The Bid-Ask Spread of Attention

Everyone bids. No one asks.

When you swipe right, you're posting a bid. You're saying, I'd take this, at the prevailing price, where the price is respond to my message and let's see. But the other side's ask is invisible. You have no idea what they'd accept, or whether they're even selling. The platform has already collected your bid and moved you to the next listing.

The asymmetry is the whole problem. Bidding is free and frictionless, so it inflates: tens of thousands of right-swipes per day per user, almost none of them carrying any information. The ask side has no infrastructure to match. There's no way to credibly say I'd accept a 38-year-old founder in London, no kids, who can hold a conversation about Roth and Gale-Shapley, so nobody does. Selectivity stays private, the bid stack overflows, and the market never clears.

In trading we have a name for what happens when one side floods the book with low-information orders: toxic flow. The market makers' first job is to filter it out. The dating equivalent (bots, low-effort profiles, mass-swipers, anyone bidding without conviction) is what every serious user is silently trying to filter when they open the app. The platform isn't built to help with that. It's built to make the toxic flow look like liquidity.

Engagement ≠ effectiveness

Tinder pioneered the slot-machine engagement loop and every major app copied it, because it was easy to instrument and easy to monetize. The metric that gets optimized is time on platform, which is a proxy for revenue. The metric that matters, did this person meet someone they'd want, is invisible to the product. Worse, it's anti-correlated with engagement: a user who finds someone leaves the app.

This is the equivalent of a market venue optimizing for number of orders submitted rather than fill quality. You'd never run a real exchange that way, but for ten years the entire dating product surface has been built on it.

What an actual market maker does

In a real market, market makers exist to provide liquidity by quoting both sides. They post bids and asks, narrow the spread, and absorb adverse selection in exchange for the spread they capture. Their value isn't matchmaking in any romantic sense; what they do is translate opaque private preferences into a standing, executable price.

The dating equivalent of a market maker is a matchmaker: someone who knows both sides' real preferences (explicit and revealed), synthesizes them, and produces a single quote, here's the introduction; here's why it clears, that either party can lift or pass on.

For the last decade, this role didn't exist on the apps. It lived offline, in the form of $50,000-a-year human matchmakers serving a few thousand wealthy clients. Software couldn't do it because software couldn't yet understand a sentence like I want someone ambitious but not anxious about it. That changed in the last eighteen months.

Aria as a quote engine

Aria is the AI matchmaker inside the app I'm building, Aureve. The product surface is deliberately small: one explained introduction a day. You can chat with Aria the way you'd chat with a thoughtful friend (show me more people like the last one I liked, but less intense), and she translates that into structured preference updates. She also reads behavioral signals (dwell time, scroll depth, pass speed) the way an order book reveals real intent versus quoted intent.

Then she does the market maker's job: she quotes. Not swipe through these forty profiles, but I think this person fits because of X, Y, Z. Take it or pass.

The "explain why" part isn't decoration; it's price transparency. In markets, an opaque quote is worth less than a quote with disclosed reasoning, because the reasoning lets the counterparty update their beliefs. Why is this match here? is the same question as why is this bid here?, and a market that can answer it produces less regret on both sides.

The microstructure analogues, made explicit

Once you adopt the market frame, every product decision in dating maps to something a microstructure researcher would recognize:

• Adverse selection. Low-effort users disproportionately consume the attention of high-effort ones. The matchmaker's first job is filtering toxic flow.
• Inventory management. One introduction per day is position sizing. Forced scarcity prevents the user (and the market) from overtrading.
• Latency-aware quoting. Behavioral signals (dwell, scroll, pass speed) reveal real preferences, not stated ones. The book updates from action, not declaration.
• Spread compression. Pricing transparency (the "why" on each match) tightens the spread between what people say they want and what they'll accept.
• Subscription as screen. A paid tier isn't a luxury good; it's a Spence-style signal. Paying $30 a month reveals seriousness in a way swiping for free can't. You don't see noise traders paying for terminal access.

The honest tradeoff

This design compresses choice. Some users will hate it. The free-feed apps deliver a hit of optionality that one-introduction-a-day cannot match.

The bet is that compressed-but-explained choice has higher information content per unit of attention than infinite-but-blind choice, and that the segment of users who are trying to meet someone (rather than collect dopamine) will pay for the difference. Same reason an institution pays for a Bloomberg terminal when Yahoo Finance is free: the price of a bad decision dwarfs the cost of better infrastructure.

We'll find out if that bet is right. But the math of why the current apps fail isn't mysterious. They're running unrefereed venues with no market maker, no explained quotes, no toxic-flow filtering, and an objective function (engagement) that's misaligned with the only outcome that matters.

What dating learns from finance

The reason this language helps is that finance has spent a century thinking carefully about how markets clear in the presence of asymmetric information, signaling, and noise. Roth's work on stable matchings, Glosten-Milgrom on adverse selection, Kyle on informed traders versus liquidity traders, Spence on costly signals. None of it was written about dating, and all of it applies.

The next decade of dating product will be built by people who treat it as a market design problem rather than a UI problem. The winners aren't going to be the ones still iterating on the swipe; they'll be the ones who finally build the market maker.