The Metropolis blog

Let's geek out on economic theory for a moment ... Online dating — at least, the heterosexual kind — is what's known as a two-sided market . In these markets you're trying to match two groups of people, Group A and Group B. They might be buyers and sellers, drivers and passengers, or men and women. The usual dynamic of a two-sided market is that people in Group A will join the platform with the most members of Group B, and people in Group B will join the platform with most members of Group A. Because that's where they have the best chances of finding someone suitable. This causes a positive feedback loop, where growing numbers make it even more popular, which is known in the venture capital world as a flywheel . Usually it results in a winner-takes-all market, with one dominant player, plus a handful of minnows serving niches. A broken flywheel At least, that's the way it's supposed to work. In the world of online dating, however, the flywheel is broken. Inste

Hi, I'm Matt, and welcome to the Metropolis blog. Metropolis is an online dating app that uses machine learning to predict whether two people will be attracted to each other. The purpose of this blog is to talk about the app, and online dating in general. But first, a bit about myself. Until recently I was a software engineer at Google, working on Google Maps in the Sydney office. Prior to that I've been a hedge fund quant, a codebreaker with the Australian defence forces, and was a co-founder of MessageMedia, which recently sold for USD 1.3 billion (I was a contractor, so no equity unfortunately!) I also have an MBA from Melbourne Business School and a PhD in Geospatial Science. LinkedIn profile here . The origin story While I was at Google, Sundar announced that we would become an AI-first company (pivoting from being mobile-first). All of us engineers were expected to become proficient in machine learning. After completing the training I found there wasn't much immediate