How We Broke Dating

Listen to the essay read by the author

0:00

/1984.342676

1×

Online dating was meant to widen the field. It sorted it instead — and the conclusions both sides have drawn from what remained are, by the design of the system, almost guaranteed to be wrong. A long argument from inside the frustration.

Let's talk about Hinge.

Their tagline — printed on the marketing, threaded through the brand, drilled into every product update — is "Designed to be deleted." The intention is generous, even noble. The app, they're telling you, is so confident it can help you find your person that it's actively rooting for you to leave. We don't want you here forever. We want to help you out the door, into someone's arms, and into whatever you build next.

The irony is that almost everyone I know who deletes Hinge deletes it for the opposite reason. Not because they found their person. Because they couldn't bear another evening of not finding them. Because the matches weren't coming, or weren't going anywhere when they did. Because every profile was starting to look like the last profile. Because the hours they were spending were producing no result that resembled the result they'd come for. They deleted Hinge in frustration, not in fulfilment. The product is designed to be deleted — just not, it turns out, by the mechanism the slogan implies.

I am one of those people. I have been single long enough to hate the dating apps.

This essay is for the men who already know exactly what I mean by that — and for the women who, in slightly different language, are saying the same thing back. I'm one of those men. I know plenty of those women. And the longer I listen, the more convinced I am that both halves of this conversation are describing the same room from opposite sides of it — and that both halves, for reasons we'll get to, are drawing exactly the wrong conclusion from what they're seeing.

Let me tell you what it looks like from where I'm standing.

I've had brief forays onto Hinge over the last few years, and not one of them has lasted more than a couple of months before I've deleted the app, again, in frustration. The reason is the one Scott Galloway named on Diary of a CEO last March: the average man, on Tinder, has to swipe right roughly two hundred times to get a single coffee date. And of those coffee dates, a real proportion end in being ghosted before the kettle is even on. (DOAC, March 2025) That number does more to describe what modern dating actually feels like than any think-piece I've read in the last five years. Two hundred swipes. For one coffee. With another human being. Whose name you don't yet know. And whose interest in you, even after the swipe, will probably evaporate before either of you turns up.

The women I am presented with — not matched with, I hardly match with anyone — sell themselves, almost as a one, on Sunday roasts and a long walk in the park, ideally with a dog. The dog is sometimes theirs. Often it isn't. The walk is rarely as specific as the photo suggests. The roast is fine, I'm sure. But it's the seventh profile in a row this evening, and the woman is not the roast, and the roast is not telling me anything about the woman.

When somebody does write something different — something with a bit of texture, a bit of risk, a bit of recognisable interior life — that person, almost by the laws of physics, has thousands of likes already. The small economy of distinctive self-representation gets absorbed at speed by the few people brave enough to attempt it. Everyone else is passed over not because they're uninteresting, but because they sound like everyone else, and the medium rewards sounding like roughly as much as it punishes being it.

So I've arrived where, I suspect, a great many men have arrived. Matching rarely. Knowing that a match is usually the start of a conversation that doesn't go anywhere. Knowing that the road from match to date to anything actually unfolding is so improbable, given the volume of effort to traverse it, that more often than not I just close the app. There is no chance, on Hinge or Tinder or whatever Bumble has become this month, of matching at a meaningful rate. No chance of making eye contact, so to speak. No chance of being seen — because being seen, on these apps, depends almost entirely on either being in the top fifth of one's gender's visual hierarchy, or having mastered a particular skill of light, fluent, knowing self-presentation that I don't believe is the same skill as being a good partner.

And the women I know who are still in the apps — the women I trust, the women I'd want any of my closest friends to meet — describe an experience that mirrors mine almost exactly, except inverted. Too much volume, of the wrong kind. Lots of attention, none of it warm. Hours of conversation that goes precisely nowhere. The same exhausted resignation, just at a different end of the same broken thing.

But here is the thing that bothers me most, and which most of the discourse about dating apps misses entirely.

The people I think I'd most want to meet — the men and women who are genuinely ready for a relationship, who've done the inner work, who'd be wonderful to be with for anyone lucky enough to find them — many of them aren't on the apps at all. Some have deleted and stayed deleted. Some never signed up in the first place. They're put off by the reputation, even when they don't have a single first-hand bad experience to draw on. The rep, by now, precedes the reality. The very people the apps need most — the deep, considered, ready-to-meet-someone people — are voting with their feet and walking quietly away from the venue altogether. The dating apps are losing exactly the audience that would make them work.

All of that is true at once. That is the thing I want to walk into.

I. The sorting

The first thing online dating did, almost by accident, was sort everyone.

