Summary

  • draft-eggert-procon-chair-delegate-02 would permit, but neither require nor recommend, the IETF Chair to delegate responsibilities to other Area Directors. It specifically reaches the IESG chair role, the General Area Director role and the Chair's full IAB membership.
  • The draft requires public transparency about which role went to which Area Director and for what time. That is a useful minimum for workload allocation, but it does not by itself resolve the authority source, consent, participation, voting, conflicts and responsibility that follow a full institutional seat.
  • Current RFC 2850 makes the IETF Chair one of thirteen full IAB members while excluding the Chair from approving IESG members and hearing appeals against IESG decisions. Public list discussion has therefore questioned whether membership is a delegable task and whether IAB consent is needed. Those are individual views, not an IETF or IAB decision.
  • If the proposal advances, it should distinguish an operational delegation from an institutional-seat substitution. A public delegation charter should preserve scope, exclusions, consent, acts, responsibility, correction, revocation and the return of authority without exposing private workload or health information.

A small notice with a large verb

The attractive version of delegation is easy to picture. The IETF Chair has several jobs. One Area Director can run a meeting, another can carry a bounded General Area portfolio, and the Chair remains available for the decisions that cannot sensibly move. Work is distributed, the institution gains fault tolerance, and no one has to pretend that a single person has unlimited time.

Lars Eggert's The IETF Chair May Delegate begins from that practical problem. It says the role has historically been filled by one person, although the foundational documents never precisely collected all of the Chair's duties in one place. Few documents expressly permit delegation. A full-time Chair may still confront periods in which higher-priority tasks crowd out other duties. A single bottleneck, the draft says, creates obvious scaling and fault-tolerance problems.

Revision 02 would let the Chair decide whether to delegate, which responsibilities to move, to which Area Director and when. It does not require delegation. It does not even recommend it. It creates permission. It also says the community must be able to see which roles were delegated to whom and for what time, perhaps on the IESG website or through public email. Operational details would remain with the IESG.

That notice is a sensible floor. Secret delegation is incompatible with reviewable authority. But three fields—role, person, time—do not carry the same information on every surface. “Chair the Tuesday session until Friday” is an assignment of performance. “Serve as General Area Director for this portfolio” may carry a defined set of Area Director powers. “Hold the IETF Chair's full IAB membership” reaches a separate body's composition. The same verb hides three different constitutional moves.

The distinction matters even though this is not yet policy. At the 28 August research cutoff, the Datatracker classified revision 02 as an active individual Internet-Draft. It had no RFC stream, no responsible Area Director, no telechat date and the IESG state I-D Exists. The document header seeks Best Current Practice status, but the status page assigns no intended RFC status. PROCON lists it as related work rather than an active Working Group draft. It has no formal standing and has changed no RFC.

The bundle the draft would unbundle

Current documents accumulate several offices in one person. RFC 9281 says the IETF Chair chairs the IESG and is Area Director for the General Area. The next section says the Chair is also an IAB member. RFC 3710 similarly names the Chair as General Area Director when the area is active. RFC 2850's IAB charter goes further: the IAB consists of thirteen full members, the IETF Chair and twelve sitting members.

Revision 02 would loosen all three joints. It would remove RFC 9281's sentence combining the IETF, IESG and General Area roles. It would allow another Area Director to chair the IESG. It would let the Chair share or delegate the General Area Director role. And it would amend the composition language so that the IETF Chair or the Chair's delegate is an IAB member.

There is a strong organizational case for the first two changes. Chairing and area management consume attention. Their outputs can be described: convene, set an agenda, shepherd a defined portfolio, sign a named class of procedural act. A principal can assign those tasks and remain answerable for the assignment, provided the public can identify the delegate and challenge an act.

The IAB provision is structurally different. RFC 9281 describes the IAB as providing architectural oversight, reviewing proposed Working Group charters, confirming IESG candidates and serving as an appeal board for complaints about the standards process. Its members deliberate as a collegiate body. A seat is not merely a queue of chores. It is a position inside the body that produces institutional judgment.

RFC 2850 already recognizes that the Chair's position carries conflicts. The IETF Chair may participate in official IAB actions except approval of IESG members and appeals against IESG decisions. Revision 02 does not announce that those exclusions disappear. Its proposed edits would put the delegate into the Chair's composition clause and replace another reference with “the IETF chair or their delegate.” A careful implementation would therefore have to preserve the exclusions and show how they bind the substitute. A generic public notice cannot leave readers to infer that result.

Selection and onward selection are different acts

The IETF Chair is not simply a manager allocating private staff time. The Chair reaches office through the NomCom selection and confirmation structure. A voter or selector can assess one candidate's judgment, availability and ability to carry the bundle. If that person later transfers substantial parts of the bundle, the institution acquires an onward-selection problem: which qualities were selected in the Chair, and which qualities are now being supplied by someone the Chair selected?

This does not make delegation illegitimate. Every durable institution needs substitutes, deputies and division of labour. It does mean the first mandate cannot be treated as an unlimited power to manufacture identical mandates for others.

Joel Halpern raised this point in public PROCON discussion after revision 02 appeared. A NomCom, he observed, may select someone without knowing which parts of the job the candidate intends to delegate. Choosing a person to do work and choosing a person who will choose someone else require different capabilities. He suggested that NomComs could ask candidates what they expect to delegate.

