What Happened at Nominet

I was a DNS registry system admin at Nominet, the .uk registry, from 2017 to 2019 — inside the building while the pressure that produced the 2021 revolt was building up. The vote itself came after I left, and that part is public record, linked as usual. Where I am talking about what it looked like from inside, I say so.

The companion post to this one is about ICANN, and it keeps arriving at the same question from different directions: when nobody holds a contract over a registry, what actually makes it behave?

.uk is a good place to answer that, because nobody does. There is no ICANN registry agreement over it, no Specification 6, no compliance function, no external monitoring, and no regulator in the ordinary sense. Ofcom does not run it. The government does not run it.

Its members do. And in March 2021 they used that.

What Nominet Actually Is

Nominet is a company limited by guarantee. It has no shareholders. It has members — registrars and other interested parties who pay a subscription and get a vote — and it was set up to run .uk for public benefit rather than for profit.

That structure is the entire story. A company with no owners to enrich still takes in money, and a registry running a national namespace with no competitor takes in a lot of it. What that surplus is for is a question the constitution answers vaguely and the board answers in practice.

The members are the only check. There is nowt else. Which is fine while the answers line up, and becomes the whole game when they stop.

The Estate, From Inside

Here is the part people outside a registry rarely picture, and it matters for everything after.

Nominet was never just .uk. While I was there the platform carried:

• .uk, the country-code domain (ccTLD), under no ICANN contract at all.
• .cymru and .wales, generic top-level domains (gTLDs) Nominet holds in its own right — and those are under ICANN registry agreements.
• Other people’s gTLDs. In April 2016 Minds + Machines handed Nominet the back end for up to 28 of its strings — .london, .work, .law, .fashion, .cooking among them — which put Nominet in the top tier of registry operators by number of TLDs run. Add dot-brands like .bbc and .bentley.
• ICANN’s emergency role. Nominet is one of ICANN’s Emergency Back-end Registry Operators, the outfits ICANN hands a gTLD to when it takes it off whoever was running it. It joined in 2014, and in December 2017 ICANN used it — Nominet became emergency interim operator of .wed after its operator’s registration data service failed.
• The resolver end. Nominet built and ran Protective DNS for the National Cyber Security Centre (NCSC), the recursive resolver UK public sector bodies query, which declines to resolve names known to be malicious.

Worth one clarification, because “Nominet runs .uk” is looser than it should be.

Nominet runs the .uk registry and most of what sits under it — .co.uk, .org.uk, .me.uk and second-level .uk itself. It has never run all of it. .ac.uk belongs to Jisc, successor to the academic network that named uk in the first place, which has administered and registered names under it since 1996. Through the years this post covers, .gov.uk was Jisc’s too — Nominet only took it over in 2024, and even now the approvals sit with the Central Digital and Data Office rather than the registry.

It goes further than that, and in a direction you would not guess. Jisc also provides the administration for .gov.scot, and for .gov.wales and .llyw.cymru — both of which live inside .wales and .cymru, the two generic top-level domains Nominet holds in its own right. Nominet runs those TLDs. Somebody else runs the governments’ corner of them.

So even inside one country’s namespace the authority is split, and it is split by history and convention rather than by anybody’s design.

So one organisation sat in four different relationships to the name system at once. Outside ICANN’s reach entirely for .uk. Inside it, under contract and continuously measured, for the gTLDs. The instrument ICANN reached for when it needed to take a top-level domain off somebody else. And the resolver deciding what a government department was allowed to look up.

None of that is a contradiction. It is what the structure actually looks like once you stop reading org charts and start reading contracts. Authority here attaches to individual delegations, not to companies.

Two Regimes, One Platform

Now the bit you only see from inside, and it is the reason the estate matters rather than being trivia.

For the gTLDs, Nominet signs the ICANN registry agreement like anybody else. Specification 6 applies — no wildcards, no synthesised answers. So does Specification 10, and ICANN measures compliance with it continuously from outside: DNS resolution, the shared registration system, the Whois service, escrow deposits and correctly signed zones are all monitored against thresholds, and falling through one is a compliance event rather than merely an outage.

Count the zones. On one side, .uk, with no contract. On the other, several dozen gTLDs, every one under a registry agreement and a monitoring probe. The un-contracted zone was outnumbered on its own platform by something like thirty to one.

