Summary

  • RFC 2026 §6.5.4 obliges the person who files to set out "a detailed and specific description of the facts of the dispute" inside a two-month window, then leaves every deciding stage free to define its own procedure and asks only for a decision "within a reasonable period of time" — a heavy input duty and almost no output duty.
  • Julian Mehnle's Sender ID appeal produced two dated dispositions at two stages: an IESG reply of 8 December 2005 that conceded merit in his concerns and changed nothing without naming a test, and an IAB decision of 2 March 2006 that named a process test and denied the appeal under it.

One paragraph, two different duties

The document that creates the appeal route is RFC 2026, the Internet Standards Process statement that carries BCP 9, published in October 1996 and written by S. Bradner. Its Section 6.5 is titled "Conflict Resolution and Appeals" and divides into working-group disputes, process failures, questions of applicable procedure and the appeals procedure itself (RFC 2026 index record). The escalation ladder is in Section 6.5.1: working-group chair, then the Area Director or Directors, then the IESG as a whole, then an appeal to the IAB, whose decision the text calls "final with respect to the question of whether or not the Internet standards procedures have been followed and with respect to all questions of technical merit" (RFC 2026, HTML rendering). Section 6.5.2 adds that the IESG "shall issue a report on its review of the complaint to the IETF", after which an appeal may be lodged with the IAB, which may annul an IESG decision and recommend action but may not pre-empt the IESG's role.

Read those two sections and the route looks symmetrical: somebody decides, somebody reviews, somebody finishes. Section 6.5.4 breaks the symmetry. It addresses the filer in the imperative twice and the decider in the imperative once, and the two burdens are not comparable. The filer must provide "a detailed and specific description of the facts of the dispute" and must file "within two months of the public knowledge of the action or decision to be challenged". The deciders get discretion instead of a duty: "At all stages of the appeals process, the individuals or bodies responsible for making the decisions have the discretion to define the specific procedures they will follow in the process of making their decision." The only thing the text demands of the answer is timing — a decision, and its communication to the parties, "within a reasonable period of time" (RFC 2026 §6.5.4).

That is the whole structural finding, and it can be stated before a single appeal is opened. A person who wants to challenge an IETF decision must hand over his theory of the case in writing, on a clock, in enough detail to be answered. The bodies that answer him may write their own rules at each stage, and the text gives them no obligation to state the standard they applied. An appeal route built this way can produce a perfectly legitimate outcome that a stranger cannot check. Whether it does is an empirical question, and one file answers it.

The subject of this briefing is the IETF itself, whose directory entry carries the body's institutional record. Prior briefings on that card have read standards documents and a working-group case file; none has read the appeal instrument. The delta here is the route and one named filer's passage through it.

The filing the paragraph produced

On 25 August 2005 a formal appeal went to the IESG, addressed to its chair, Brian Carpenter, and opening: "as per the Internet Standards Process, section 6.5, and on behalf of the SPF project, I am filing a formal appeal on the IESG's approval on 2005-06-29 to publish the draft-lyon-senderid-core-01 I-D as an Experimental RFC as-is" (Mehnle appeal, record 109). The grounds were technical and specific in exactly the way the paragraph demands: the Sender ID draft, in treating records beginning v=spf1 as though they meant spf2.0/mfrom,pra, collided with draft-schlitt-spf-classic-02, whose published records had been written with MAIL FROM and HELO in mind. The remedy was equally specific: amend section 3.4 so that v=spf1 means spf2.0/mfrom, and withhold publication until the conflict was resolved.

Two record facts have to travel with that quotation. The retrieved copy of the appeal stops before the end of its justification and its references, so what is quoted here is what the record returned, not a summary of an argument nobody read. And the IESG's own appeals index prints this row dated 2005-08-25 while the artifact links inside the same row read "Appeal - 2005-08-24" and "Response - 2005-12-08" (IESG appeals index). The filing date used here is 25 August 2005 because the row and the artifact title both carry it. A register that disagrees with itself by one day is worth reporting as exactly that, and nothing more.

Merit, conceded, with no standard attached

The IESG answered on 8 December 2005. Its reply, as the record returns it, opens: "The IESG has reviewed the contents of the appeal and subsequent email discussions, and while we have found merit in Julian Mehnle's technical concerns, we will not change our decision to publish the draft as an Experimental RFC without change to its technical content" (IESG response, record 110).

That sentence does the damage to any theory that an appeal forces reconsideration: the body agreed with the appellant about the substance and left the document untouched. The rest of the reply reasons rather than concludes. It points to a lower bar for Experimental RFCs, records that the conflict had been considered, notes a caveat added to the documents, and treats the requested edit as having "amounted to a substantive technical change".

What it does not contain is a formula of denial, and what it does not name is a test — no articulation of the kind that would let a reader ask why this objection failed and a different one might have succeeded. That absence is a reading of the reply as returned, not a claim about a document nobody has seen; the reply's own words are the evidence, and they are quoted above.

It is also the most useful kind of negative result. A body that rejects an appeal and says nothing has told the appellant nothing. A body that agrees with the appeal's substance and rejects it anyway has told him that the standard of decision lies somewhere else, and has not said where.

A standard appears at the second stage

Mehnle went up a level. The IAB records receiving the appeal on 8 February 2006 and issuing its response on 2 March 2006, expressly "According to the procedures in Section 6.5.2 of RFC 2026", and disposes of it in one sentence: "The appeal is denied and the IESG's decision is upheld" (IAB response, 2 March 2006).