That is an individual contribution, not adopted procedure. It nevertheless identifies a real information asymmetry. A candidate need not lock in an operating plan before circumstances exist. But the selecting body can distinguish ordinary contingency from a planned structural split. “I may occasionally assign bounded tasks” is different from “I expect another AD to chair the IESG and occupy the IAB seat.” The first is resilience. The second changes what the selection actually purchases.

The public record should therefore have a before-and-after dimension. Before selection, candidates can state any expected structural delegation at a useful level of generality. After selection, the Chair can publish the actual charter. If circumstances change, the difference should be visible rather than treated as misconduct. Transparency is not a promise that plans never move. It is a record of who is now exercising which authority and why the change remains within mandate.

The receiving body is not furniture

Delegation is often described from the sender's perspective: the Chair owns a responsibility and passes it to another AD. That description may work for work internal to the Chair or IESG. It becomes incomplete when the delegate enters another institution.

The IAB is not a spare room in the Chair's office. Its charter defines its membership and functions. A proposal to substitute a full member therefore has at least two sides: the Chair's power to designate and the IAB's authority to recognize the designation. Public discussion quickly found this edge.

Halpern argued that changing IAB participation should not be left to the IESG without explicit confirmation from the IAB. Mirja Kuehlewind wrote that IAB membership is not merely a role or responsibility that a member can delegate, especially where voting rights are concerned. Brian Carpenter, who had first proposed a broad delegation rule, replied that the IAB-seat aspect at least needed IAB consent.

These messages do not establish a legal answer. They are not an IAB resolution, a Working Group consensus call or a change to revision 02. They establish something more modest and useful: informed participants do not agree that “full membership” can be analyzed as ordinary work allocation. That disagreement deserves an explicit decision before a final text is treated as operationally complete.

Consent is not ceremonial duplication. It answers which institution recognizes the substitute and under which rule. It can also settle whether the delegate is a member for participation, voting, quorum and internal duties; whether existing conflict exclusions attach automatically; and how a contested act is corrected. Without those fields, public transparency might identify the human being while leaving the authority invisible.

Responsibility needs a return address

Carpenter's first message proposed a shorter general rule: a Chair could delegate a specific part of the role to another AD unless a BCP forbade it; delegations would be time-bounded and announced; ultimate responsibility would remain with the Chair. Revision 02 does not contain that final responsibility sentence. It is a participant's proposal.

The proposal is attractive because it gives every act a return address. A Chair cannot dissolve responsibility by distributing tasks. If the delegate misses a deadline, exceeds scope or creates a conflict, the community knows who selected and supervised the delegate.

Yet “ultimate responsibility remains” may still be too compressed for a seat. The Chair can remain responsible for choosing, defining and revoking a delegate. The delegate may be responsible for an individual vote or statement. The IAB may remain collectively responsible for its decision. A reviewing body may be responsible for interpreting the charter. These are not rival answers; they are layers of responsibility attached to different acts.

A good record should refuse the choice between “the Chair owns everything” and “the delegate owns everything.” It should show assignment responsibility, performance responsibility and institutional decision responsibility separately. That avoids two familiar failures. The Chair cannot say the act belonged entirely to the delegate. The delegate cannot say the title supplied authority without personal judgment. The body cannot describe a contested decision as a private arrangement between two officeholders.

Revision 02 drew one useful boundary

Earlier versions combined voluntary delegation with an emergency stand-in for an incapacitated Chair. Revision 02, posted on 29 June 2026, removed that section. A separate draft published the same day now proposes emergency stand-in succession.

The split improves the analysis. A capable Chair voluntarily distributing work is not the same state as a Chair unable to act. The first has a live principal who can define, supervise and revoke the assignment. The second needs a trigger, a succession order and a rule for determining when the principal returns. A mid-term vacancy under RFC 8713 is different again because it selects a replacement rather than a temporary performer.

This article stays on the voluntary side of that line. It makes no claim about the current Chair's health, workload or performance. It does not assess the separate stand-in design. The document history is evidence that one word—delegation—was already carrying too many states. Splitting incapacity from voluntary distribution was the right move. Splitting operational work from institutional membership is the next question.

A charter, not a nameplate

The minimum public object should be a delegation charter. For ordinary operational delegation it should bind the controlling RFC or draft version, principal, delegate, exact function, included and excluded acts, start and end, renewal, acknowledgment, conflicts, recusal, subdelegation, correction, revocation and the event that returns authority to the Chair. A reason class can explain whether the driver is workload, continuity, expertise or scheduling without publishing private health or personnel details.

Material acts should then receive small receipts. Who chaired the meeting? Who approved the procedural step? Under which charter? Was anyone recused? Was the act corrected or superseded? This is not a demand to publish deliberation. It is a way to prevent a later reader from reconstructing authority from job titles and guesswork.

If the delegation reaches an institutional seat, a second class of fields becomes mandatory: the separate authority source for substituting membership; consent or confirmation by the receiving body if the final rule requires it; participation and voting scope; preserved appeal and confirmation exclusions; quorum treatment; allocation of responsibility; and a return receipt when the seat reverts.

The distinction keeps the common rule thin. Every meeting assignment should not need a constitutional ceremony. Every seat substitution should not be hidden inside a meeting assignment form. Governance becomes resilient when work can move quickly and authority can still be traced precisely.

Sources