Summary
- Six IANA protocol-parameter registry pages read on 21 November 2026 each print a registration procedure naming a policy defined in RFC 8126; none of the six prints any per-entry field naming the party that approved a registration, the body that decided it, or an approval act.
- That missing field is not one fact but four: under First Come First Served there is nothing substantive to review, under Private Use the record is withheld by design, under Expert Review a designated expert's approval exists and the entry does not carry it, and under IETF Review the decision survives as a published RFC.
- The durable repair is cheap and unglamorous: one per-entry field naming the route an entry travelled, printed beside the reference these pages already carry.
Open the IPv4 Special-Purpose Address Space registry and pick a row. You get an address block, a name, an RFC reference, an allocation date, a termination date, and five flag columns: Source, Destination, Forwardable, Globally Reachable, Reserved-by-Protocol. Everything about the space is there except the thing a careful reader asks first. Who decided this block belongs on the list?
The page answers a nearby question instead. Its registration procedure line reads IETF Review, and it cites RFC 5736, RFC 6890, RFC Errata 6404 and RFC 8190 (IANA, IPv4 Special-Purpose Address Space). Nowhere in the row does it print a decider, a deciding body, or an approval action. The two dates are not that record either. The registry's own note defines Allocation Date as the date on which the special-purpose address block was allocated (plain-text rendering, registry note), and an allocation date answers when the space was handed out, not who decided it was special.
The rulebook itself is published, and it is short on excuses. RFC 8126 is BCP 26, dated June 2017, and obsoletes RFC 5226 (RFC Editor, RFC 8126, plain-text edition). Its section 4.5 says of Expert Review that review and approval by a designated expert is required, and that the registry's definition needs to make clear to registrants what information is necessary. Section 4.8 says that under IETF Review new values are assigned only through RFCs in the IETF Stream that have been approved by the IESG as having IETF consensus. Section 4.9 limits Standards Action to Standards Track or Best Current Practice RFCs in the IETF Stream. Section 4.4 says something different again: for First Come First Served there is no substantive review beyond checking that a request is well-formed and does not duplicate an existing assignment. And section 4.1 is blunt about a fourth case: IANA does not record assignments from ranges carrying the Private Use policy, and such assignments are not generally useful for broad interoperability.
Four procedures, then, and four different things a missing approval field could mean. Now look at what the six pages actually print.
The Message Headers registry prints Registration Procedure(s) followed by Expert Review and cites RFC 3864. It also prints an Expert(s) line — Mark Baker (primary), Murray Kucherawy (backup) — and publishes entry columns Header Field Name, Template, Protocol, Status, Reference (IANA, Message Headers). The URI Schemes registry replaces the single line with a range table: Permanent carries Expert Review, Provisional carries First Come First Served, Historical carries Expert Review, the CRI Scheme Number field 0-999 carries Expert Review and 1000-60000 carries First Come First Served. Its Expert(s) line names Mark Baker (primary) and Murray Kucherawy (backup) for the registry, and Henk Birkholz and Marco Tiloca for the CRI Scheme Number field. Its columns are URI Scheme, Template, Description, Status, CRI Scheme Number, Well-Known URI Support, Reference, Notes, and its page text states that requests for permanent registration must be preceded by mailing list review, per section 7.2 of RFC 7595 (IANA, URI Schemes).
The HTTP Status Code Registry prints Registration Procedure(s) followed by IETF Review against RFC 9110 section 16.2.1, and publishes Value, Description, Reference (IANA, HTTP Status Code Registry). The DNS Parameters page gives each subregistry its own range table rather than one line for the whole page: DNS CLASSes prints 0 Standards Action, 1-127, 128-253 and 256-32767 IETF Review, 32768-57343 and 57344-65279 Specification Required, 65280-65534 Private Use, 65535 Standards Action, with columns Decimal, Hexadecimal, Name, Reference; the Resource Record TYPEs table adds a per-entry date and labels it Registration Date, with columns TYPE, Value, Meaning, Reference, Template, Registration Date (IANA, DNS Parameters). The TLS Parameters page does the same with ranges: for ClientCertificateType Identifiers, values 0-223 whose Recommended value is Y or D require either Standards Action with Expert Review or IESG Approval, values 0-223 with Recommended N and not transitioning require Specification Required, and 224-255 are Reserved for Private Use, over columns Value, Description, DTLS-OK, Recommended, Reference, Comment (IANA, TLS Parameters).
