Dating apps and what they do to judgement

A large proportion of relationships now begin on an application designed by a company with commercial interests that are not identical to the user's. That is worth thinking about carefully rather than either dismissing or accepting.

What the format does

It foregrounds appearance. Not because the designers are shallow but because a photograph is the only thing that can be assessed in the time the interface allocates.

Which means selection happens on the dimension that brief exposure permits, and the dimensions that predict long-term satisfaction are unavailable at the point of choice.

It creates the appearance of abundance. Research on choice generally finds that very large option sets can reduce satisfaction with the option chosen and increase the difficulty of choosing at all.

The evidence for choice overload is more mixed than the popular account suggests, and the effect appears under specific conditions rather than universally. But the specific conditions — options that are difficult to compare, no clear criteria, high perceived stakes — describe dating applications closely.

It encourages assessment rather than interaction. The mode is evaluative from the first moment, which is a poor state for meeting anyone.

It makes rejection cheap and constant. Both directions. The volume of low-cost rejection is unlike anything in ordinary social life, and it has effects on how people treat each other.

The incentive question

Stated plainly because it is frequently treated as a conspiracy theory when it is simply a business model.

Subscription applications earn revenue from active users. A user who forms a lasting relationship stops being an active user.

This does not mean anyone is deliberately preventing matches. It means the metrics being optimised — engagement, session length, return visits — are not the same as the outcome users want, and where those diverge the metric wins.

Which is a general property of attention-funded products rather than a specific accusation.

Using them without being used by them

Move to meeting quickly. Extended messaging builds an impression that the meeting then has to survive, and it usually does not.

A week of conversation followed by coffee is a reasonable pattern. Six weeks of messaging is a different activity.

Reduce the option set deliberately. Fewer conversations, taken more seriously, produce better outcomes than many conducted shallowly.

Set a cap on time. The applications are designed to be used indefinitely, and the marginal value of the twentieth minute in a session is negative.

Write something specific. Profiles composed of generic positives are indistinguishable from each other. A specific and slightly odd detail does more work than any list of qualities.

Do not treat the response rate as information about you. The base rates are extreme in both directions and reflect the format rather than the person.

What the evidence says about outcomes

Relationships that begin online appear to be broadly comparable to those that begin offline on measures of satisfaction and stability, in the studies that have looked.

Some analyses have found small differences in either direction, and the samples are not random, so this should be held loosely.

The clearer finding is about who meets whom. Online meeting has substantially increased the proportion of couples who did not share a prior social network, and increased pairings across some social boundaries.

Which is a genuine benefit and the strongest argument for the format.

The alternative that still works

Worth mentioning, since the applications have crowded it out of the conversation.

Meeting people through repeated shared activity — work, sport, volunteering, hobbies, friends — has properties the applications cannot replicate.

You observe behaviour over time rather than assessing a presentation. There is no evaluative frame. And there is social accountability, which affects how people behave.

It is slower and it requires showing up somewhere regularly for reasons other than meeting someone.

For many people the reasonable answer is both, with the applications treated as one channel among several rather than as the entire method.