Dating

Dating is a two-sided matching market with search costs, private information, and no clearing price. Neither side posts a number, both sides can decline, and the traits that are cheapest to observe up front have the weakest relationship to whether the match works over years. Everything strange about modern dating follows from that gap between what is searchable and what is decisive.

Search and selection are different problems

The market has two layers, and they fail for different reasons.

Search is the problem of encountering plausible candidates at all. It is bounded by geography, social graph, and time. This is the layer apps solved. They took a market that used to run on friends-of-friends and workplaces and made it thick, indexed, and continuously available.

Selection is the problem of predicting, from limited evidence, whether a specific person is a good long-run partner. Apps did nothing for this. They arguably made it harder, because they front-load the variables that photograph well — appearance, height, occupation, the ability to write a witty line — and defer the ones that decide outcomes: how someone handles conflict, what they do when bored, whether their stated values survive contact with cost.

The result is a market that sorts efficiently on observable traits and inefficiently on everything else. Assortative mating is the visible consequence: people pair with those close to them on education, income, and background, partly by preference and largely because those are the dimensions the market can actually see and sort on.

Costly signals and the interview problem

Unverifiable claims are cheap, so they carry almost no information. Anyone can say they are kind, ambitious, or emotionally available. Signals worth reading are the ones that would be expensive to fake: long friendships, a track record of finishing things, how they treat people with no power over them, what they have actually given up for something they said they valued.

The deeper issue is that early dating measures performance under observation. Both people are running a version of themselves optimized for evaluation, and both know it. The variable you need is behavior under low observation and real cost, which by definition the early market cannot show you. This is why time, shared logistics, and mild adversity are more diagnostic than any number of good conversations, and why the failure rate concentrates in matches made entirely on the searchable layer.

Where the common understanding is wrong

The advice industry treats dating as a volume problem: get out more, match more, put yourself out there. Volume improves search, which is usually not the binding constraint. Past a modest threshold, more options degrade decisions rather than improving them, because a large candidate set pushes people into comparison mode, where they evaluate on ranked, quantifiable attributes instead of asking whether they like being around this specific person.

The second error is self-optimization aimed at the searchable variables. Improving the profile improves match volume and does nothing to the selection function. If the last five relationships failed in the same way, the failure is in what you select for and how you behave inside a relationship, not in the top of the funnel.

The third is treating chemistry as a signal of fit. Early intensity mostly measures novelty and uncertainty. It is real, it is pleasant, and it predicts very little. The traits that carry a partnership are boring to observe and slow to reveal, which is exactly why the market underprices them.