Running two operational regimes on one platform is painful, so you do not do it. Two escrow processes, two monitoring regimes, two sets of runbooks, two on-call procedures, two answers to the same question depending on which zone the ticket happens to be about — that is how mistakes get made at three in the morning. When ICANN mandates something for the gTLDs, you do not build it twice. You build it once and run everything on it.

Which means ICANN’s requirements landed on .uk as well. Not because ICANN had any authority over .uk — it had none — but because the cheapest safe way to satisfy a rule binding most of your estate is to apply it to all of it, and because deliberately maintaining a split so the ccTLD could do things the gTLDs may not buys you nothing except a second way for the platform to break.

No contract was signed for that. No consultation happened. Nobody in the UK was asked. A requirement written in Los Angeles for generic top-level domains shaped how the country’s own registry ran, by way of a build decision.

For the record, that also made ICANN a real presence on the on-call rota rather than a line in a policy paper. For part of the estate it was a counterparty with a contract, a probe pointed at our infrastructure, and an escalation path.

Selling .uk to Pay for the Rest

The commercial logic of all this was the thing members eventually objected to.

.uk is a monopoly. There is exactly one place to buy a .uk domain, the demand is close to inelastic, and the margin funds whatever the board decides to fund. From 2016 The Register was making the argument out loud that .uk registrants were being overcharged to subsidise the rest of the operation.

Under CEO Russell Haworth, appointed in 2015, Nominet pushed hard into cyber security — the NCSC contract among other things — on the reasoning that a registry sitting on national DNS infrastructure was well placed to sell security services.

That was the direction of travel for the whole of my time there. The revolt did not come out of nowhere in 2021. The conditions for it were laid down years earlier, in plain view of anyone working in the building.

The Charity Went First

In January 2018, while I was there, Nominet withdrew from its own charitable foundation.

The Nominet Trust had been funded by the registry since 2008 — £44 million over that period, £4 million in 2016, £5.4 million in 2017. It gave money to technology-for-good projects. It was, in a fairly direct sense, the public benefit in a public benefit company. It went independent and in May 2018 became Social Tech Trust.

Haworth’s stated reason was that “the grant-giving, single funder model we set up in 2008 was not the most effective route to greatest impact.”

What was announced alongside it was a Cyber Advisory Panel — chaired by Haworth — targeting government and enterprise business, backed by a marketing programme and, in Nominet’s own words, potentially an acquisition.

Put those two next to each other, because they were announced together. The money leaving the building for an arm’s-length charity stopped. A commercial venture chaired by the chief executive started. The members were consulted about neither.

One of them, Andrew Bennett, asked the obvious question at the time: where are all future operating profits going to be spent?

Three years later the membership answered it.

This is also why the paragraphs below are not just my impression. The single largest movement of public benefit money in Nominet’s history went from an independent trust with its own governance to a panel the chief executive chaired, in one announcement, without asking the owners. Whatever you make of the intent, the direction of the money is on the record.

“Profit With a Purpose”

That was the phrase. It was the line for the whole strategy and we heard it a great deal.

The trouble with it was not that it was ambitious. It was that the two halves had come apart. The profit was real, growing, and came from a captive market that had nowhere else to buy a .uk. The purpose was a slide in a deck.

You can hold a company to a slogan like that, and eventually the members did. Inside, it mostly produced the particular tiredness that comes of being told at every all-hands that the commercial push is the public benefit, while the actual public benefit line goes down.

Where the Money Went

Three things ran while I was there, and none of them is DNS for the United Kingdom.

A radio spectrum registry. Nominet built a TV White Space database — a register of which radio frequencies are free to use in a given place at a given time — and got itself approved by the FCC as a database administrator in the United States. The argument was that a registry is a registry, and a company good at one kind of lookup could sell another.

A DNS security product. NTX, threat detection built on inspecting DNS traffic, sold to governments and enterprises. In other words a competitor to OpenDNS, which is to say a competitor to Cisco, entered by a British domain registry.

Driverless cars. Along with drones and the internet of things, held up as the next great wave of things that would need naming and registering.

What happened to them is the answer to whether they were a strategy. The spectrum business was sold to RED Technologies when Nominet refocused. The cyber work produced the technology behind the NCSC contract, which Nominet then lost in 2024. The driverless cars never arrived and neither did the domains they were going to need.