You may have heard the headline figure, in some version: that twenty per cent of men get eighty per cent of the matches. The number is a slogan, not a study, and the truth underneath it is messier and more interesting. But the slogan is broadly right.

The cleanest data is from inside the apps themselves. In 2017, an engineer at Hinge published an internal analysis of how "likes" were distributed across the platform. The picture was lopsided in a way that surprised even the company. The top one per cent of male users on Hinge were absorbing roughly sixteen per cent of all the likes coming from women. The top ten per cent of men were absorbing about half. Half of all the female attention on the platform, going to ten per cent of the men. The corresponding figures for women were noticeably less extreme. (Quartz / Aviv Goldgeier, 2017)

Economists have a tidy way of measuring this kind of unfairness — they call it a Gini coefficient. Skip the maths; the only thing you need to know is that zero means everyone gets the same and one means a single person gets everything, and that real-world economies sit somewhere in the middle. A reasonably equal country like the Netherlands runs at around 0.3. A deeply unequal one — say, post-apartheid South Africa — runs around 0.6. The Hinge analyst calculated the inequality among men on the platform at 0.73. That is not a national-economy number. That is a number you'd reach for if you were describing a kleptocracy. The female side of the platform was at 0.63 — bad, but recognisably within the range of an actual country.

A separate, less rigorous but consistent piece of analysis on Tinder data published a couple of years earlier reached a similar place. It found that the bottom eighty per cent of men on Tinder were effectively in competition for the bottom twenty-two per cent of women, and that the top seventy-eight per cent of women were funnelled into competition for the top twenty per cent of men. (Worst Online Dater, 2015)

The 2025 picture is, if anything, worse. A recent dataset of around 7,000 Tinder profiles found women's match rates running roughly 8.4 times higher than men's — 44.4% versus 5.3%. The median male match rate is around 2%. (SwipeStats, 2025)

You can argue with any one of these figures in isolation. You can't really argue with the picture they paint together. Online dating, as currently constituted, doesn't distribute attention. It concentrates it. A small subset of men receive a torrent of attention they can barely read; the great middle of men receive close to nothing; and a smaller but real subset of women are functionally invisible at the bottom of the same distribution. The system is performing the function of a sorting hat. The problem is that it's sorting on the wrong things — and that the sort, repeated over a decade, has driven away the very populations that would have made the apps work.

II. The Bumble paradox

The cleanest counter-argument to all that — and the one I find most poignant — is Bumble.

Whitney Wolfe Herd founded Bumble in 2014, after leaving Tinder, around a single piece of moral architecture: in heterosexual matches, only the woman could send the first message. The point was to interrupt a culture of women being deluged with low-effort, often hostile messages from men they hadn't agreed to engage with. The point was to give women back a little of the agency the swipe economy had quietly stripped away.

For about five years, it worked. Bumble grew, IPO'd at over seven billion dollars, and became, especially among younger women, a moral preference, not just a product. (CNN, April 2024)

Then, slowly, the floor came up.

What women started telling Bumble — in feedback, in research, in the volume on the support inbox — was that having to make the first move every single time felt less like empowerment and more like another full-time job. Tens of matches a week, all of them sitting there expectantly, all of them silently asking go on, then. By 2024, Bumble was openly describing this as "burden" in its own communications. The company introduced a feature called "Opening Moves" — a pre-set prompt women could leave for men to respond to — and the gendered "first move" requirement was effectively dropped. (NPR, May 2024)

The market punished this clarity. Within a year of the change, Bumble laid off thirty per cent of its workforce, replaced its CEO, and brought Wolfe Herd back. (Yahoo Finance, January 2025) Wolfe Herd, in interviews around her return, said something quite striking: that over the last decade, the dating-app industry — including her own product — has left a great many people feeling judged, rejected, and worse off than they were when they signed up. She wasn't blaming users. She was looking at her own product honestly.

The Bumble story isn't really a story about whether women should make the first move. It's a story about a deeper assumption — that the architecture of the swipe could be tweaked into kindness if only the right gender held the gate. It can't. The problem isn't who clicks first. The problem is what people are clicking on, and what they're being asked to evaluate, and which people have decided the whole exercise is no longer worth the time.

III. What both sides have wrong

Here is the part of the argument I most want to land, because it is the part with consequences.

Two beliefs have ossified, over the last decade, on either side of the heterosexual dating divide. On the female side: most men on the apps are shallow, performative, allergic to commitment, and looking for one thing. On the male side: most women on the apps are dismissive, demanding, judging on photos, and quick to ghost. These beliefs are now so entrenched that they read, in conversation, as observations of natural fact. That's just how men are. That's just how women are.

