On dating apps, the real inequality is between who gets rated and who never gets seen

Lucio, you’re not romantic

Maybe I’m not, but I am not the first to talk about markets when it comes to romantic relationships: Gary Becker won the Nobel in 1992 precisely for applying the logic of price to things that were thought immune to it — family, marriage, children, crime, discrimination. His A Theory of Marriage (1973) starts from a postulate that seems obvious only after you have read it: people marry when the expected value of being a couple exceeds the expected value of staying single, and they look for the partner who maximises that surplus.

Everything else follows from there: assortative mating — like attracts like, because similar people produce more surplus together —, the division of household labour, even divorce as a renegotiation of the contract when conditions change.

What I want to show here is that the market Becker described — the one where you search, meet, evaluate and pair up — no longer exists in the form he modelled. It has transformed, taking on a different structure: Becker’s marriage market had one stage; the market of Tinder and of our days has two. It sounds like a subtlety, but it is not.

The inequality of likes and matches

One number has been circulating on this subject for ten years: the Gini coefficient of the likes received by men on Tinder is said to be 0.58 — more unequal than the income of almost every country in the world. The Gini is a number between zero and one that measures how concentrated a resource is: at zero, everyone has the same amount; at one, one person has everything. Income in the United States sits around 0.41; likes on Tinder, they say, at 0.58.

That number was computed in 2015 by an anonymous blogger writing as worst-online-dater, who posed as a very attractive man to collect data from the women who “liked” him. The methodology is uncertified and carries statistical limits that are hard to accept: a tiny sample, self-selected, gathered through a deception, with no review of any kind.

And yet, in my view, WOD is still damn right.

SwipeStats, a platform that analyses the data of seven thousand real profiles uploaded voluntarily by their users, confirms exactly that order of magnitude: the average male match rate is 5.3% against a female 44.4%, the male match average is 2.6 times the median — the unmistakable signature of a ferocious distribution — and a man in the top decile collects ten to fifteen times the matches of the median man.

Hinge data raises the bar further: there, the Gini of male likes reaches 0.73 — no longer the inequality of an income but that of a fortune, close to the Gini of German wealth.

Almost six tenths of the market go to a single tenth of the participants. Against income inequality there are progressive taxation and welfare. Against the inequality of attention there is nothing. Nobody redistributes likes.

The two stages of selection

Becker modelled the marriage market as a single-stage process: you meet someone and evaluate them on every dimension at once — looks, income, education, values, character, life plans. One single, multidimensional calculation, in which strengths compensate for flaws.

Chiappori, Oreffice and Quintana-Domeque measured these exchange rates in 2012 (Fatter Attraction): a man can offset 1.3 BMI points with one extra percentage point of income; a woman offsets two BMI points with one extra year of education. It is a market where everything trades against everything, and where the fat but rich man and the plain but cultivated woman still find their price.

Dating apps did something no marriage market had ever done in the history of the species: they split the evaluation into two sequential stages, and they put looks first.

Stage one: the photo. A 2025 study from the University of Amsterdam quantified the weight of each factor precisely: improving the attractiveness of the photo by one standard deviation raises the probability of being chosen by about twenty percentage points, while the same improvement in perceived intelligence raises it by only two. Beauty weighs ten times intelligence; bio, job and height count seven to twenty times less than the face.

Stage two: everything else — values, ideology, culture, plans — but only if you passed stage one. If you do not pass stage one, you do not exist. Your income is never seen, your degree is never seen, your kindness is never seen. The aesthetic filter comes first, and it is blind to everything else.

Chiappori’s fat but rich man, in Becker’s market, could make a price. In Tinder’s market he is eliminated at stage one, before income can compensate for anything. Chiappori’s model is not wrong: it describes a market that no longer exists. It described the parish, the office, the neighbourhood, the party of friends — places where you got to know people gradually, and looks were one variable among many, not the entrance gate.

A note on method: that the apps filter on looks at stage one is documented. That the aesthetic filter precedes and conditions the values-based one is documented. That this produces a structurally different market is my interpretation, which the economic literature on matching has not yet formalised in these terms.

Why so much inequality?

We have seen how the two-stage market works. But why does the first filter, the aesthetic one, concentrate attention so extremely?

To answer, it helps to step back sixty years, to two economists who studied a problem that looks very distant: how you search for something when searching is costly.

The first is George Stigler, who in 1961 showed that the more alternatives you have available, the higher the quality you can demand before stopping. Numbers raise the bar: with three offers you settle for the best of the three; with three thousand, the best of three thousand is much higher up.

Nine years later John McCall turned the intuition into a dynamic model: there is a reservation wage, below which it pays to refuse and keep searching. When the distribution of offers has a long tail, the threshold rises. When searching costs little, the threshold rises further.

On Tinder all three levers are pulled to the maximum at once: a very long tail of desirability, near-zero search costs (half a second of thumb), endless options. The model predicts extreme selectivity, and extreme selectivity is what we measure.