Bidding for Australia

There was a fourth, and it says most about the ambition. In 2018 Nominet bid to run Australia’s registry.

auDA, which administers .au, had put the registry operations out to tender. Nine bids came in from around the world, three were shortlisted, and Afilias took over on 1 July 2018, ending sixteen years of AusRegistry running it. auDA never published who else bid, so you will not find this in the record — but Nominet was in it, and I was there while we went for it.

Hold that against the same year’s other news. In January 2018 Nominet withdrew from the charitable foundation it had given £44 million, on the grounds that single-funder grant-giving was not the most effective route to impact. In the same twelve months it was bidding to run the domain registry of a country on the other side of the world.

It is also worth noticing what the .au tender is, because it is exactly what most people assume .uk must be. auDA can put its registry out to competitive tender and hand it to somebody else, and in 2018 it did. There is no equivalent for .uk. Nominet’s position is not a contract that comes up for renewal — which is a stronger position than Afilias won in Australia, and as such it is worth remembering when weighing how much pressure the members’ vote actually represented. It was the only lever there was.

And here is what was happening to auDA while Nominet was bidding to work for it.

Alongside the tender, the Australian government was running a review of auDA itself. In April 2018 it reported that auDA’s management and governance framework was “no longer fit-for-purpose”, issued 29 required reforms as new terms of endorsement, and put a senior officer from the Department of Communications onto auDA’s board to watch the work. It also said, plainly, that it would transition the delegation for .au to another provider if auDA could not deliver.

That is a national government stating in writing that it will move its country’s domain if the body holding it does not sort itself out. auDA’s own members were mutinying at the same time — a petition for a special general meeting to remove four of its leadership, three years before Nominet’s members did the same thing to five of theirs.

Two national registries, two not-for-profits holding a country’s namespace, both accused of governance failure inside four years of each other. It is not a Nominet quirk. It is what this structure does when nobody is watching it closely enough.

What the Members Saw

As to why those and not others — I will be careful, because I can tell you what it looked like and not what was in anybody’s head. The view widely held among staff at the time was that funding tracked the enthusiasms of the people approving it. I cannot show you a ledger and I am not going to pretend I can.

What I can point at is that three years later the membership’s formal complaint was a more evidenced version of the same suspicion — that a body with no shareholders and a public benefit purpose was spending the surplus from a national monopoly on things that suited the people running it, and that the giving which was supposed to justify the whole arrangement had been cut while it happened.

The members’ complaint was not that diversification is wrong. It was arithmetic. .uk prices rose more than 50 per cent. Charitable and public benefit giving fell, while the organisation was making monopoly margins on a national asset. Operating performance was going backwards despite cost-cutting. Executive pay and bonuses went up through all of it. And member feedback, offered through the channels the constitution provides, went nowhere for years.

A company with no shareholders had started behaving like one with impatient shareholders, and the surplus from the national namespace was paying for it.

The Members Revolt

The campaign was PublicBenefit.uk, organised by Simon Blackler of the hosting company Krystal. Its resolution was blunt: remove named directors.

Nominet’s response is the part worth recording, because it tells you what the organisation had become.

It campaigned against its own members using the members’ own money — emails, phone calls and mailings urging rejection. It refused to engage with the campaign’s substance. When the campaign exercised its right to members’ contact details in order to make its case, Nominet did not send a spreadsheet. It sent a physical package with the details printed across more than 500 sheets of paper, with the email addresses left out.

It also blocked a second resolution that would have installed two qualified caretaker directors, on the argument that it was not legal — and then criticised the campaign for having no succession plan.

On 22 March 2021 it went to a vote. Turnout was 53%. The resolution carried with 52.7%, and five of the eleven board members went:

Haworth resigned hours before the vote rather than lose it. Rob Binns became acting chair.

Here is what I think that pair amounted to, and I am flagging it as opinion because that is what it is.

Wood and Haworth ran Nominet the way a venture capital firm runs a portfolio company. Not as a registry that happens to produce a surplus, but as a balance sheet with an underworked asset bolted to it — a captive monopoly throwing off cash that could be put to better use somewhere with more upside.

Every documented thing above is consistent with that. You take the reliable income and raise its price, because the customers have nowhere else to go. You stop the outgoing that produces no return, which is the charity. You put the difference into a portfolio — spectrum, cyber, a foreign registry, driverless cars — on the theory that one of them comes good. You pay the people directing it at the rate that sort of work commands. And you keep going until somebody with the standing to stop you does.

