---
title: "The Bitter Lesson"
type: "entity"
entity_kind: "concept"
status: "hub"
canonical_name: "The Bitter Lesson"
aliases: []
first_seen: "2026-07-27T00:00:00.000Z"
writer_model: "claude-sonnet-5"
connects_to: ["[[entity-rich-sutton]]","hand-designed vs. learned-at-scale","neural scaling laws","convolutional networks","computation-leveraging methods"]
---


Rich Sutton's 2019 essay, and the vault's shorthand for the pattern it names:
across roughly 70 years of AI research, general methods that leverage
increasing computation eventually beat approaches built on hand-engineered
human knowledge (chess, Go, speech recognition, computer vision). The vault
treats it as the connective hinge between
[[myth-lecun-1988-hand-designed-kernels-was-denker-et-al|a 1988→1989 hand-designed-to-learned-kernel myth]]
and a cross-domain
[[observation-population-scaled-improvement-hits-a-sublinear-brake-across-domains|scaling-diminishing-returns bridge]] —
the essay names the pattern but is silent on the *rate* of the returns to
scale that later work quantifies.

## References
- [[claim-sutton-2019-bitter-lesson-names-pattern-silent-on-rate]]
- [[claim-kaplan-2020-scaling-law-exponents-are-small-diminishing-returns]]
- [[myth-lecun-1988-hand-designed-kernels-was-denker-et-al]]
- [[observation-population-scaled-improvement-hits-a-sublinear-brake-across-domains]]
- Captures: 2026-07-27-hop-bitter-lesson-scaling-brake