Summary

  • AFPUB-2026-GEN-001-DRAFT01 would require PDWG co-chairs to check that an author’s presentation slides match the proposal text before an AFRINIC Public Policy Meeting.
  • If the slides differ, the draft requires an updated proposal on the mailing list to notify the working group, but says the changes may still not be considered. Publication is therefore not the same event as admission.
  • AFRINIC should attach a version receipt to each presentation: proposal and deck hashes, any variance, the notification message, the version’s procedural status and the exact text eligible for a later decision.

A rule about the object in the room

Policy procedure is often written as a sequence of verbs: submit, discuss, present, decide, appeal. The nouns are easier to neglect. What, exactly, has been submitted? Which text is being discussed? What did the slides explain? Which version can the chair put to a consensus test? A process can satisfy every visibility ritual and still fail to give one answer to those questions.

AFRINIC’s current proposals list records AFPUB-2026-GEN-001-DRAFT01 as under discussion. The proposal page dates the submission to 20 May 2026 and names Grégoire EHOUMI, Noah Maina and Adeola A. P. AINA as authors. The draft is ambitious: it would replace or amend parts of the Consolidated Policy Manual governing the Policy Development Working Group, and it ranges across chair responsibilities, appointments, consensus, recall, appeals and continuity.

One pre-meeting duty is unusually specific. Proposed section 3.3.2.1(f) tells the co-chairs to read the presentation slides, familiarise themselves with the details and ensure that the slides match the proposal text. If there is a difference, an updated version of the proposal must be submitted on the mailing list to notify the working group, “even if these changes will not be considered”.

The provision is not yet a binding AFRINIC rule. It belongs to a draft that remains under discussion. It is nevertheless a precise recognition of a real governance problem: a slide deck can become a parallel policy instrument without ever declaring itself one.

Slides are designed to compress. They turn qualifications into bullets, tables into takeaways and exceptions into footnotes, if the exceptions survive at all. That is not misconduct; it is the form. A speaker also adds examples and answers questions in language that may be clearer than the filed text. The danger begins when clarity is mistaken for amendment. A room can leave believing it heard one bargain while the durable proposal still contains another.

The draft’s answer is not to ban explanation. It creates a boundary. The proposal text remains the object; the slides must be checked against it. A difference triggers an updated submission and notice to the list. Then the final phrase applies a second boundary: the fact that the update was published does not mean its changes were eligible to count.

Notice is not admission

That distinction deserves more attention than the grammar gives it. An updated proposal posted after a slide mismatch can do at least three jobs. It can document what appeared in the presentation. It can preserve language that the authors want the working group to consider next. Or it can become the operative version for a later consensus decision. Those jobs have different legal and procedural effects even if they use the same PDF or web page.

A mailing-list message proves that a file was announced at a time. It does not by itself prove that the required review period restarted, that participants had enough notice, that objections addressed the new language, or that the chairs treated that language as eligible. A URL proves availability. It does not prove status.

The draft itself supplies the warning. If changes can be posted and still “not be considered”, then the system recognises a class of public but non-operative text. The public record must make that class visible. Otherwise a reader arriving later sees Draft02, a deck and meeting minutes, but cannot tell whether Draft02 records the meeting, governed the meeting, or only prepared the next one.

This is where version laundering becomes possible without deception. A later document may inherit the legitimacy of an earlier public discussion simply because the archive presents them as one continuous proposal. The new language looks discussed because its version history sits beside the meeting. The meeting looks more complete because the later language repairs its ambiguities. Neither inference is warranted unless an admission event joins them.

The required event need not be theatrical. It can be a small status transition: received for notice; admitted for discussion after a stated date; eligible for consensus at a named meeting; superseded; withdrawn. What matters is that the transition is explicit and attached to an immutable version.

What AFRINIC-37 did—and did not do

AFRINIC’s 37th Public Policy Meeting minutes are useful precisely because they narrow the possible claim. The meeting took place on 24 June 2026, with AFPUB-2026-GEN-001-DRAFT01 scheduled from 10:15 to 11:05. The minutes link the author presentation, and AFRINIC’s presentation index identifies the corresponding 496 KB PDF. The 20-page deck names the same proposal and authors.

