Summary

  • Four documents sit behind the directory entry: California corporation number C2114757, a reported 501(c)(3) public-charity classification under EIN 68-0373204, the organisation's own account of its work on internet exchange points and traffic exchange, and one public number-resource record.
  • Each answers a narrow question. Whether a business-registry entry exists. Which federal tax category applies. What the organisation says it does. Whether a name has been associated with a number resource. None of the four establishes registry authority, and none establishes that anyone depends on the organisation.
  • The entry assembles them into a wider claim — a digital infrastructure institution, region Global, performing registry functions for internet number resources on the basis of that single record.
  • Not obtained in this review: articles of incorporation or bylaws, current corporate standing, any Form 990, financial statements, a board or officer list, and any contract, attestation or incident record naming a counterparty.

Packet Clearing House, Inc. is not a mystery. It has a name in a business registry, an entry in a tax compilation, a website written in complete sentences, and a file in a public infrastructure directory. What it does not have, in any document this review could obtain, is an account of how it is governed, what it operates at what scale, or who depends on it. The four things it does have are worth reading one at a time, because the whole problem is that they get read as one thing.

Four documents, four different claims

Start with the directory entry, because that is where four separate documents become a single statement. The entry is filed under the region value Global, describes the organisation as a digital infrastructure institution, and states that it performs registry functions for internet number resources. Its recorded evidentiary basis is one public number-resource record (directory entry Only for Strategic Circle).

That sentence is the object of this article. Not because it is false — no document obtained here contradicts it — but because it is wider than anything underneath it. Behind it sit a corporation number, a tax classification, a homepage and a single numbering record, and each of those documents proves something narrower than the word the entry uses.

The distinction is not pedantry, and it is not a technicality that only registry lawyers care about. Number resources are one of the few places in internet governance where the vocabulary is exact on purpose. A registry is a body that holds authoritative records and is answerable for them. A corporation number proves that a state has a file. A tax classification proves what kind of exempt purpose a revenue authority recognises. A homepage proves what the organisation chooses to say. A number-resource record proves that a name and a number appear together somewhere.

Four documents. Four different kinds of proof. One label.

It is worth saying plainly what these documents are here, because the difference between a document and a description of a document is exactly the kind of gap this article is about. All three external pages reached this review as summarised descriptions dated 4 November 2026, not as full text read line by line. That does not make them worthless — a summary of a registry record is still a registry record with a number on it — but it does mean the claims below are claims about what these pages report, and no more.

What a corporation number proves

C2114757 is the least glamorous document in the file and the most frequently over-read. A third-party compilation of California business-registry data reports an incorporated entity bearing the name Packet Clearing House, Inc. and that corporation number (OpenCorporates record).

That is the entire claim. A record exists, under that name, with that number. What the number does not carry is equally specific. It is not the articles of incorporation, so it does not tell you what powers the entity has or what its stated purpose is. It is not a certificate of good standing, which is a separate document with its own date and its own meaning. It says nothing about directors, officers, decision rights or internal control. Registration is an act of filing. It is not a description of governance.

What a corporation number does do is travel. It is short, stable and citable; it fits in a column; it survives being copied from a register into a compilation and from a compilation into a directory. That portability is why it ends up doing work it was never meant to do. A number that proves an entry exists looks, in a table, exactly like a number that proves an institution exists. The difference lives in the document nobody fetched.

What a tax classification proves

The second document is a tax classification: EIN 68-0373204, recognised as a 501(c)(3) public charity, as reported by a third-party compilation of United States tax data (ProPublica Nonprofit Explorer record).

A 501(c)(3) determination places an organisation in a category of federally recognised exempt purpose. It is a real categorisation with real consequences for how an organisation is treated, and it is not nothing. It is also not a financial statement. It does not disclose revenue, expenditure, reserves, liabilities or donor concentration. It does not name a board. It does not say whether the organisation runs a surplus or a deficit, or whether its obligations are funded a year ahead or a month ahead.

The document that would begin to answer those questions is an annual return. None was obtained in this review. That is a statement about this file, not about the organisation's filing conduct, and the difference matters: a return missing from a review is not a return missing from a revenue authority's records. But the practical consequence is the same for a reader. Where a 501(c)(3) line sits without a return behind it, the financial picture is not thin. It is absent, and the classification is being asked to stand in for it.

This is the most common substitution in institutional due diligence, and the least examined. Charity status is read as reliability because the two words sound adjacent. They are not adjacent. One describes why an organisation is exempt; the other describes whether it can pay for what it promised.

What a homepage proves

The third document is the only one the organisation wrote itself. Its site describes Packet Clearing House as a non-profit providing operational support, research and analysis to the global internet infrastructure community, focused on internet exchange points and traffic exchange (pch.net).