I want to argue that both beliefs are wrong, and that the way they are wrong is structural, and that the way they are wrong tells us something important about what an actual answer would have to do.

Start on the male side, and follow the arrows.

The dating apps select powerfully. The men who do well on them — who get matched, messaged, asked out, courted — are concentrated in the top fifth of the visual distribution. They receive more attention than they can usefully process. They have, by definition, abundant choice. Men with abundant choice tend, on average, to behave less attentively than men without. They have less to lose by being non-committal. They can afford to be slow to text back, ambivalent, half-in. The very behaviour women complain about is, in part, a behaviour the system is selecting for and rewarding. Sometimes it's called the fuck-boy loop. The phrase is cartoonish but the mechanism is real.

Below the top quintile, the rest of the male population is splitting into two camps. There are the men who haven't yet given up — the men still pushing through their two hundred swipes for a coffee, who, by the time they get a match, are often arriving at the conversation with the desperation of a man who has been ignored ninety-nine times in a row and is now putting everything into the hundredth. And there are the men who have given up — who have deleted, often years ago, and are not coming back. These are, by and large, the steadier, more considered, more relationally-ready men. The men who would be wonderful partners. The men women say they want. They are not on the apps at all. They have no economic reason to be there; the system has been telling them, with the brutal clarity of empty match queues, that there is nothing for them in this venue.

So when a heterosexual woman opens Hinge tonight, the men she sees fall into two rough buckets. There are the top-quintile men, in whose inboxes she is one of hundreds of conversations, who treat her with the casual ambivalence of people who have abundant choice. And there are men below that line, who she will probably not match with, but whose enthusiastic messages she will sometimes receive and sometimes find off-putting in their own way. What she does not see, in any meaningful number, is the third population: the men who would make wonderful partners and aren't there. The settled ones. The thoughtful ones. The men her best friend's older sister married. They've left, or never came, and the app has no mechanism to show her what's missing.

From this incomplete sample, she draws a conclusion. The conclusion is that men on the apps are shallow, performative, allergic to commitment — that all men, basically, are bastards. The conclusion is, in a strict logical sense, correct about the data she has. The men she has seen, in the volume she has seen them, do exhibit those traits, and she is not imagining it. What she's wrong about is the inference from the sample to the population. The men available to her on the apps are not a fair sample of men. They are a sample that has been filtered, by the architecture of the apps themselves, to exclude precisely the kind of men she actually wanted to meet.

Now run the same story in reverse on the male side. The man who is still on Hinge, sending message after message into a void, is not seeing the calm, deep, considered women either. He sees the women whose attention has been so deluged that they are, by the time he reaches them, jaded and short of patience — sometimes for very good reason. He sees the women still on the apps after years of disappointment, whose tolerance for low-effort openers has run thin in ways he is now meeting at the front edge of her response. He sees women writing the same Sunday roast bio because the ones who wouldn't write it left a long time ago. From his sample he draws a conclusion: that women are demanding, dismissive, full of themselves. The conclusion is, again, correct about the data he has. The women he has seen do, in number, exhibit those traits, and he is not making it up. What he's wrong about is the inference from the sample to the population.

So both sides arrive at parallel beliefs — the other side is awful — that are simultaneously well-evidenced from inside their own experience and structurally guaranteed to be wrong. The data they have is bad. The data they have is bad because the people the apps would need to be present, in order for the data to mean anything, are gone. Both sets of conclusions are sound logic on bad data.

Most of the public conversation about dating treats the all men are X / all women are Y exchange as a culture war. It isn't, mostly. It's a sampling error. And recognising it as a sampling error matters, because it changes what you do about it. You don't fix a sampling error by arguing with the conclusion. You fix it by changing the sample.

IV. The photograph was always the problem

Underneath the sorting, underneath the burden, underneath the loop — there is a deeper architectural choice, and it deserves its own paragraph, because it's the thing driving everything above it.

Online dating, as a category, runs on photographs.

Tinder is the purest expression: a photo, a swipe, a one-bit decision. Hinge gives you prompts, but the photos are still the headline. Bumble likewise. The first second of evaluation, in every product on the market, is a face. The bio is a tiebreaker, at best. Recent profile-optimisation data, oddly enough, confirms it: short bios outperform long ones by 73%. Men who don't list a job get more matches than men who do. Bartenders out-match software engineers by 3.5 to 1. None of this is because nobody cares what someone does for a living. It's because the photo has already done the work, and the bio is just adding noise. (SwipeStats, 2025)

Photographs are an extraordinary medium for some things. For the question would I be in a relationship with this person? they are a remarkably bad one. They tell you almost nothing about how a person speaks, how they listen, how they fight, how they grieve, how they organise a Saturday morning, what they are like to live next to. A single image cannot convey character; it can only suggest a body. And when an entire dating culture is built on bodies suggested through pixels, what you optimise for is the body. The character — the thing you would actually have to live with — is left to be discovered later, often by surprise, often unhappily.