My own count from those six pages, and the count is mine rather than any registry's: six of six print a registration procedure naming at least one RFC 8126 policy; zero of six print any per-entry field naming the approving party, the deciding body or an approval act; two of six print a named Expert(s) line at registry level only; two of six print a per-entry date field whose own label says registration or allocation rather than approval.
Here is why that last number matters more than the zero. The same silence is four different silences. On the URI Schemes page, a Provisional entry sits under First Come First Served, where RFC 8126 requires no substantive review — the absent approver field is not a gap, it is a complete record of a clerical act. On ranges marked Private Use, the absence is deliberate and published as such: the rulebook says IANA does not record those assignments at all.
On a Permanent URI scheme or a Permanent message header field, Expert Review applies and a designated expert's approval is required — the decision exists, and the entry table simply does not carry it. On the HTTP status codes and the IPv4 special-purpose space, IETF Review applies and the decision exists as an RFC approved by the IESG, so the record lives in a different document entirely. A reader who treats all four as one finding has learned nothing. A reader who can tell them apart has found the exact place where the published record stops.
The TLS range table sharpens it. Values 0-223 with Recommended set to or transitioning from Y or D may be approved by Standards Action with Expert Review or by IESG Approval — two routes, one range, printed side by side. Not one column records which route any individual value took. That is not concealment; it is a design choice about what a registry table is for. But it means the table cannot answer the question a security engineer asks two years later, when a value turns up in a deployed stack and someone wants to know what scrutiny it passed through.
Two of the six pages show what a partial answer looks like. Message Headers and URI Schemes name real people, four individuals across two lines. That is a registry-level statement about who may review. It is not a per-entry record of who did, and reading it as one is the mistake this piece exists to prevent. The URI Schemes page goes one step further and points outside itself to a mailing list review required before permanent registration — the only reviewed page that names a procedural step living in a document it does not publish.
The rest of the argument belongs to judgement rather than to the pages. A protocol-parameter registry is, at its best, a thin coordination layer: its deliverable is an accurate public record that lets strangers agree on what a number means without ever speaking to each other. An authorising act that leaves no trace at the entry is recognition power exercised one level above that record. Naming it is not an accusation. It is a description of where the record layer ends.
What I did not read bounds everything above. No designated-expert roster or appointment page, no per-entry approval record, no mailing-list archive, and no front matter of any RFC cited by these registries was read for this piece, so the field a stranger cannot check is named and the contents of those documents are not characterised. The Specification Required definition used for the DNS CLASSes ranges rests on a secondary HTML reproduction of RFC 8126 section 4 rather than the RFC Editor's own rendering (RFCinfo, section 4 reproduction) and is labelled as second-hand wherever it appears. One further limit: no Expert(s) line appeared in the HTTP page text I inspected, and an absent line in one capture is not a statement by IANA that no expert is designated there.
This desk has read IANA's protocol registries before, and the delta is the point. An earlier briefing followed the appeal clock after a rejected registration request, where the appeal carries a two-month deadline and the decision itself carries no date. This piece starts after the clock has run: it reads a successful entry and asks whether the approving act is printed anywhere at all. A second earlier briefing read an aggregate service tally with no per-case identifier; this one reads a per-entry table with no per-decision field. Same institution, three different shapes of the same question about the claim that a register is a record — the BTW directory entry for the registry operator, IANA in the directory, groups the earlier work.
The recommendation is one column wide. These pages already publish per-entry tables and offer CSV downloads; adding a field that names the route an entry travelled — expert, mailing list, IESG, clerical — would cost an editor's afternoon and would let a stranger trace any value back to a decision instead of to a date. Registries are built to make numbers mean the same thing to people who never meet. A decision field is the cheapest way to keep that promise for the decision too.
Member Briefing
Read the full trend briefing
Members read the full briefing with the source notes behind every finding.
Strategic Circle
A daily briefing in your language, built from changes we have verified.
Join for USD 20/monthCancel anytime.
Members also get the full research library, a daily briefing on the networks they follow and the private forum.