Of the four, this is the richest and the most limited. It is rich because it is the only place where the organisation places itself in a technical landscape: exchange points, traffic exchange, operational support. It is limited because a homepage is a genre with conventions. It announces a mission to people deciding whether to make contact. It does not audit the mission, count the sites, name the staff or publish the contracts.

There is a more interesting thing in this sentence, though, and it sits in the comparison rather than in the claim. The organisation describes its work as support for internet exchange points and traffic exchange. The directory describes registry functions for internet number resources. Those are not the same activity. Supporting traffic exchange is an operational and analytical role in peering; holding registry functions is an authoritative role in number-resource records. A reader who takes the entry as a summary of the homepage has been handed a promotion the homepage never asked for.

That is not a scandal. It is classification drift, and it happens because the words are close enough to pass through an editorial process without anyone stopping to ask which function is actually being described. But the drift is what a relying party inherits.

The label that travels

Assemble the four documents and you get the entry: digital infrastructure institution, region Global, registry functions for internet number resources, one numbering record as evidence.

Two words in that assembly deserve separate attention.

The first is registry. In number-resource governance, registry is not an honorific. It is a function with a specific shape: authoritative records, a defined population, rules for change, and a party answerable for the record being right. A single number-resource record establishes an association between a name and an identifier. It does not establish authority over the identifier, and it certainly does not establish a regime for keeping it right.

The second word is Global. Directory regions are attributes; they exist so that a reader can filter and compare. Put Global next to registry functions, though, and it stops being a filter and becomes something like a jurisdiction. A corporation number is issued by a state. Global is not a state. No registry issues it. The pairing quietly moves a California filing into a planetary role, and it does so without a single document changing.

This is how a label travels further than its evidence: not by lying, but by adjacency. Each field is defensible on its own. Region: the organisation's work is described as global. Function: a numbering record exists. Sector: infrastructure. Read the fields together and a reader receives a categorisation that none of the four documents states.

The cost of that is not abstract. It lands on the next reader, who has less time than the first one.

Where reliance actually sits

Network operators, procurement teams and registry staff screen institutions constantly, and they screen cheaply, because direct verification is expensive and slow. Legal form and directory presence are the cheapest signals available. A corporation number and a charity classification together look like a safe screening proxy, and in most files they are one.

The moment a proxy becomes an expectation is the moment this file stops being four documents. Someone reads the entry, sees corporation, charity, registry, Global, and files the organisation as infrastructure rather than as a vendor. That is a rational shortcut. It is also a decision made on a label rather than on a dependency.

Whether any particular operator currently depends on Packet Clearing House is not established here. That is precisely why the label matters. The risk the entry implies is a continuity risk — a service that other people's networks assume is running — and continuity risk is invisible until the service is not running. An entry that implies the risk without documenting it asks the relying party to hold a position they cannot price.

Two obligations get tangled at this point, and they belong to different parties. The organisation's obligation is to describe its own work accurately on its own pages; nothing obtained in this review suggests it has failed to do that. The relying party's obligation is to test a dependency before building on it, and that is the obligation that gets skipped when the directory has already done the screening. A label that reads like a finding ends the enquiry one step too early.

The boundary of this review

On 4 November 2026 this review spent two retrieval attempts on the organisation's governance and operational disclosure: articles of incorporation and bylaws, an annual return, board records, counterparty records. One attempt failed after a bounded wait because provider capacity was unavailable. The second returned a report that no retrieval capability was available in that session. Both failures were on the retrieving side, and the accurate finding is that these documents were not obtained in this review — not that they do not exist, and not that anyone declined to publish them.

What was not obtained, concretely: articles of incorporation, amendments and bylaws; current California standing documentation; any Form 990; audited or unaudited statements; a current board and officer list with appointment rules; conflict-of-interest, procurement and accountability policies; a service catalogue, architecture, facilities or staffing evidence; any contract, memorandum, attestation or incident record showing a named operator, registry or exchange point relying on the organisation; service-level commitments, performance statistics and continuity arrangements; and the exact relationship between the organisation and the number-resource record its directory entry cites.

That list is long, and its length is the finding. A reader who wants a wide description of Packet Clearing House has four documents and a directory label. A reader who wants to know what to do when the service is unavailable has nothing in this file at all.

What would change the reading is specific and ordinary. One annual return with a revenue figure would turn a tax classification into a financial fact. One charter page naming a board would turn a corporation number into a governance structure. One attestation from a named exchange point describing what happens when the organisation's service is unavailable would turn a label into an operational dependency. Any single one of those would move Packet Clearing House from a name with four documents to an institution with a record, and the directory entry would stop being wider than its own bibliography.

None of those documents is in this review's file. Until one is, the honest reading is narrow: a documented California corporation, a reported federally recognised public charity, an organisation that describes its own work around internet exchange points and traffic exchange, and one numbering record that a directory has used to say more. The label is doing the work. Anyone who needs the work done should ask for the fifth document.