Greg Clark stood up in the Commons on 17 January 2011 and called it the biggest transfer of power to local communities in a generation. The Localism Bill, he said, would sweep away the regional quangos and hand planning back to the people who lived with the consequences of it. Councils got a "general power of competence" — the right, in the Act's own strange, almost philosophical phrasing, to "do anything that individuals generally may do." Neighbourhood forums could draw up their own plans, put them to a referendum, and see them written into law. It read, if you squinted, like the most sincere devolution project since Attlee's councils built the postwar estates.

Fifteen years on, the government is giving mayors the power to overrule the councils that Act was supposed to empower.

That is not a contradiction so much as a confession. Read the Localism Act closely and the tell was always there, buried in Part 5 alongside the neighbourhood plans: the London Development Agency abolished, its housing and funding powers folded not into boroughs but into the Greater London Authority, and a new power created for the Mayor to establish Mayoral Development Corporations. The Act that promised to move power down to communities was, in the same clauses, moving it sideways to a regional executive with a single elected head. Localism and mayoralism arrived in the same bill, wearing the same rhetoric, and it was never quite clear which one Whitehall actually believed in.

The neighbourhood-planning provisions were real, and they are worth being fair to. A parish council or a neighbourhood forum could designate an area, draft a plan, take it through an independent inspection, and put it to local people in a referendum; if it passed, it became part of the statutory development plan, binding on the local authority. This is not nothing. In villages and suburbs across England it produced plans that genuinely shaped where housing went, precisely because it gave residents something more binding than a consultation response: a vote that counted. The people who now sneer at localism as pure theatre should have to explain the mechanism away, and mostly don't try.

But a right to be consulted on where the houses go is not the same as a right to decide whether they get built at all, and it was never going to survive contact with a housing crisis the government could no longer pretend not to notice. Neighbourhood plans could shape and steer. They could not, on their own, produce the volume of housing that thirty years of undersupply required. When volume became the political emergency, the architecture built to protect local preference became the thing standing in the way of the target.

Jules Pipe saw this within two years. As Mayor of Hackney in 2013 he said the Act "does not challenge the deep-rooted centralisation in the UK," which from a Labour mayor of a borough that had actually used the powers was not a partisan jab but a field report. The financial constraints were the real story underneath the constitutional one. A council can be handed the general power of competence and still be unable to do very much with it if the Treasury controls the revenue support grant, the borrowing caps and the business-rate retention formula. Localism Act, meet the spending review. The spending review wins every time, because the spending review was never devolved to anyone.

Greater Manchester's experience is the cleanest illustration going. In November 2011 the Combined Authority used the Act to push for a further transfer of powers over transport and housing, real devolution on their own terms. It took three years and an Act of Parliament of its own before that transfer materialised in November 2014. The lesson was not that devolution is impossible. It is that devolution requires central government to keep choosing it, repeatedly, against its own institutional instincts, and a government that has to keep re-deciding to give away power will eventually stop.

Which brings us to the mayors, and to the strongest case for taking power back off councils — the one the Act's own architects would recognise. Councils genuinely have struggled to build. Planning committees defer to local objection with a reliability that would embarrass a weather forecaster; the average council has neither the planning officers nor the political appetite to push through the housing numbers the country needs; and a mayor with a strategic, city-region mandate can plausibly claim a democratic legitimacy that a single ward's nimby majority cannot. If the argument is that somebody has to be able to say yes when fifty councils say no, that argument has merit, and it is roughly the argument Whitehall made for the London Mayor in 2011 too.

The trouble is that this is centralisation admitting localism was never going to be permitted to produce an answer it didn't like. The Localism Act gave communities a voice on the theory that voice would produce consent, and consent would produce housing. When it produced neither in anything like the volume required, the state's response was not to fix the incentives that made councils cautious — the funding formula, the political cost of building, the absence of land-value capture — but to insert a mayor above the council who can be leaned on more directly by the one department that was never devolved: the Treasury.

Here is the honest genealogy. In 1947 the postwar planning act nationalised development rights entirely, and Whitehall decided directly. In 2011 it pretended to give that decision to the parish and the neighbourhood forum. In 2026 it is giving it to the regional mayor, who answers to a bigger electorate but still, in the end, to the same spending review. Each version has looked more local than the last. None of them has moved the actual lever — money, land value, the power to say yes and make it stick — outside the M25. The word localism did the devolving. The power stayed in London, and merely changed desks.

Somewhere in one of the thirteen mayoral combined authorities that will shortly acquire this power, a planning committee is about to spend an evening doing what planning committees do: reading objections about drainage, parking and the loss of a hedge, weighing them against a housing target nobody in the room set, and voting. Under the rules announced this week, that vote will be advisory. The mayor can call the application in and decide it anyway. The committee will still meet. It just won't matter.

The government's case, delivered by the housing minister Matthew Pennycook, is that this is "essential" — his word — to delivering new homes and regeneration, and it is worth taking seriously before it is taken apart. Nearly 180,000 children are living in temporary accommodation in England. The 1.5 million homes target that Labour set itself is, in Pennycook's own admission, "really stretching," and he is right that a smaller target would not be commensurate with a housing shortage this severe. Councils do sometimes say no to things a country that needs homes cannot afford to keep saying no to: a scheme for 400 units blocked over a heritage view corridor, a warehouse conversion killed by eleven letters about traffic. Andy Burnham, who spent nearly a decade as Manchester's mayor before becoming Prime Minister, puts the argument plainly. Challenge "cuts both ways," and sometimes it is the council putting forward something unacceptable to the community it claims to speak for. There is a real housing crisis and a real record of local obstruction, and the instinct to route around both is not a stupid instinct.

