[rpd] [Last Call] Draft Policy Proposal - Hierarchical Names for New AS-SETs (AFPUB-2026-ASN-001-DRAFT02)

[Tshepo Masuku]

Dear Nishal and Seun,

I disagree with both responses.

First, this is not about whether the proposal has persuaded one particular participant. The objection is that those seeking mandatory registry enforcement have not demonstrated that compulsion is necessary, proportionate, and technically sufficient.

“Incremental benefit” is not the correct threshold for mandatory policy. Many operational practices may offer some benefit. That does not mean every useful convention should be elevated into a compulsory registry rule. The common policy layer should contain only what running networks genuinely require for uniqueness, interoperability, security integrity, proof of control, and continuity.

Second, citing deployment in other regions does not settle the matter. Adoption elsewhere is evidence, not authority. AFRINIC should not govern by institutional imitation. The fact that other RIRs have implemented hierarchical naming does not prove that this proposal is the minimum necessary solution for AFRINIC operators, nor that the same operational outcomes could not be achieved through thinner mechanisms.

Third, hierarchical naming does not solve the whole problem being used to justify it. It may identify the ASN holder responsible for creating an AS-SET. It does not prove that every ASN, route, or customer relationship listed inside that set is accurate or authorised. It authenticates the maintainer of the container, not the truth of its contents.

That distinction is not academic. Prefix-generation tools consume the membership of the AS-SET. An authorised maintainer can still publish stale, incomplete, excessive, or incorrect membership. The proposal therefore improves attribution, but it does not transform AS-SET expansion into verified routing authorisation.

Fourth, the suggestion that only those who “live and speak BGP” may raise valid objections is misplaced. Operational experience is valuable evidence. It is not a mandate to dismiss questions about proportionality, institutional scope, implementation risk, or policy design. Technical competence should discipline power. It should not become a credential used to silence scrutiny.

A policy room is not an operators’ guild. Participation is not conditional on passing an informal test imposed by those supporting the proposal.

Fifth, objectors are not required to produce a replacement policy before opposing this one. That reverses the burden. The party proposing a new mandatory rule must prove that:

- the harm is clearly defined and evidenced;
- the proposed rule materially prevents that harm;
- less restrictive mechanisms are inadequate;
- residual risks are understood;
- and the additional registry authority is proportionate.

Possible alternatives have already been identified: source-qualified references, collision detection, stronger validation, clearer tooling, warnings, explicit object provenance, and voluntary hierarchical adoption. Supporters may disagree with those alternatives, but they cannot pretend no alternatives have been raised.

Seun’s suggestion that participants with similar concerns should become co-authors of another policy misses the point. Similar objections do not become invalid because several people independently reach the same conclusion. Participation is evidence, warning, and technical judgment. It is not weakened by agreement.

Nor should the last-call stage be reduced to a demand that objectors prove harm caused by the proposed rule while supporters are excused from proving necessity for compulsion. Last call is precisely where unresolved questions of scope, proportionality, technical sufficiency, and mandate must be tested.

The Internet was not built by making every preferred practice mandatory. It was built through thin common rules, operator judgment, implementation, and adoption. Policy should describe operational reality where a true invariant requires coordination. It should not declare a preferred future and then use the registry to compel it.

The proposal may offer a useful convention. That is not the same as proving a legitimate mandatory rule.

For those reasons, I remain opposed to AFPUB-2026-ASN-001-DRAFT02.

Regards,
Tshepo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.afrinic.net/pipermail/rpd/attachments/20260719/30fb908c/attachment.html>