Summary

  • ICANN’s published contention CSV lists 13 application IDs under the same .agent string, the largest group by row count in that file’s 263 numbered sets.
  • Counted by exact organization-name strings across distinct set numbers, XYZ.COM LLC appears in 100 sets, Intercap Registry Inc. in 54, Radix Technologies Inc SEZC in 41, Charleston Road Registry Inc. in 13 and ECHO Domain Co L.L.C. in 11. The .agent set is included in each total.

A collision group has a portfolio behind it

The first public view of .agent is a list of thirteen applications. The more consequential reading is that the applicant names do not arrive with equal footprints elsewhere in the same record. Five of the thirteen listed entities appear in at least eleven numbered contention sets; one appears in a hundred.

The count is reproducible from ICANN’s published CSV. It contains 891 applicant rows across 263 distinct set identifiers. The .agent rows share set identifier 3 and contain 13 different application IDs. Among those names, XYZ.COM LLC is listed in 100 distinct sets, Intercap Registry Inc. in 54, Radix Technologies Inc SEZC in 41, Charleston Road Registry Inc. in 13 and ECHO Domain Co L.L.C. in 11. OPENAI OPCO, LLC appears in nine. These figures count distinct set identifiers attached to an exact organization-name string; they are not counts of likely winners, customers or domains that will be delegated.

That difference changes what the headline can support. A thirteen-way collision establishes that thirteen applications in the published record seek the same string. It does not establish thirteen independent operating businesses, thirteen product plans or thirteen measures of user demand. Conversely, a company name that recurs across many sets is evidence of a broad application footprint in this dataset, not proof that one corporate group controls every similarly named applicant.

What the record can and cannot connect

The ICANN announcement says the publication includes applicant identities, applied-for strings and the contention-set file, and dates its snapshot to 7 October. The Applicant Guidebook’s contention module explains how identical or variant strings can enter a set and how set membership can change as evaluation proceeds.

Neither source turns a row into a corporate-family chart. The CSV supplies applicant names and application identifiers. It does not, by itself, prove ultimate ownership, common management, financing, an intended registry operator, a product roadmap or commercial readiness. Those are separate questions requiring separate evidence. Treating a legal applicant label as a verified corporate relationship would make the dataset say more than it does.

Lu Heng’s Bill of Rights of Uniqueness Coordination offers a useful governance test: coordination records should preserve uniqueness and accuracy, without being mistaken for a grant of broader authority. Applied here, the contention record is a map of competing claims to a name. It is not a verdict on which claimant deserves the namespace or proof that the namespace has found a market.

The useful comparison is across applicant identities

For investors, registries and infrastructure executives, the practical signal is not simply that .agent drew the largest group in the file. It is that several of its applicants also recur across many other groups, while others appear in only one to four. That uneven footprint can shape capacity planning, diligence priorities and scenario design. It cannot answer who will prevail or whether any applicant will ultimately operate a registry.

A sound market model therefore needs two linked but distinct ledgers. One should preserve the exact applicant name, application ID and each dated contention-set appearance. A second should record any verified parent, affiliate, service provider or operator relationship, with its own source and confidence. Until evidence joins the ledgers, analysts should not merge them.

The distinction matters because an early portfolio map can become an anchor. If a large applicant footprint is treated as proof of control, executives may overstate concentration or mistake application activity for demand. If the identity trail is preserved but marked unresolved, the same count can guide questions without hardening into an unsupported conclusion.

Sources