Singapore's government dating app reportedly uses Gale-Shapley stable matching

What the post claims

A post on X by the user marianne, which surfaced on Hacker News's front page, claims that FirstDate — a dating app run by Singapore's government — is built on the Gale-Shapley stable matching algorithm, the 1962 result commonly introduced as the solution to the "stable marriage" problem. According to the post, the app is currently limited to public servants aged 21 to 35 and verifies every user through Singpass, Singapore's national digital identity service.

The post also sketches the product: one match per cycle rather than an endless feed, a 72-hour window to decide, and contact details revealed only when both people say yes.

How Gale-Shapley works

David Gale and Lloyd Shapley published the algorithm in 1962. The setup: two groups of participants, each with a ranked preference list over the other group. The goal is a stable matching — an assignment with no two people who would both rather be with each other than with the partner they received.

The procedure, as the post recounts it for FirstDate, is straightforward. User preferences and stated dealbreakers are converted into ranked lists. One side proposes to its top choice. Each person on the receiving side holds the best offer received so far and rejects the rest. Rejected proposers move to their next choice, and the loop repeats until everyone is matched.

The payoff, in the post's framing, is that a familiar dating-app grievance — "the app never showed us each other" — becomes structurally impossible: if two users would genuinely prefer each other to their assigned matches, the algorithm would have paired them. The guarantee only holds for the preferences the system actually knows, which makes honest preference and dealbreaker input the load-bearing part of the design.

Work on stable allocation earned the 2012 Nobel memorial prize in economics, awarded to Shapley and Alvin Roth, and the algorithm still sits inside real infrastructure: the US program that assigns medical graduates to hospital residencies, school-choice placements, and kidney-paired exchange programs.

The proposer advantage

The post also flags a well-known asymmetry: Gale-Shapley is proposer-optimal. Whichever side does the proposing is guaranteed its best outcome among all stable matchings, while the receiving side lands on its worst stable outcome. In labour-market deployments this is a live policy issue. In a dating app it becomes an open design question the post does not answer: how does FirstDate decide who proposes? Whoever draws that role is mathematically favoured, cycle after cycle.

Built to be deleted

The sharpest contrast the post draws is economic. Commercial dating apps monetise attention, so their interfaces optimise for time on app — infinite scroll and endless queues of candidates. A government-run service has no comparable incentive. Per the post, FirstDate's constraints (a single match, a hard decision window, verified identities) push users toward commitment rather than browsing, and the app's real success metric is users deleting it after finding someone.

Caveats

Every specific detail above — the algorithm choice, the eligibility rules, the 72-hour window, the Singpass integration — comes from a single X post amplified on Hacker News. No official documentation or government statement appears in the source material, so the implementation claims should be treated as unconfirmed. Running Gale-Shapley over a dating population also raises operational questions the post leaves unanswered: how preferences are gathered, what happens when someone exits mid-cycle, and how ties or incomplete lists are handled.

Why it matters

Textbook algorithms are rarely pitched as consumer product features. If the claims hold, FirstDate is a notable example of an app marketing a mathematical guarantee — stability — that users can feel without understanding the proof. It is also a counter-model to engagement-driven software: a matching product designed to shrink its own user base. And the proposer-optimality wrinkle deserves attention at national scale, because whoever is assigned the proposing role quietly receives the best stable outcome while the other side absorbs the worst.