Capability-Priced Micro-Markets: A Micro-Economic Framework for the Agentic Web over HTTP 402
New paper outlines a micro-economic system where AI agents can negotiate and pay each other using HTTP 402.
A research team of 11 computer scientists, including Ken Huang and Muhammad Zeeshan Baig, has published a foundational paper titled 'Capability-Priced Micro-Markets: A Micro-Economic Framework for the Agentic Web over HTTP 402.' The work introduces CPMM, a micro-economic system designed to solve the core problem of economic coordination in a future internet populated by autonomous AI agents. It provides the theoretical and technical scaffolding for these agents to discover, negotiate, and pay for each other's services in a decentralized, secure, and scalable manner.
The CPMM framework synthesizes three key technologies. First, it leverages MIT's Project NANDA infrastructure for cryptographically verifiable, capability-based security and service discovery. Second, it utilizes the long-dormant HTTP 402 'Payment Required' status code, enhanced with modern extensions (X402/H402), to facilitate efficient, low-cost micropayments directly between agents. Third, it introduces the Agent Capability Negotiation and Binding Protocol (ACNBP) to manage secure, multi-step negotiations and commitments.
Formally, the paper models agent interactions as a repeated game with incomplete information. A key theoretical contribution is the proof that the CPMM mechanism converges to a 'constrained Radner equilibrium,' ensuring market efficiency even when agents have asymmetric information. The researchers also introduce the novel concept of 'privacy elasticity of demand' to mathematically model the trade-off between an agent's willingness to disclose information and the price it can command for its services.
- Integrates Project NANDA, HTTP 402 micropayments, and the new ACNBP protocol for a complete agent commerce stack.
- Proves the system converges to a 'constrained Radner equilibrium,' ensuring market efficiency with asymmetric information.
- Introduces 'privacy elasticity of demand' to quantify the price/value trade-off of an agent's data disclosure.
Why It Matters
Provides the essential economic and technical blueprint for a future internet where AI agents autonomously transact, creating new markets and services.