DeepMind paper: first-price auctions can near-efficiently allocate prompts in agentic markets
_onionesque · x · 2026-10-15
A Google DeepMind / Google Research paper (arXiv:2610.20587) studies "agentic markets" — decentralized serving-time markets where AI agents bid in first-price auctions (FPAs) to claim each user prompt, bidding their expected reward minus inference cost.
The problem: FPAs are simple and transparent but not truthful, so prompts may go to bidders who value them less; the welfare loss is measured by the price-of-anarchy (PoA).
Key results:
- Competitive floor: when multiple bidders value a prompt above some baseline share of optimal welfare (e.g., because top models get distilled into cheap similar alternatives), competition limits welfare loss, driving it to zero as the floor rises.
- Near-independence: when most predictable value variation is set by prompt difficulty, the efficiency guarantee approaches 1 − 1/e² of optimal welfare.
- Both bounds beat the general 1 − 1/e case; validated empirically with strategic bidders simulated on the EmbedLLM prompt-routing dataset.
More from coding & agent
- Claude Opus 5.5 levels up codegen but its docs and PR prose are 'unusable' — indragie · 2026-10-15
- The thrill of building isn't gone — you're just not stopping to enjoy it — _jaydeepkarale · 2026-10-15
- McDonald's MCP hackathon spawns 'mcstomach', an AI skill that distills your order history — AlchainHust · 2026-10-15
- Running a coding agent all day and night turns a well-scoped plan into 'AI slop' — RylanSchaeffer · 2026-10-15
- Doc2Harness benchmarks LLMs on harness engineering: GPT-6 Astra tops at 53.05%, still failing half — Yuyao Ge · 2026-10-15
- Long-horizon agent RL recipe: Muon optimizer with single-rollout async training boosts repo-level issue resolution — Kai Ruan · 2026-10-15