---
id: "quote-equimarginal-principle"
type: "quote"
source_timestamps: ["§ Ask the Bot"]
tags: ["microeconomics", "data-mixtures"]
related: ["concept-equimarginal-principle", "concept-data-mixture-weights"]
speakers: ["E. Glen Weyl", "Raul Castro Fernandez"]
speaker: "E. Glen Weyl and Raul Castro Fernandez"
sources: ["tail1"]
sourceVaultSlug: "hbr-seg-tail1"
originDay: 1
articleStem: "hbr-tail-109-ai-pay-fair-rates-content"
sourceUrl: "https://hbr.org/2026/06/how-ai-companies-can-pay-fair-rates-for-the-content-they-need"
sourceTitle: "How AI Companies Can Pay Fair Rates for the Content They Need"
---
# The Equimarginal Principle applied to AI

> "Here we can apply the \"equimarginal principle,\" which is one of the oldest results in the economics of production. It states: if a builder has optimized the mixture, then the last token drawn from each source contributes roughly equally to performance. If news articles were pulling more weight per token than web text, the builder would use more of them until the contributions evened out."

— [[entity-e-glen-weyl|E. Glen Weyl]] and [[entity-raul-castro-fernandez|Raul Castro Fernandez]]

## Context

The authors map economic theory onto the technical realities of training a neural network. This is the plain-language statement of [[concept-equimarginal-principle]] and the justification for treating [[concept-data-mixture-weights]] as a relative-value signal.