Where to Look?

Chance favours the prepared mind.

— Louis Pasteur

A thousand profiles in one evening, a thumb worn out from swiping, and a feeling of productive work: the more you scroll, the closer you get to love. The search has turned into a conveyor belt with people on it, and you sort them — this one left, this one right. It feels like the skill of meeting people. In fact it is the skill of avoiding meeting people without looking away from the screen.

Platforms really do widen access, and that is their honest and considerable value. But profiles and compatibility percentages do not solve the next problem: they say almost nothing about what will happen between two particular people after they meet. So the question “where to look?” is not about a magic place but about how a channel is built: whom it gathers, what it lets you see, and how quickly it turns a profile into live interaction.

A profile here works as a ticket in: it should get you to a conversation, not replace one.

The channel pre-sorts people

The place where you meet selects people in advance by features that may be invisible to the participant. A university gathers people of similar age and educational path; a professional environment gathers people with similar jobs and schedules; a religious community gathers people who share some values; a circle of friends gathers people with overlapping reputations and social ties. Sociologists call the tendency of ties to form between similar people homophily. What matters is that part of the similarity comes not from conscious taste but from the structure of opportunity: a person chooses among those an institution has already seated nearby.

This explains the old effectiveness of intermediaries without romanticizing the past. Relatives, neighbours and mutual acquaintances narrowed the choice, sometimes excessively, and brought advance information. You could not easily invent a different biography in front of someone who knows your sister and your former boss. The price was control, closeness and the exclusion of anyone who did not fit local norms.

Online dating removed part of that constraint. In a national American survey, meeting through the internet had become by 2017 the most common way heterosexual couples met, displacing introductions through friends and family. This widened access to weak ties and to people outside the usual circle. Along with it, part of the built-in vetting disappeared. A platform delivers a stranger more efficiently than a neighbour does, but it knows less about his behaviour in ordinary life than the neighbour does.

The result is a trade-off between reach and advance verification. A closed environment gives fewer options and more context. An open one gives more options and more uncertainty. No channel is best in general. If the constraint is the absence of suitable people, you need wide reach. If the problem is the impossibility of judging reliability, what helps is recurring environments, shared projects and ties where behaviour is visible before any romantic offer.

A place is worth judging by how many contacts and observations it creates. A one-off party gives many first impressions. A course, a volunteer project, a sports group or a circle of friends gives fewer new faces per evening, but lets you see a sequence: whether the person shows up on time, how he treats those he does not need, whether he tolerates boredom and keeps promises. The environment becomes part of the measuring instrument.

The channel is solving its own problem too

An intermediary is never just a pipe. A service has its own goals and metrics: active users, subscriptions, ad impressions, time in the app or successful matches. Sometimes these indicators coincide with the person’s goal, sometimes not. A user may need one good contact and an exit from the system; a platform that earns on attention benefits when the search continues.

A divergence does not require a conspiracy of developers. Different metrics are enough. Views, likes, messages and returns are visible immediately; the quality of a match shows up late and, if things worked out, removes both participants from the observed sample. A channel therefore measures activity inside itself better than results beyond its borders.

It is useful to judge a meeting place not by how long it holds you, but by whether it leads to the information you need for your own decision. A good intermediary shortens the path to the next appropriate step. A bad one quietly turns the search itself into the final product.

Reach without a conveyor belt

A search requires two things that are easy to confuse: wide enough reach and real contact. In that sense it resembles surveying terrain: it helps to go where the right meetings are possible at all, and to check more than one plot. The analogy is about distributing attempts, not about rating people.

It rests on three assumptions. Where mutual interest will appear is poorly known in advance, so varied meetings help. Attempts are not independent: reputation, fatigue and your own behaviour link them together. And every meeting has a cost, so endlessly widening reach quickly starts taking attention away from acquaintances that have already begun.

The limit is simple: wide reach must not turn people into inventory. As soon as volume becomes a conveyor belt of candidates, the search is a catalogue again, where profiles are sorted fast and almost nobody is known.

You should look not where there are more people in general, but where the channel corrects your current constraint. With a narrow circle you need reach; with a deficit of trust you need a recurring environment and shared ties; with endless texting you need a quick move to a short meeting. A good channel leads to varied live contacts and lets you see a person outside the prepared shop window. Breadth is needed before contact, depth after it.

Main sources

Finkel, E. J., Eastwick, P. W., Karney, B. R., Reis, H. T., & Sprecher, S. (2012). Online dating: A critical analysis from the perspective of psychological science. Psychological Science in the Public Interest, 13(1), 3–66. On access as the real value of platforms, see also the chapter “A market that is not quite a market.”

Joel, S., Eastwick, P. W., & Finkel, E. J. (2017). Is romantic desire predictable? Machine learning applied to initial romantic attraction. Psychological Science, 28(10), 1478–1489. More on the unpredictability of the “spark” in the chapter “Can the spark be computed.”

McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415–444.

Rosenfeld, M. J., Thomas, R. J., & Hausen, S. (2019). Disintermediating your friends: How online dating in the United States displaces other ways of meeting. Proceedings of the National Academy of Sciences, 116(36), 17753–17758.