Part II: The six domainsChapter summary

Trust between an organization and an agent must run both ways. The organization wants to know which agent is calling, who runs it and on whose behalf it acts. The agent, and its principal, want to know that the organization is what it claims to be and that its answers and actions are genuine.

Identity is moving from guesswork to proof. Sites have long identified bots by IP address and user-agent strings, which are easy to fake. Proposals such as Web Bot Auth, developed in the IETF, let an agent sign its requests so a site can verify who sent them. In the other direction, an organization can publish signed descriptions of itself and its endpoints at well-known addresses, so that an agent can check it is talking to the real thing.

Agent

Acts for a principal

Signs each request: who runs it, whom it acts for

Publishes signed facts and endpoints at well-known addresses

A written policy decides what each verified agent may read, ask and do.

Organization

A named person answers for it

Figure 8. Identity both ways. The agent proves who it is and whom it acts for; the organization proves it is the real thing.

Identity makes policy possible. An organization can decide that verified agents may read prices, that agents acting for signed-in customers may book, and that only agents with a payment mandate may buy. A written policy for agents states these rules, along with what data agents may keep and what the organization will disclose about its own agents.

Behind every policy stands a person. The Framework’s last principle, keep a person answerable, means that when an agent makes a commitment for an organization, a named human is accountable for it. Agents may act; people remain responsible.

Read the whole chapter

This is a summary. The full chapter is in the book, which is coming soon. One email when it is published.