← yourmood.social⌸ Newsstand
Wire ResponseWednesday, September 30, 2026

The Dispatch

WHAT THE MATCHING ALGORITHM ACTUALLY OPTIMIZES FOR

Users assume the algorithm is searching for their best match; an engineer explains what it is really trained to maximize, and why those are not the same thing

In plain terms: Dating app algorithms are optimized for engagement — keeping you swiping and returning — not for finding your best match, since a perfect match would end your usage. MOOD removes the opaque matching algorithm entirely: you declare intent and see who's compatible, no black box.

PORTUGAL — Here is the most important thing a user can understand about dating apps, and almost none of them do: the matching algorithm is not trying to find you your best match. It is trying to maximize engagement — your time in the app, your swiping, your return visits, your subscription renewal — and I say this not as a conspiracy theory but as a description of how these systems are literally trained. An algorithm optimizes whatever objective it's given. The objective it's given is engagement. Your happiness is, at best, a proxy the system uses only insofar as it drives engagement — and often, the two point in opposite directions.

Follow the logic to its uncomfortable conclusion. If the algorithm's true goal were to find your ideal match as fast as possible, it would succeed — and success means you leave, delete the app, and stop generating revenue. A dating algorithm that worked perfectly would be a business catastrophe, churning its best customers by satisfying them. So the system is not, and cannot be, optimized for your swift success. It is optimized to keep you engaged in the search — which means doling out enough promising matches to keep hope alive, while never quite delivering the resolution that would end the profitable searching. This isn't malice. It's just what an engagement objective mathematically produces. The algorithm is doing exactly its job; its job was simply never your happiness.

And this reframes every frustration users report. The frustratingly good match who never quite works out. The sense that the app shows you just enough to stay. The feeling of being strung along by a machine — that feeling is accurate. You are being optimized, and the thing being optimized for is not you.

MOOD's answer is radical in its simplicity: remove the opaque matching algorithm entirely. There is no black-box system deciding who you get to see based on a hidden engagement objective. Instead, you declare your intent — your MOOD — and you see the people near you who share it. The "matching" is transparent and user-controlled: it's your stated intent plus proximity, not a secret model optimizing metrics you can't see against interests that aren't yours. This is, admittedly, less magical-sounding than "our AI finds your soulmate." But it's honest, and it can't be secretly turned against you, because there's no hidden objective for it to serve. The most trustworthy algorithm, it turns out, is the one you can fully see — or better yet, the one you don't need, because you told the system what you wanted directly.

The Burnout Diaries

Entry submitted by O., 32, data scientist.

Day 120. I build recommendation systems for a living, so let me just say what everyone suspects: yeah, the dating algorithm is optimized for engagement, not for you. I'd stake my career on it. It's the obvious design if your revenue is subscriptions and time-in-app. And knowing that from the inside made it impossible to keep using the things — every "great match!" started to feel like a slot machine paying out just enough to keep me pulling. I'm a person to them the way a number is a person to a spreadsheet. Can't unsee it. Won't use it.

Reprinted with permission, spelling intact. Submit your own to [email protected].

Readers Also Ask

What do dating app algorithms optimize for? Engagement — your time in the app, swiping, and subscription renewals — not finding your best match, since a perfect match would end your usage.

Does the dating algorithm want me to find someone? Not quickly — your success means you leave and stop generating revenue, so the system is designed to keep you searching, not to resolve the search.

How is MOOD different? It removes the opaque matching algorithm entirely — you declare intent and see nearby people who share it. Matching is transparent (intent plus proximity), not a hidden engagement-optimizing model.

Read Edition No. 26 as a paper →