There is nothing unusual about running a company that way. It is no way to run a public benefit body holding a national asset in trust, because that surplus was never capital in search of a return. It was the thing the whole arrangement existed to produce, and the constitution said so.

The members eventually said the same, in the only language that was available to them.

It is worth being honest about the margin. 52.7% on a 53% turnout is not a landslide, it is a narrow win on a split membership, and the organisation fought it with every resource it had. It still lost.

What Happened Next

Nominet pulled back from the commercial cyber direction and turned towards the registry and public benefit work. Then the numbers arrived.

The core business peaked the year I left. .uk domains under management topped out at 13,348,378 in 2019. By January 2023 that was 11,045,559. By January 2024, 10,688,932. A fifth of the base, gone, and still falling.

The diversification lost money. In its last full year before the reckoning the cyber business unit turned over £12.6 million and posted a £2.4 million loss. That is the answer to whether the projects the surplus paid for were an investment or an indulgence, and it is Nominet’s own figure.

Then the contract went. In 2024 the NCSC retendered Protective DNS — a service handling around half a trillion queries a year — and Nominet lost it. The work went to Cloudflare with Accenture from September 2024, on a deal reported at about £30 million. Chief executive Paul Fletcher said the government had picked a cheaper competitor.

And the backend book churns. In April 2021, weeks after the EGM, MMX sold its portfolio to GoDaddy Registry for $120 million — and the 28 strings that had put Nominet in the top tier of registry operators went with the buyer. .blog had already left for CentralNic in 2019.

It is only fair to say this runs both ways. Amazon moved the bulk of its 54 gTLDs onto Nominet’s platform in 2019, and Microsoft later shifted .skype and .office across from GoDaddy. Nominet wins portfolios as well as losing them.

But that is the point about the business, not a defence of it. Backend registry services is revenue you do not control. It arrives and departs on somebody else’s corporate transaction — MMX did not leave because Nominet ran the platform badly, it left because MMX was sold. Building the finances of a public benefit body on a book of business that can walk out on a Tuesday because its owner took $120 million is a strategic choice, and it was made.

And in the same year, it won .gov.uk.

.gov.uk had been run by Jisc for years, pro bono, under a legacy memorandum of understanding — an arrangement the government eventually concluded was not meeting internationally recognised standards. The Central Digital and Data Office ran a procurement through Crown Commercial Service, Nominet won it in November 2023, and the transition completed on 26 June 2024, timed a week before the general election to keep voter registration out of harm’s way. Jisc’s own registry page now records the date flatly — as of 26 June 2024 it no longer manages the .gov.uk namespace, though it stayed on as a registrar for some customers. Twenty-eight years running a national namespace, ending in a line on a web page.

The stated requirements were resilience, compliance with ICANN’s DNS standards, and meeting the NCSC’s Cyber Assessment Framework.

Read that against the argument earlier in this post. The reason Nominet was a credible bidder for the government’s own namespace is that it already ran to ICANN’s standards — standards it had adopted because most of its estate was contractually bound by them, and which reached .uk because nobody maintains two regimes on one platform. The discipline that arrived sideways, through the gTLD contracts, is what qualified it to run .gov.uk.

So 2024 was not simply a bad year. It lost the biggest contract it had and won the one with the government’s name on it.

Two things about that win are worth noticing, because they are the difference between operating a namespace and holding it.

Government kept the authority. Nominet runs the registry. The Central Digital and Data Office still manages and approves the applications — who is allowed a .gov.uk remains a government decision, not the registry’s. That is not how .co.uk works, where accredited registrars sell to whoever turns up. The technical operation was outsourced. The say over the namespace was not.

And it is a contract. .gov.uk was procured through Crown Commercial Service, which means it has a term and an end date, and the government has already demonstrated exactly once what it does when it decides the arrangement is not good enough: it moved the whole thing off Jisc after twenty-odd years.

So Nominet holds .gov.uk on terms it does not hold .uk on. One can be taken back at the end of a contract by an official deciding not to renew. The other has no contract, no term and no renewal — and the only people who have ever managed to discipline it had to organise a members’ vote to do it.

In March 2024, with PDNS gone and .uk shrinking, Fletcher announced a restructure with up to 70 roles at risk.

And then the line that closes the circle. Fletcher noted that domain pricing “cannot be held at the level set in January 2020 indefinitely.”