This time there is a named test. The IAB read Sections 4.2.1 and 4.2.3 of the same document as permitting the publication of conflicting experimental work when a disclaimer accompanies it, found that the IESG "gave due consideration to the technical issues" and "reached a decision according to the process specified by RFC 2026", and noted that one of its voting members, Bernard Aboba, was recused.

Set the two replies side by side and the asymmetry the paragraph permits becomes visible in a single file. Stage one produced a decision with no stated standard and an outcome that left the appellant's specific request unfulfilled. Stage two produced a decision with a stated standard and an explicit disposition. Both are decisions communicated within a reasonable period of time. Both satisfy Section 6.5.4. Only one of them can be argued with by someone who was not in the room, which is the difference between an answer that is published and an answer that is reviewable.

The route is not decorative

The obvious objection is that two dated replies prove nothing except that paperwork generates paperwork. The record argues otherwise, in both directions.

On 15 February 2003 the IAB responded in the Elz matter that the IESG's decision and its handling of the appeal "were incorrect", and held that draft-ietf-ipngwg-addr-arch-v3-11 must not be published as a Draft Standard in the form then before it (IAB response, Elz). The retrieved rendering is partial; its grounds included a failure to verify two independent interoperable implementations. The appeal route, in other words, has produced a finding that a decision was wrong.

It has also produced point-by-point engagement. In 2023 John Klensin appealed a revision to the IESG's guidance on in-person and online interim meetings, and the IESG's reply of 24 August 2023 answered his specific complaints one at a time — that announcements need not be limited to working-group lists, that interim meetings cannot be closed, that Area Director approval for online interims had been dropped in the 2008 revision, and that RFC 2418 does not require more precision — on the basis that the statement "does not violate any community decision-making principles" (IESG reply, record 41). The reply artifact carries the 15 August 2023 appeal date; the 24 August date comes from the appellant's later filing to the IAB.

At the second stage in that file the IAB concluded of 19 December 2023 that IESG statements do not require community consensus, that the revision did not violate existing RFC policy, and therefore that "the IESG did not violate a process in publishing the statement and did not violate a process in handling the appeal" (IAB response to Klensin). It declined to annul. Unlike the 2006 Mehnle decision, the text as returned does not carry an explicit formula of denial, and a concluding sentence may sit below what the record displayed.

Earlier still, on 5 October 1999, the IAB declined to annul the IESG's approval of RFC 2451 as a Proposed Standard while recommending a thorough technical and editorial review (IAB response, Simpson). That text is partial too, and whether it names a standard of review was not established.

Four files, four different shapes of answer: a reasoned rejection that conceded the point, a reasoned rejection with a named test, a decision found incorrect, and a refusal to annul accompanied by a recommendation. The route decides things. What varies is how much of the deciding it publishes.

Scope matters here. Both appeal registers rendered only in part on 19 November 2026. The IESG index showed appeals from 2023, 2025 and 2026 naming John Klensin, Aijun Wang, Timo Gerke, Dan Harkins, Andrew Lee and a long tail of D. J. Bernstein complaints; the IAB index named Klensin, JFC Morfin, William Allen Simpson, Timothy McSweeney and Shyam Bandyopadhyay (IESG appeals index; IAB appeals index). No total count is established from what was displayed, and none is claimed. The observation is that the route is used, repeatedly, by named individuals.

What the body's own page adds

The IAB maintains a page describing appeals to it, restating Section 6.5.2 and listing three requirements for anyone filing: a detailed and specific description of the facts, a preferred remedy, and initiation within two months of public knowledge (IAB appeals page).

The first and third match the instrument. The second does not appear in the Section 6.5.4 text read for this briefing. That makes it a discrepancy between the body's own page and the paragraph that authorises the body — a small thing, and the kind of small thing that matters most, because a filer who reads the page before filing will believe he owes a remedy he may not owe, and a filer who reads the RFC will not know what the page expects. Two accounts of one requirement, neither reconciled on the record.

The same care applies to a document that looks like an appeal route and is not. RFC 8713, which carries BCP 10 and dates from February 2020, governs how IAB and IESG members and some IETF Trust and LLC office-holders are selected, confirmed and recalled (RFC 8713 index record). Its own dispute clause is a confidential arbitration whose results are "final and binding. There is no appeal", expressly limited to the circumstances the document indicates. It is not authority for anything in the file above, and conflating the two would invent an appeal route that does not exist in one instrument while ignoring the one that does in another.

What a stranger can check

Here is the shape of the thing, stated as plainly as the documents allow. An individual outside the institution wrote out his case in detail, on time, in the form the text demanded. One stage answered by agreeing with him and doing nothing. The other answered by naming a standard he had never been told was the test, applying it to the record, and upholding the decision. Two dates, two signatures by role, one file.

Nothing here establishes that either decision was wrong: the Sender ID and SPF drafts were experimental work, the IAB's reading of Sections 4.2.1 and 4.2.3 is on the page, and a reviewer who thinks the technical judgement was correct is not contradicted by anything above.

The gap is elsewhere. Section 6.5.4 could have required the deciding body to state the provision it applied and why the facts failed it. It does not, and so the accountability of an IETF appeal rests on the habits of whoever answers rather than on the instrument. When those habits hold, as in the IAB's 2006 decision, the record becomes checkable after the fact. When they do not, the appellant receives a decision he cannot argue with and the rest of us receive a paragraph we cannot audit.

What is missing is not reasons in general — reasons appear, sometimes in detail — but the stated ground of refusal, which is the one thing a reader needs in order to disagree.