Summary

  • The Internet Society agreed in November 2019 to sell Public Interest Registry, the operator of .org, to Ethos Capital; on 30 April 2020 the ICANN Board declined consent to the change of control, and the sale did not proceed.
  • The written transfer standard, the operative text of that decision and any condition binding the operator afterwards were not obtained in this review, which leaves the episode precedent-like for transaction risk rather than a documented precedent that adjudicates later transfers.

The transaction and the refusal

Public Interest Registry operates the .org generic top-level domain. Its parent, the Internet Society, announced in November 2019 that it had agreed to sell the registry to Ethos Capital, a private equity firm, according to EFF's account of the episode. The proposal drew sustained opposition from non-profit organisations and internet advocacy groups, and the Internet Society's own statement confirms the shape of the deal: consent was sought for a change of control of the .org registry agreement, transferring the registry to an entity controlled by the buyer (Internet Society statement).

The end came at a consent point rather than in a courtroom. As EFF reports it, the ICANN Board formally declined its consent to the change of control on 30 April 2020, which halted the sale, and the decision cited the impact on the .org community and the long-term stewardship of the registry (EFF). The Internet Society acknowledged the decision, expressed disappointment, and said it had viewed the sale as a route to long-term financial stability for its own initiatives (Internet Society).

The transaction also drew a state regulator. The California Attorney General, Xavier Becerra, welcomed the Board's decision and noted that his office had been investigating the transaction; he framed the concern around the effect on the charitable sector and the transparency required in the oversight of an asset of that kind (California Attorney General).

Three accounts, not one test

It is tempting to read those three statements as a single verdict on registry ownership. They are not the same claim.

The seller's account is financial: a charitable parent holding a commercial registry subsidiary, seeking durable funding for its programmes. The advocacy account is about community and stewardship: what happens to a domain space used largely by non-profit organisations when its operator becomes a for-profit acquisition. The regulator's account is about charitable assets and oversight: whether an asset held for public benefit can be converted without a transparency that protects the public interest.

Each may be true. None of them is the legal test. Which of these framings, if any, actually governed the refusal is not something the reachable record establishes, and treating the most quotable one as the operative standard would be an assumption dressed as a finding.

Where the power sat

A change-of-control consent right is a specific instrument. It does not regulate prices or service levels; it decides whether an owner may be replaced at all. Where such a right exists, practical leverage over the transaction moves from the buyer and the seller — who can agree whatever they like — to the party whose approval must be obtained before the deal closes.

That is a structural observation about where authority sits, not a finding about how it was used. It matters because it explains the shape of the episode: a commercial transaction, negotiated between a willing seller and a willing buyer, terminated by an approval that was never the seller's to give. And it explains why the absence of a published test is not a technicality. When the gate-holder's criteria are visible, a party can prepare for them, contest them, or price them. When they are not, the gate-holder can move transaction risk without giving anyone outside the room a reproducible way to anticipate or challenge the result.

The dated boundary of this review

This article cannot tell you what the .org registry agreement says about assignment or change of control. It cannot quote the resolution of 30 April 2020, identify the standard the Board applied, describe the Internet Society's board approval, summarise the Attorney General's findings, or name any condition that survived the collapse of the sale.

That is a limitation of this review, recorded on 18 October 2026, and it should be read precisely. Two consecutive attempts to retrieve those instruments failed at the provider: one returned no retrieval capability and zero sources while stating that this was not a finding that the documents do not exist, and one returned a service error with an unknown outcome. The correct conclusion is that the instruments were not obtained here. It is not that they are absent from the public record, and it is certainly not that no written standard exists.

Reconstructing a clause or a resolution from press coverage would produce a document that no one wrote.

What stewardship has to mean to be auditable

Stewardship was the word used to describe what was at stake, and it deserves a definition that can be checked rather than invoked.

Stripped of sentiment, stewardship of a registry is accountable custody of continuity and public-interest function. It is distinct from title — you can hold the contract without holding the trust — and it is distinct from an unreviewable moral veto, which is what stewardship becomes when it is asserted but never specified. The difference between the two is a chain of five things: a disclosed source of authority, a defined test, a reasoned application to the facts, disclosed conditions, and a review or continuity route for those affected.

Measured against that chain, the .org episode is legible at the first link and dark at the rest. There was an authority to consent, and it was exercised decisively. What test it applied, why the facts met it, what conditions attached, and where an affected party could go are all outside what this record shows.

The finding that survives

A refusal that ends a transaction is not the same thing as a rule that governs the next one. For an outcome to work as precedent — to be something a future buyer, seller or regulator can actually reason from — its authority, its test and its reasons have to be identifiable enough to apply to new facts. Here the outcome is known and the reasoning is not, so the episode functions as precedent-like memory: a data point that raises the assessed risk of any future registry change of control, without telling anyone what threshold will be applied.

That has a cost, and the cost is not borne by the parties who lost. It is borne by the next seller who has to guess, the next buyer who has to price an unknowable approval, and the registrants whose continuity is the stated object of the exercise. The honest reading is neither that the refusal was wrong nor that it was arbitrary. It is that a decisive exercise of consent power left behind an outcome without a published test — and in a system that decides who may own a registry, an unpublished test is itself a governance fact worth recording.

Public Interest Registry remains the operator of .org. Its directory record sets out the entity as this article's subject.