so a shy person never has to write an advert for themselves. Contact is a wave: each side signals interest privately, stored as a salted hash even the relaying server cannot read, and only when both have waved does an introduction open. Nobody messages you cold, and there is no inbox to buy your way into. And romance is one intent among seven, alongside friendship, collaborators, mentorship and the rest — facets of one primitive rather than an app of its own. That last choice matters more than it looks. Carving romance off into a separate product with its own growth target is how the meter got built the first time.

Honesty about the gap, as ever: what ships today is the local half — the derived profile, the wave, friends-of-friends discovery. The federated directory that would let strangers find each other across communities is designed and not yet built, and the moderation bill that comes with it is a real cost that the paywall, for all its sins, currently pays elsewhere. We would rather tell you that plainly than imply a dating service exists.

The part worth writing home about is the part we made impossible. Our continuous-integration gate, the same one that already fails any build containing a streak counter or an infinite scroll, now also fails any build that introduces a metered connection. An identifier like boostPrice, paidVisibility or payToReveal stops the merge. The refused-rents list in our public charter carries the entry in plain words: no rent on introductions, ever, with a test pinning the receipt. A promise with a lint rule is a different kind of promise.

There is a deeper reason the good version has to be unmonetised rather than differently monetised. Suppose you tried to rebuild the shadchan’s incentive honestly and charge only for outcomes. A matchmaker paid on the marriage has to observe the marriage. In a village the observer was a neighbour; at software scale it is surveillance, and the cure is worse than the fee. Which leaves one honest place for connection to live: as a feature of ground you already own, riding a flat bill for hosting, worth nothing extra to the operator either way. The matchmaker cannot be a business. It can be infrastructure.

Designed to be deleted was always the right idea in the wrong tense: an aspiration wearing the clothes of a description. The version that can actually be built is older than the apps and plainer than the slogan — a matchmaker whose incentives face the couple, a room you will be in again next week, a no that lands on the format instead of on your face, and a coin that stays in your pocket, because nothing on the far side of the glass was ever scarce to begin with.

Sources

Figures are as reported by the sources above, most recently checked August 2026. The people-matching layer described here is a workspace feature under active development, not a dating service; the essay says plainly which parts ship today and which do not.