.uk price rises were among the things the members revolted over. Five years on, with the diversification wound down at a loss, the flagship contract lost to a cheaper bidder and the registrations falling, the answer being prepared is to put the price of .uk up again.

And the View From Inside Has Not Recovered

The members got their board change. Whether they got the organisation back is a separate question.

As of 25 August 2026 Nominet’s Glassdoor rating stands at 2.9 out of 5 across 97 reviews, with 31% saying they would recommend it, 22% positive about the business outlook and 29% approving of the chief executive. The lowest-scoring category is senior management, at 2.3. Reviewers rate their colleagues and the work highly. What they do not rate is the layer above them.

That is a mediocre score rather than a damning one, and it should be read as what it is: a self-selecting sample on a public site. But the shape of the complaint is recognisably the one the members made in 2021 — strategy nobody can explain, reward flowing upward, and the people running the national registry not being asked.

Different chief executive. Same complaint.

What It Says About Who Governs a ccTLD

Come back to the question at the top.

ICANN could not have done any of this. It had no contract over .uk, no standing, and no mechanism beyond writing a letter. Everything in the ICANN post about Specification 6, compliance and monitoring applied to Nominet’s gTLDs and not to the zone that actually matters to the country.

The UK government did not do it either, and it is worth being clear about why, because the common assumption is wrong. .uk is not awarded on a government contract. There is no tender, no renewal date and no competitor waiting to bid for it. Nominet holds the delegation from IANA, the same way every other country-code registry holds theirs, and nobody hands it out every few years.

What the government does have is a reserve power, and almost nobody knows it is there.

Sections 19 to 21 of the Digital Economy Act 2010 let the Secretary of State act where there is a “serious relevant failure” at a qualifying internet domain registry — a failure adversely affecting the availability or reputation of UK communications, or the interests of consumers or the public. Nominet is that registry. After notice and a chance to make representations, the Secretary of State may appoint a manager over the registry, or apply to court to alter its constitution.

That is a good deal more than ICANN has ever had over any ccTLD. And here is the part worth sitting with: those two sections sat dormant for fourteen years, and were commenced on 6 April 2024 — the same year Nominet lost the NCSC contract and announced 70 redundancies.

I am not going to claim those facts are connected, because I do not know that they are. What is on the record is that the power to put a manager into the .uk registry became live in 2024, and had not been for the fourteen years before it.

And that is only the domestic route. There is a second one, and it is the reason no ccTLD operator anywhere holds its delegation by right.

A country-code delegation can be moved. IANA redelegates ccTLDs, under RFC 1591, ICP-1 and the GAC Principles, and it has done so repeatedly — .kz, .iq, .za, .gd, .gw and others. The GAC Principles hold that each government carries the ultimate responsibility within its own territory for national public policy, and in practice IANA treats the view of the recognised government as a major consideration in any transfer of its country’s domain.

Australia, as above, put that in writing in 2018 — reform or the delegation moves.

So a UK government that decided .uk needed to be somewhere else would not be blocked. It would not be instant, it is not a power exercised by announcement, and the local internet community would be consulted. But the machinery exists, the precedents exist, and the government’s opinion is the heaviest single input to it.

Which puts .uk in a position worth stating plainly. Communications is one of the UK’s critical national infrastructure sectors, and the state treats DNS accordingly — the NCSC has been buying protective DNS for the public sector for years. A registry running critical national infrastructure, whose government can appoint a manager over it domestically and whose word would carry the day in a redelegation internationally, does not hold its delegation as property. It holds it on sufferance, and the sufferance is conditional on not embarrassing anybody.

What disciplined it was a membership vote, narrowly, six years into the problem, after a campaign run by one hosting company that had to be handed its own members’ details on 500 sheets of paper to make its case.

That is a better accountability mechanism than anything ICANN has. It removed the chief executive and the chairman of a national registry, which no ICANN process has ever done to anybody. It is also slow, adversarial, dependent on someone deciding to spend a year of their life on it, and it came within a couple of percentage points of failing.

Which is roughly where DNS governance is everywhere. The mechanisms that work are the ones somebody with resources chooses to operate. .cm wildcarded a whole top-level domain for years because nobody with standing objected. Freenom stopped only when Meta sued. .uk changed course because a hosting company organised a vote.

None of that is governance in the sense the word implies. It is who happened to turn up.

A Free Thought to Finish

Everything above is either sourced or marked as my own experience. This last part is neither. It is what I think, and you are free to disagree with it.

