Netflix can't afford content you want, so it scours a limited library for things you may like
Netflix's dumbed-down algorithms — Alexis Madrigal has a rollicking investigation into Netflix's movie genres — all 76,897 of them, from category #1 (African-American Crime Documentaries) …
Context & Ripple Effects
Reuters is picking up Alexis Madrigal's investigation of Netflix's 76,897 micro-genres from The Atlantic three days earlier, and the pickup traveled unusually wide through tech commentary circles — Eugene Wei, Tren Griffin, Dave Chensky and others all shared it within hours. The framing lands on an uncomfortable arithmetic: Netflix cannot license everything its subscribers want, so the taxonomy exists to squeeze maximum relevance out of a deliberately finite catalog.
That logic builds on Netflix's own earlier disclosure, back in April 2012, that its recommendation engine guides roughly 75 percent of what members actually watch. In other words, the company has spent two years telling investors that curation, not acquisition, drives consumption — Madrigal's genre map is the first detailed look at the machinery underneath.
First-order effects
- Netflix's product surface shifts further toward personalized rows built on the micro-genre taxonomy, meaning subscribers increasingly experience the licensed catalog as a tailored shelf rather than a browsable library — the algorithm, not the A-Z list, is the interface.
Second-order effects
- If matching efficiency substitutes for catalog depth, Netflix's spending calculus tilts further toward data infrastructure over incremental licenses — visible already in its December 2013 open-sourcing of Suro, the real-time pipeline tool feeding exactly these systems — while rival services face pressure to match the personalization bar or compete purely on exclusive titles.
Third-order effects
- Media distribution structurally splits between supply-rich players who buy breadth and demand-side players like Netflix who win by measuring and routing taste — making viewer-data systems a core competitive asset on par with content rights themselves.
The trend: Streaming is becoming a matching business, where the value of a finite licensed library depends less on what you own than on how precisely your recommendation system routes it to each viewer.