Summary

  • ICANN, acting through IANA, administers the global pool of autonomous system numbers and allocates blocks of them to five regional internet registries, which distribute numbers locally. Its own network appears in the public record as a single autonomous system number, AS13, registered to ICANN (IANA).
  • The same organisation is named as operator of the L-root server, an anycast service answering root-zone queries. Every other network reaches its number through a regional registry; the records reviewed here show ICANN holding a number and a root letter, and do not show the allocation route either took.

A number with a two-digit name

AS13 is small. Not metaphorically small — small as a number, one of the lowest in a pool whose members run into six digits. It identifies a routing domain, the unit by which BGP speakers decide whose prefixes they will accept from whom, and by that measure ICANN's own network is one of the oldest slots in the addressing scheme rather than a recent addition to it.

The whois record retained for that number is short enough to read in a breath. It reports AS13 as an autonomous system number registered to ICANN (IANA), and describes the number as associated with ICANN's infrastructure and L-root operations (AS13 whois record). Two sentences, one holder, one loose connection to a piece of DNS machinery. Nothing about when the number was issued, nothing about which registry issued it, nothing about who may route it.

That is the first fact of this article, and it is a fact about documentation rather than about intent. The organisation that allocates the global pool appears inside the pool's own public record with a number, and the record says what the number is and who holds it. It does not say how the holder came to hold it.

The chain every other number travels

For everybody else the route is explicit. ICANN's own explainer on autonomous system numbers puts the administrative role with IANA, which manages the global pool and allocates numbers to the regional internet registries, which in turn distribute them to local entities (ICANN on autonomous system numbers). Read that sentence as a chain of custody and it has three links: a global administrator, five regional registries, and the operator at the end who actually runs the network.

An operator on the last link knows the middle one intimately. The regional registry assigns the number under a policy that its members voted on, records the holder in a public registry, prices the membership or the allocation, and governs what happens when the number changes hands. The number is not just a label; it is the tip of a documentary chain that runs back through a regional registry to a global administrator.

ICANN's own number skips the middle link. As recorded, AS13 is registered to ICANN (IANA) — the administrator itself — and no regional registry appears in the entry. That is an observation about what the record shows, not a claim about how the allocation was made. It is possible that AS13 was issued long before the current allocation framework, and possible that some other path accounts for it. What the reviewed records establish is narrower and still interesting: the allocator sits in its own registry as a holder, and the middle of the chain that normally connects a holder to the administrator is not visible from the entry.

The letter L, which is a different object

Confusing AS13 with the L-root server is the easy mistake, so keep them apart. A number identifies a routing domain; a root-server letter identifies a name server that answers queries about the root zone. ICANN holds both, and they do different work.

The IANA root-server list designates ICANN as the operator of the L-root server, L.ROOT-SERVERS.NET (IANA root servers). The L-Root project site describes the service as anycast-based, operated by ICANN to serve root-zone DNS information globally, with many nodes distributed worldwide for stability and resilience (L-Root). A second reading of the IANA page adds a detail with more bite than it first appears to carry: the L-root address 199.7.83.42 sits in address space assigned to ICANN for operating this infrastructure.

Put the two entries side by side and the shape of the institution's own footprint emerges. One number, one letter, one address inside space the record attributes to the holder. Not a portfolio, not a network of subsidiaries with their own allocations — a single registered AS and a globally distributed service running under it. For an institution of ICANN's size and role, that is a strikingly small footprint, and it is a documented one.

What these entries are, and what they are not

Both entries are registry pages. They carry fields, not provisions. The whois record lists an identity; the root-server list names an operator; the project site describes a deployment. None of them is an instrument, and none of them tells you which review, if any, applies to the holdings they describe.

That distinction matters more in this case than in most, because the subject is the institution that writes the rules for everyone else's entries. When a regional registry lists a member's allocation, the member's obligations are set out elsewhere: in policy, in a service agreement, in transfer rules. For ICANN's own entry, the bylaws provisions on community review of budgets and the address supporting organisation's role would be the natural places to look for the equivalent obligations, and no provision-level text was obtained in this review.

So the honest reading of the thinness is bounded. The reviewed pages are short because they are registry pages, not because nothing else exists. Reading them as evidence of an accountability vacuum would be sloppy; reading them as evidence that the allocator documents itself in the same terse format it uses for everyone else is exact, and it is enough to support a question worth asking. Where, in the documentary chain, does the allocator's own holding acquire the obligations that a regional registry attaches to its members' holdings?

The asymmetry an operator can actually see

Consider the view from a network engineer with a regional registry allocation. Her number arrives inside a contract-like relationship: a designated registration, a published holder record, a policy that governs transfer, and a registry that can be asked to correct an error. If she wants to know who controls a peer's number, she consults the same machinery and gets an answer with a name behind it.

Faced with AS13, she gets a holder name and little else. The record tells her that ICANN (IANA) holds the number and that it is associated with L-root operations. It does not tell her which of the five regional registries, if any, stands behind the assignment, what policy governs a future transfer, or where a dispute about the number would be heard. Her own allocation comes with those answers pre-installed; the allocator's does not.

That is the asymmetry worth naming, and it is not an accusation of concealment. It is a structural observation about where documentation ends. A registry that records the world in contractual detail records itself in a couple of lines, and the difference in density is the difference between being inside the review chain and being the place the chain terminates.

Institutional reflexivity, stated plainly

The most defensible frame for this material is reflexivity. The institution that coordinates the global pool is itself represented inside the registry architecture — by AS13 as a holder, and by L-root as an operator of one of the machines that makes the root zone resolvable. It is not outside its own system looking in. It is a node in it.

Reflexivity is not a scandal. Every standards body, every clearing house, every allocation authority has to exist somewhere in the system it administers, and the alternative — an allocator with no network of its own — would be an allocator that cannot test what it prescribes. The interesting question is not whether the institution appears in its own registry. It is whether the entries describing the institution carry the same informational weight as the entries it requires of others.

On the evidence reviewed, they do not. The entry for AS13 names a holder and gestures at infrastructure. The entry for L-root names an operator and a service model. Neither states provenance, review status or the route by which the holding is governed. Those absences are gaps in a particular record, and gaps in a record are fixable in a way that gaps in an institution's authority are not.

What was not obtained, and why it does not collapse the finding

The bylaws provisions on community review of budgets, the address supporting organisation's procedures, and the IANA registry table of autonomous system numbers were not obtained in this review; the retrieval attempts made for them returned no documents. That is a boundary on the article, not a conclusion about ICANN. Nothing here should be read as a statement that those instruments are silent, or that no review applies to the holdings described above.

The finding that survives is documentary and modest. The allocator holds one registered autonomous system number and operates one root-server letter, both visible in public registry pages, both described in a few lines. For a reader who relies on registry records to understand control of numbering resources, that is the useful takeaway: the record of the allocator's own network is legible, small, and thinner than the record the allocator keeps on everyone else.

What would change the meaning of the entry

A registry entry means something different when it carries provenance. If the AS13 record stated the allocating authority, the date of assignment and the policy under which the number is held, the entry would answer the question this article can only pose. If the L-root entry bound the address space to a named review process, the letter would stop being just an operator designation.

Until then, the entry means what it says: the institution that allocates the world's network numbers is one of the networks it numbers. That is a small sentence and a large fact, and it belongs in the record beside the number itself.