The minutes say discussions were ongoing on the Resource Policy Discussion mailing list. More importantly, they say the authors were not ready to seek consensus and had requested extra time to present the concepts and policy details. Alain Aina described the purpose as presenting the concept and gathering feedback. In the closing account, the proposal was sent back to the mailing list.

That chronology prevents an easy overclaim. There is no basis here to say that a slide mismatch changed an AFRINIC-37 decision, because the authors did not ask for a consensus decision on the proposal. There is no basis to declare the session invalid, accuse the authors or chairs of using a different text, or infer that any participant was misled. The article’s subject is the rule proposed for future meetings, not an allegation about this presentation.

The RPD announcement adds another useful coordinate. On 25 May, the co-chairs told the list that they had received Draft01, linked the proposal and asked the community for support, opposition, points of ambiguity and possible improvements. That is a recognisable receipt for the submitted draft: an identifier, authors, public location, date and invitation to review. The presentation layer should inherit the same discipline.

AFRINIC’s current Policy Development Process and Consolidated Policy Manual provide the institutional baseline against which a future version would move. The new proposal seeks to specify working-group machinery more fully. Its slide-matching clause should therefore be read as procedural design, not as evidence that the present manual already imposes the entire proposed control.

The comparison needs coordinates

“Ensure they match” sounds binary. Documents rarely are. A slide may shorten a sentence without changing it. It may omit a condition because there is no room. It may replace “shall” with “should”, change a deadline, add an example that behaves like an exception, or introduce a diagram whose arrows imply authority that the prose never grants. A meaningful check cannot end with a green tick beside two filenames.

The comparison needs a declared unit. Some differences are editorial: typography, ordering, abbreviations. Some are explanatory: a faithful example, a simplified chronology, a diagram of the same roles. Others are substantive: changed actors, thresholds, dates, powers, burdens, appeal rights or conditions. The receipt should classify the difference and cite both the slide and the proposal section. It need not reproduce every formatting change.

Hashes solve one part of the problem. A SHA-256 digest can identify the exact proposal bytes and deck bytes reviewed before a meeting. It cannot decide whether the words are equivalent. That judgment still belongs to accountable people. But without the hashes, even a careful judgment can later float free of the files it addressed.

Time matters as much as identity. The record should say when staff received the deck, when a co-chair completed the check, when a difference was reported, and when any updated proposal reached the list. If the update arrives after the meeting begins, its role cannot be silently backdated. If it arrives before the meeting but after a minimum discussion window, availability still may not equal eligibility.

The record also needs a negative. When no substantive variance is found, say so and bind that result to the two hashes. Silence is ambiguous: perhaps the files matched, perhaps nobody checked, perhaps a difference was found and resolved informally. A negative attestation turns absence from a gap into evidence.

A practical presentation-to-text receipt

The receipt can be compact enough to publish beside the deck. It begins with the proposal ID and version, canonical URL, submission timestamp and content hash. It records the deck URL, receipt or publication timestamp, page count and hash. It identifies the responsible co-chair and completion time for the comparison.

If there is a difference, a variance table names the slide number, proposal section and a short description of the change. It distinguishes editorial, explanatory and substantive differences. For substantive differences it links the required mailing-list message and the updated proposal, each with its timestamp and hash.

Then comes the field that most archives omit: procedural status. Is the updated text notice-only? Is it admitted to discussion from a particular date? Can it be considered at the next PPM? Did it restart a discussion interval? Is the earlier version superseded for future work but still the version that governed the recorded meeting? The answer should not be inferred from a higher draft number.

The meeting record should cite the exact eligible version. A consensus announcement, if one eventually occurs, should do the same. Corrections to minutes, slides or proposal text should create new records rather than overwrite the evidence that participants saw. This is elementary document control, but here document control is constitutional infrastructure: it tells a distributed community what object its consent can attach to.

The receipt is Theo March’s recommendation, not an artifact AFRINIC says it has promised. Nor does the reviewed public material prove what comparison logs AFRINIC keeps internally. The recommendation makes a narrower claim: when a proposed rule turns matching into a co-chair duty and expressly separates publication from consideration, the public record should show how that duty was discharged.

Sources