romance didn't survive being made efficient.

the numbers can be read as a subscription problem, and some of them are. a payer count tells us who stopped spending, though, not what made the experience feel thin. the more interesting possibility is that the apps kept their promise too well. they delivered reach, speed and legibility on a scale no friend or matchmaker could approach. what they could not preserve was the part of meeting that had never looked productive: the missed signals, the second look, the evening that seemed to be about something else. efficiency removed the inconvenience and, with it, some of the evidence that an encounter mattered.

before the profile grid, romance often arrived with an intermediary attached: a friend, a workplace, a neighbourhood, a class. in the nationally representative 2017 u.s. survey behind stanford's study, meeting online had overtaken meeting through friends among heterosexual couples.⁶ that shift widened access, especially beyond the limits of a person's immediate circle, and deserves more credit than the present backlash allows. it also displaced a quiet form of social information. the friend of a friend came with context; the stranger in a grid came with a claim about themselves. one was slower. the other was easier to scale.

a profile is useful because it makes a stranger legible. it is also reductive for the same reason. height, work, politics and three polished prompts can be compared cleanly; timing, ease and the way a person changes the temperature of a room cannot. a 2012 review of online dating warned that access to very large pools may encourage an assessment mindset quite unlike face-to-face interaction.¹² there is evidence that our own lists are shakier than they appear. in speed-dating research, the traits participants said they wanted beforehand did not predict whom they actually desired at the event.¹¹ the profile asks a sensible question too early: is this person right for me? a room permits a smaller and often more useful one: do i want to keep talking?

some attraction is immediate. much of it is not. robert zajonc's classic work found that repeated exposure can increase liking for unfamiliar stimuli.⁸ that is not proof that proximity manufactures love. it is a reminder that familiarity has its own emotional force. in ordinary life, repetition also reveals information: how someone treats a waiter, tells the same story, behaves when tired, becomes funnier on the fourth meeting than on the first. the swipe is excellent at first impressions. most human charm is not.

in a 2024 online survey of 1,000 u.s. adults who had used a dating app in the previous year, 79% of gen z respondents said they felt exhausted by the apps sometimes, often or always.⁵ the sample does not stand for an entire generation, but the feeling it measures is familiar. romance is the clearest version of a broader design mistake: efficiency becomes corrosive when the process being compressed is also where meaning forms. a queue can be shortened without changing its destination. courtship cannot be separated so neatly from the time it takes.

the industry solved a real problem and mistook it for the whole problem. searching was expensive, awkward and bounded by geography. the apps made it cheaper, clearer and vastly larger. what they could not automate was the conversion of an available person into a particular one. a match can open the door, but it cannot supply the accumulation of small evidence by which two people come to feel chosen. the product is logistics. the relationship begins after the product has done its job.