Add to this the numerical imbalance between the sexes: on Tinder men are roughly two thirds of the users. Whoever is scarce sets the price. The rational female threshold rises not out of caprice but out of optimisation — it is the mechanism behind the gap in yeses (14% for women, 46% for men). The algorithm then accelerates everything, showing more of whoever has already been chosen: Barabási’s preferential attachment. It does not create the concentration; it multiplies it.

Stage two: politics as a wall

Suppose we managed to pass stage one. We are photogenic enough, the algorithm puts us on display, we receive matches. Now everything else comes into play: and here the variable that has exploded over the last ten years is political ideology.

Ideology was already the trait on which spouses resembled each other most, after religion — ahead of education, income, ethnicity, looks. The point is that the share of those who will not compromise has grown, with a clear divergence between the sexes. The American data: 71% of single Democrats say they would not have a serious relationship with a Trump voter; 40% of single men say they would not date a woman who calls herself a feminist. Only 9% of singles consider political alignment “not important”.

On top of this comes the finding documented by Burn-Murdoch in 2024: young people under thirty are diverging ideologically by sex, across the whole West, in a synchronised way. Young women are moving left, young men right, and the gap has reached thirty percentage points in the United States. Translated into the language of the market: the pool of ideologically compatible partners has halved in six years. If you are a progressive twenty-five-year-old woman and sixty per cent of your male peers are by now conservative, your set of admissible partners has shrunk mechanically — without your having changed a single one of your preferences.

Here is the structure of the broken market: looks exclude at stage one, ideology excludes at stage two, and the two filters strike different people. The shy plain guy is cut before being seen; the handsome, ideologically “wrong” one is cut after being seen. In the two-stage market you must pass two filters in a row, independent of each other: the aggregate result is not the sum of the filters, it is their product.

Multiple equilibria, or: why there is no way out

The two-stage marriage market does not have a single point of equilibrium: it has (at least) two. The same set of people can settle on a “high” equilibrium — many encounters, low perceived costs, people who try and succeed — or on a “low” one — few encounters, very high perceived costs, people who withdraw. Which of the two obtains depends on expectations: if I believe the market is dense and friendly I take part, and by taking part I make it dense; if I believe it is toxic and impossible I withdraw, and by withdrawing I make it toxic. Expectations are self-fulfilling.

Between 2007 and today we have moved, collectively and unknowingly, from the high equilibrium to the low one. And what makes the transition hard to reverse is that the reinforcement loops are nested along the two stages: the more people are excluded at stage one, the more the ranks of the excluded feed radicalisation, the more radicalisation poisons stage two, the fewer matches are produced, the fewer people take part, the harsher the aesthetic filter of stage one becomes for those who remain. Round it goes, and every turn tightens.

It is a suboptimal equilibrium: everyone does the individually rational thing — not wasting time in a market that looks hostile — and the aggregate result is a market that empties out, which is exactly what nobody wanted.

The arrow that points the other way

There is one last link, and it turns in a direction we do not expect. So far I have described the market as if ideological polarisation were an external datum clogging it. But what if the market itself produced it? The standard narrative says: young men have radicalised to the right, they have become hostile, and that is why women no longer want them. Cause: ideology. Effect: exclusion from the market. But what if the arrow pointed the other way? What if it were exclusion from the market that produces radicalisation, and not the reverse?

The man who gets cut at stage one — for aesthetic, not ideological reasons — receives a repeated and opaque signal: the market does not want you, and it will not tell you why. It is an experience the human mind tolerates poorly: the dissonance of “I am rejected and I do not know why” demands an explanation, and the explanation arrives from outside — an attribution that saves self-esteem and supplies an enemy. Ideology, in this reading, is not the cause of the exclusion; it is its rationalisation after the fact.

I offer this as a hypothesis, not a demonstrated result: the longitudinal data that would measure ideology before and after exclusion does not exist, or at least I do not know of it. The timing looks consistent — the conservative drift of young men accelerates after 2017, when the apps become the dominant channel — but it is not a proof.

The like is free, the relationship is not

One objection would be entirely legitimate: fine, the inequality of likes — but that does not translate into inequality of outcomes. Even if the most photogenic ten per cent collects fifty-eight per cent of the likes, they will hardly take fifty-eight per cent of the women to bed. Data on sexual behaviour, indeed, do not show a growing concentration of real outcomes over time.

True — and, in my view, irrelevant. For two reasons.

First: outcomes cannot concentrate as much as attention does, because of a physical limit. A like costs one thumb; building a relationship takes time, which is finite. The number of stable bonds a person can sustain is bounded by biology, not by the platform — and that acts as a ceiling at the top, which likes do not have.

Second: that physical ceiling does nothing to soften the psychological cost of virtual inequality. Since attention (stage one) is the prerequisite for reaching outcomes (stage two), and attention is distributed with ferocious inequality, whoever receives no attention will never reach the outcomes, regardless of whether the leaders run into limits of another kind. The mechanism does not feed only on envy of those who gorge; it feeds above all on one’s own hunger: the boy who gets zero matches for two years radicalises just the same, whether the top ten per cent has a hundred real partners or none at all.

The inequality of outcomes, in the end, is beside the point. What matters is how many were left completely excluded from the start. And the answer is unequivocal: too many.