This is not a moralistic point. It's a structural one. The medium shapes the outcome. If you want a different outcome, you have to change the medium.

V. Why everyone sounds the same

The Sunday roasts I opened this essay with — the identikit bios, the dogs that aren't theirs, the ten thousandth person to declare the hallmark of a good relationship is a good flirt-to-roast ratio — are downstream of all this.

There's a piece in Stylist from a couple of years ago about exactly this phenomenon, and the line that stayed with me is roughly that every single person on dating apps appears to have morphed into the same person, or at least their bios would have you believe so. Sunday roasts. The Office. "Partner in crime." "Looking for someone who doesn't take themselves too seriously." "Down for adventures." Travel photos in the same five countries. (Stylist on dating-app cliché)

It's tempting to call this laziness. It isn't, mostly. It's something more interesting and more tragic.

When you sit a person down with a blinking cursor and ask them, in 150 characters, to be themselves — really, properly themselves, the version that another human being might fall in love with — most of us reach for the nearest cliché not because we're lazy but because we're frightened. The blank box doesn't give you a structure for going inward. It gives you the exposure of inwardness without the scaffolding for it. And when humans are exposed without scaffolding, we pattern-match. We write the line we've seen everyone else write, because at least then we won't say something embarrassing. The Sunday roast is a defence mechanism. So is first round's on me if it's tequila. They're a way of being on the platform without actually being seen on it.

Which is, when you think about it, the whole dilemma in miniature. People come to dating apps to be found. They write profiles that ensure they can't be.

VI. What's actually underneath

The case Galloway and Logan Ury made on Diary of a CEO last March — and Ury's wider work over the past decade — sits underneath all of this and deserves naming. Their argument, simplified, is that the difficulties on dating apps are downstream of a much larger crisis in how young men in particular are growing up. Fatherlessness, the decline of male role models, a cultural narrative about masculinity that careens between toxic and absent without ever quite arriving at worth pursuing. A generation of young men are arriving at the age where they're meant to be partners not having been, in any deep sense, taught how. (DOAC, March 2025)

Ury, on the same podcast, made an observation I keep returning to: that on the emotional skills required to actually be in a relationship, men are running roughly a decade behind women — women in graduate school, men still in primary. That gap is not a moral failing; it's a developmental fact. And no dating app, no matter how cleverly engineered, can close it.

I want to be careful here, because this argument is easy to weaponise. It can become "men are broken, women are right" or its opposite, "women's standards are unrealistic, men are owed a chance." Both of those collapse the actual picture. The actual picture is that we are mid-way through a long, painful renegotiation of what men and women are to each other now that the old scripts have stopped working — and we're trying to do that renegotiation through an interface designed for thumb-speed.

It is going badly. Not because anyone in particular is at fault. Because the interface is wrong for the task.

VII. What would actually have to change

Let me lay out — honestly, as criteria — what any actual answer to this would have to do.

It would have to give people a way to surface depth, not just surface. A way to say something true about themselves that wasn't a roast and a long walk.

It would have to filter by intent, not by swipe-volume. The people who pay attention should be the ones who get found. Right now the system rewards the people best able to capture attention, which is a different and worse skill.

It would have to make profiles different from each other again. Identikit bios are a symptom of identikit prompts. You can't ask everyone the same six questions and expect twenty-seven different answers.

It would have to respect that humans will still respond to photographs — we're animals, the visual instinct isn't going anywhere — but make the photograph a companion to a person's voice, not a substitute for it.

And, most importantly of all — because this is the lever underneath every other lever — it would have to be a venue the people who've already left would actually consider returning to. Not because it was another swipe app with better marketing. Because it was something genuinely different in kind. The deep, considered, ready-to-meet-someone people who walked away from the apps will not be lured back by a new algorithm. They will only return for a place that recognisably solves the problem they walked away from. And without them, the sample stays bad, and the conclusions both sides draw stay wrong, and the whole thing rolls round again.

That is the brief. That is the challenge.

VIII. The DateKnown project

This is the work my DateKnown project is undertaking.

