Summary

  • Christopher Drake's 25 September Agent Identity Authority proposal is an individual Internet-Draft. It has no formal IETF standing, RFC-stream status or IESG approval; the proposed association is not an established regulator.
  • The draft permits a proposed global production namespace to run under an interim Registry Operator before the Authority is constituted. Its author discloses his and 1id.com's participation in a bootstrap Registrar and reference Registry service.
  • A later governance declaration would require seven groups of conditions, including independent registrar participation, a signed trust store, working escrow and a restore exercise, public comment and a two-thirds Board vote. These are proposed gates, not completed tests.

An identity may be designed to outlive the organization that first issues it. That is useful for software agents whose credentials and transactions have to remain attributable across service changes. It also creates an uncomfortable ordering problem: if identifiers become permanent while their first custodian is temporary, what evidence will let a successor verify the ledger without inheriting the first operator's authority by default?

That question is made concrete by draft-drake-agent-identity-governance-00, posted on 25 September. The Datatracker classifies it as an individual submission, without formal IETF standing or IESG approval. It is a governance proposal, not an IETF decision. Its companion registry draft describes an aid URN scheme, a uniqueness index, Registrars, a Registry Operator and a future Authority. The companion requests formal aid namespace registration from IANA. The currently published Formal URN Namespaces table, checked on 27 September, contains no aid row; its recorded update on 11 September predates these drafts. Neither the draft's word global nor a requested registration is an existing IANA assignment.

The governance proposal explicitly allows a production namespace to start under the interim operator's published commitments before a governing association has been formed. Drake says he and 1id.com participate in the bootstrap deployment, including a Registrar and a reference Registry service. That disclosure matters for evaluating who holds the initial operational keys; it is not independent proof of deployment scale or a finding of improper conduct. The draft says bootstrap participants receive no reserved seat or later operator-selection right, although they may join and compete on the same terms as others.

Its handover checklist is unusually specific. Before the Board may declare the namespace under Authority governance, the association must be legally constituted with an elected Board; an operator must be chosen through an open process and technically audited; at least three accredited and integration-tested Registrars must belong to three different ownership groups across at least two regions. It also calls for a signed Trust Store v1, two approved handle-dispute providers, flowing escrow and a successful restoration exercise, then a final 30-day public comment and a two-thirds Board vote.

The draft sketches an 18–33 month sequence as guidance, not a binding deadline. Listing a condition is not evidence that it has been met.

The crux lies before that vote. A ledger that has issued persistent identifiers can make the initial custodian practically important even if the charter promises an open succession. The draft says bootstrap IDs would remain permanent through transfer, and escrow would support a successor. A credible transfer therefore needs inspectable restore results and a way to distinguish continuity of identifiers from continuity of the current operator. That is Daniel Kade's editorial test, not an allegation that the proposed service has failed it.

Heng Lu's separate arguments about attendance not equalling mandate and a ledger not entitling its keeper to perpetual control sharpen the question; they do not constitute evidence against this draft's author.

The scope is also narrower than the phrase “agent governance” might suggest. This proposed Authority would govern registry infrastructure, accreditation, operator succession and disputes among consenting participants. It would not license models, regulate agent behavior or provide a remote kill switch. Section 14 explicitly notes that a private charter would not itself transfer IANA change control. Readers should keep three claims separate: an individual proposal for identity infrastructure, an operator's disclosed bootstrap role, and a formally recognized namespace or independent governing mandate.

Only the first two are described by these documents today.

Sources