I started at Nominet in 2017, which is nine years ago. Russell Haworth took over in 2015, which is eleven. That is long enough for an organisation to have learned something.

Has it?

On paper, yes. The board that was voted out is gone. The commercial cyber ambitions are wound down. The charitable arm went independent and is still going under its own name. The members used the one mechanism they had and it worked.

Then look at what is actually in front of you. .uk is a fifth smaller than it was in 2019 and still shrinking. The contract that the diversification eventually produced has gone to a cheaper bidder. Seventy roles went with it. The chief executive is briefing that .uk pricing cannot be held at 2020 levels indefinitely — which is the same lever that helped start the trouble in the first place. And the staff rate senior management 2.3 out of 5, with 29% approving of the chief executive. A different chief executive, and recognisably the same complaint.

So my honest answer is that the people were changed and I am not convinced the organisation was.

It is worth noting where Haworth went afterwards, because it is not a criticism and that is exactly why it matters. Before Nominet he had spent fourteen years at Thomson Reuters, in financial data. Afterwards he went to NBS, then Byggfakta, then Acclaro — subscription businesses with professional customers, recurring revenue and pricing power. NBS turned over £45m at £23.6m EBITDA in 2023. Those are good numbers, and getting them is what a commercial chief executive is for.

Which is rather the point, and you do not have to take my word for the characterisation. His own professional profile describes him as a chief executive for private equity backed B2B businesses. That is the job he does, he says so himself, and he is clearly good at it.

He is a commercial operator and he did commercial things. The error was never his temperament — it was putting that temperament in charge of an organisation whose surplus was not capital and was never supposed to be, and then leaving nobody in a position to check it for six years.

Although NBS is worth a second look, because the shape of it is familiar.

NBS sells Chorus, the specification tool most UK architectural practices work in. It is wired into Revit and ISO 19650 workflows, and there is no serious alternative — architects describe it as a vital tool they simply have to keep paying for. One practitioner’s licence went from £1,385 in 2015 to £7,350, which the Architects’ Journal reported as a 400% rise over a decade. There is no small practice rate, so a sole practitioner pays the same per seat as a 200-person firm. The words in the trade press are “rip-off increases” and “well above inflation”, alongside the observation that there is little alternative to paying.

Be careful with the attribution, because the dates do not line up the way you might want them to. Chorus launched in 2018. RIBA sold NBS between 2018 and 2020 for around £172 million, and a former RIBA president has said it should never have ceded control — a sentence with a certain echo to it. Haworth ran the business from October 2021 to October 2024. The sharpest increases reported, and the loudest complaints, are from 2024 and 2025, after he had left, and the chief executive publicly defending them now is somebody else.

So this is not evidence about a man. It is evidence about a shape.

A professional body sells the tool its members cannot work without. The buyer discovers the customers cannot leave. The price rises well beyond inflation, year after year, and the complaints go nowhere because there is nowhere for them to go. That is .uk, and it is what .org nearly became when ICANN lifted the price caps four months before a private equity vehicle bid $1.135 billion for it.

Which is the actual lesson, and it is not a comfortable one for anybody hoping this was about one executive. Captive professional customers, an essential tool, no alternative supplier: that arrangement produces the same outcome whoever is running it, unless summat is standing in the way. At Nominet the members eventually were. At NBS there is nobody — the professional body that might have played that part sold its stake and walked away with £172 million.

I do not think that is really about any individual, either. Put anybody in charge of a member-owned monopoly running a national asset with no regulator watching, and give them a surplus with no obvious owner, and the pull is always going to be towards spending it on something more interesting than the thing that earns it. Running .uk well is not a story you can tell at a conference. Bidding for Australia is.

The correction, when it comes, has to come from members — which means somebody has to give up a year of their life to organise a vote against an incumbent spending those same members’ money to resist it. That happened once, and it carried by 2.7 percentage points on a 53% turnout. Nobody would design an accountability mechanism that way.

The two real backstops sat unused throughout. The Digital Economy Act powers only came into force in 2024. A redelegation has never been seriously suggested by anyone.

Nine years on, the thing I would like to know, and cannot tell from outside, is whether somebody at Nominet today could say out loud in a meeting — we are a non-profit, should we really be spending money on this? — and still be working there a year later.

If the answer is yes, it has learned something. If it is no, then all that changed in 2021 was the names.