Pricing Information via Shapley Values for AI Agents

schwentker · x · 2026-08-24

Explores value allocation when AI agents consume vast amounts of information, using Shapley values from cooperative game theory as an estimator. A source's value is its average marginal contribution across all possible orderings. Redundant sources earn little, while critical ones earn more, effectively paying for uniqueness. This raises questions about whether future web moats will be content or proof of its significance.

Related event: Search Costs Overtake Inference in Agent Workflows(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →