What's this about?
At a Midjourney meetup, swyx's "aha": knowledge graphs and ontologies are finally trending now — after failing to catch on three years ago — because "good enough" intelligence has become too cheap to meter, making it economical to build and maintain structured knowledge that was previously too expensive.
What exactly is an ontology / knowledge graph?
A structured map of how concepts relate — entities and the labeled links between them (Person —works_at→ Company). Powerful for reasoning, but historically brutal to build and keep current by hand. Cheap LLMs can now extract and maintain these graphs automatically.
Why does it matter?
21 likes — small, but a sharp framing. It's the same "too cheap to meter" flip Theo invokes about Luna today, applied to a different problem: near-free inference doesn't just add features, it revives whole approaches (structured knowledge) that the economics had killed.