Suppose, for the sake of argument, that every single thing in the previous section were possible. Suppose someone built a way for a person to describe themselves — really describe themselves, in their own voice, with structure to lean against rather than a blank box to drown in. Suppose the descriptions came out genuinely different from each other, because the framework underneath them respected that human beings actually are genuinely different from each other, and refused to flatten everyone into the same six prompts. Suppose the artefact this produced wasn't another bio sitting in a swipeable grid, but a dedicated, text-first page at its own URL — the page somebody visits when they actually want to know more about you, after the photograph has done its initial work, or not.

Suppose there was a small price attached. Not free; not high. Enough to filter out the people for whom dating is a numbers game and not a search. Not so much that any serious person would be priced out. The price acts as a signal, before anyone has read a word.

Suppose this page didn't try to replace the dating apps. Suppose, instead, it was the destination they pointed at — the thing your Hinge link, your Bumble profile, your Instagram bio could all gesture toward when curiosity has been piqued and someone wants more than a face. And suppose, crucially, that the people who'd already given up on the apps — the deep, considered, ready-to-meet-someone people who voted with their feet — found this surface intelligible enough, different enough, and quiet enough that they'd be willing to come back. To make themselves visible again, in a venue that didn't punish them for being who they were.

Suppose, in short, that the bottleneck in modern dating is not really a shortage of people, or a shortage of opportunity, or a shortage of swipes — but a shortage of self-representation that's worth reading, in a place that the people worth meeting would actually go. And suppose someone built the missing surface for it.

The how — the framework that makes this kind of writing possible, the architecture of the page itself, the mechanism by which a real person becomes a piece of writing that sounds like them — is something I'm going to walk through in another piece. What I want to land here is the simpler claim:

It is possible to do better than this. It would not require dissolving the apps. It would not require everyone to suddenly become more attractive, more articulate, more emotionally evolved. It would not require men and women to suddenly become friends with each other again, after this long and difficult decade. It would only require building a different surface — a place that asked a different question, and gave people the structure to answer it well, and looked enough like a venue worth attending that the missing populations would consider walking through the door again.

That is the challenge the DateKnown project is rising to. Not the answer to the whole crisis. A specific page, for a specific kind of meeting, between people who have already realised that being chosen by more is a much worse target than being known by the right one.

Now, the honest part.

DateKnown does not solve attraction inequality. It can't. If a man is at the bottom of the visual distribution on Hinge, he's at the bottom on DateKnown too — at least in the first second of evaluation. What it changes is what happens in the second after that. On a dating app, a low-photo-tier man rarely gets the second second; the swipe is already gone. On a DateKnown profile, the page is unfolding, the writing is doing its work, and the person reading has already done enough to land there. That's a different surface. It serves a different sort of meeting.

DateKnown does not dissolve the fuck-boy loop. The men with abundant choice will still have abundant choice. What it can do is build a parallel lane for the men and women who have already realised that abundant choice isn't actually what they're after — that being known by the right one is a much better target than being chosen by more.

DateKnown does not fix the masculinity crisis. We can't solve fatherlessness. We can't teach a generation of men what nobody taught them. We can give the ones who are doing that inner work — and there are more of them than the discourse suggests — a way to be seen for the work, not penalised for not having spent the last five years cultivating a face.

And DateKnown does not pretend to be the whole answer. It is one honest bet about one specific failure of one specific medium. The bet is that when people get a tool that helps them say something true and specific about who they actually are — calibrated by a framework that respects the variety of human personalities, instead of flattening everyone into the same six prompts — a meaningful number of them choose it. And the ones who choose it, including the ones who'd long since given up on this kind of search, find each other.

That's the bet. It might be wrong. We'll know quite soon.

I started this essay by talking about Hinge — by saying that almost everyone I know who deletes the app deletes it for the opposite reason their slogan implies. They delete in frustration, not in fulfilment. The medium is wrong for the task. No amount of patience with it will fix the architecture underneath.

And neither will any conclusion you've drawn about the opposite sex while watching that architecture fail. The man you've been imagining when you say all men are X is not most men. The woman you've been imagining when you say all women are Y is not most women. The samples you have were rigged. The conclusions you drew were sound logic on bad data. There is more, on the other side of the divide, than the apps have shown you. There always was.

But there is a question I have been asking myself, late at night, for the last few years. It's the same question I keep ending coaching sessions with. And I'm going to end this essay with it too, because at this point it's the only question I really trust:

If the version of you on Hinge tonight is the version someone falls for — would you want to be in a relationship with the person who falls for that version?

For most of us, said honestly, the answer is no. We wrote a Sunday roast. We didn't mean it. The match we'd actually want is the one made by the person who isn't on Hinge at all — the one who'd recognise the actual us, if the actual us were anywhere on the page.

That's the page DateKnown is trying to be.

Not the answer to the whole crisis. Just a page.

A real one.