But look at what is actually being routed around. The power applies to any scheme over 150 homes, any commercial development over 15,000 square metres, anything above 30 metres — which is to say, precisely the large, contested developments where local scrutiny matters most and is now most easily bypassed. Mayors will also gain "up-front permission," letting building start without the developer applying at all, and a bigger say over how Homes England spends housing money in their patch. The government insists councils will "continue to decide the majority of applications," which is true and beside the point. The majority were never the argument. Nobody was fighting to keep local control of the extension on Mafeking Road. They were fighting to keep it over the 600-unit riverside scheme that changes a town's skyline and its politics for a generation. That is exactly the category now transferred upward.

Here is the part the announcement does not linger on. A council leader who blocks housing can be voted out by the ward that blocked it. A mayor who overrules that council answers to an electorate spread across a city region of two million people, most of whom have never heard of the site in question and will not connect a byelection result to a planning call-in eighteen months later.

This is centralisation wearing a mayoral chain, because a chain is easier to sell than a spreadsheet in Whitehall. The accountability has simply moved one rung up the ladder, out of reach of the people standing directly underneath it.

The honest defence of this is not that it preserves local democracy but that local democracy, at this scale of decision, has been failing and something has to give. Fine. Say that. Do not dress up a transfer of power as an enhancement of it. "Mayors will have a bigger say" is Whitehall for the money now follows mayoral priorities rather than local ones, and a council that disagrees will find its funding as persuasive as its arguments. A consultation on how the call-in power will actually work is due "next week," which means the mechanism by which a resident challenges a mayoral override does not yet exist, and the power to use it does.

There is a control group for this, and it does not flatter the current arrangement. Attlee's government built 200,000 houses a year through local authorities that had genuine housing powers and genuine local mandates, at a moment when the country had far less money and far more urgency than now. Nobody suggested the answer was to strip Salford's housing committee of its authority and hand the decision to the regional grandee. The bottleneck in 1946 was bricks and labour, and the government solved it by building capacity, not by moving the decision to someone harder to find. The bottleneck now, according to every serious account, is largely land supply, planning capacity and viability — problems a call-in power does not solve so much as relocate to a desk with a bigger map and a smaller postbag.

None of which means councils are blameless, or that NIMBYism is a myth invented by developers. Some are exactly as obstructive as the government says. But the remedy for a council that will not decide is a council that can be made to decide — through reformed planning rules, resourced planning departments, a statutory timetable that bites — not a mayor who decides for it. Burnham says challenge cuts both ways. It does. The trouble is that only one direction of that challenge now comes with an off switch for the losing side, and the mayor holds it.

Somewhere in the paperwork for the 13 mayors of England there is now a line that did not exist a fortnight ago: the power to overrule a local council and wave through a development of more than 150 homes, or 15,000 square metres of shops and offices, or anything over 30 metres tall. Housing minister Matthew Pennycook calls this an "essential part of the toolkit" for delivering homes. Nobody has yet said which electorate handed mayors this particular tool, on this particular date, for this particular purpose. The consultation on how it will actually work is due next week, after the powers have already been announced, which tells you rather a lot about the order in which this government does its thinking.

The case for it is not stupid, and it deserves stating properly before it gets buried. Nearly 180,000 children are in temporary accommodation in England. The government has promised 1.5 million homes by 2029 and housebuilders are already telling anyone who'll listen that it won't happen. Sadiq Khan has held call-in powers in London for years, and Pennycook points to schemes his office has "unblocked" as evidence the model works, a claim worth noting even as London's own housebuilding numbers have fallen. Andy Burnham, newly arrived in Downing Street after nearly a decade running Greater Manchester, wants the same tool handed to his successors and to mayors from Liverpool to Tees Valley, and his argument is not unreasonable: local planning committees can behave like small standing armies against anything larger than a bungalow, and somebody with a wider view of housing need, a city-region's need rather than a single ward's, ought to be able to say so out loud and mean it. If the choice is between a council blocking 400 flats to protect eleven parking spaces and a mayor building them, most people who have ever actually tried to buy a flat will take the mayor.

But watch what has happened to the sentence "somebody ought to be able to say so." It has become "the mayor can now say so," as though those were the same claim. Burnham was elected on transport, on policing, on being the man who stood at a podium during the floods. Nobody in Greater Manchester cast a vote last May on whether a 200-home scheme in Trafford should proceed against the wishes of Trafford's own planning committee. The mandate exists in general — it is real, and large, and the Fabian Society is right that mayors carry some of the biggest personal mandates in British politics — but it was never itemised for this. A council planning committee, whatever its faults, answers to people who live on the street in question and can be voted off in May. A mayoral call-in answers to nobody until the scheme is built, the road is jammed, and the next mayoral election is three years away and fought on something else entirely.

The Institute for Government has already called the accountability arrangements around metro mayors "underpowered." IPPR North put it more precisely still: the system is "complex and broad, but yet also manages to be insufficient," a sentence that could serve as the epitaph for most of English devolution since 2016. Neither is describing an absence of scrutiny machinery. Combined authorities have overview committees, portfolio holders, annual reports, the works. What is missing is a clear answer to the only question that matters, which is what happens to the mayor personally when the call-in goes wrong: the tower nobody wanted, built on the mayor's signature, with the mayor three years from facing anyone about it.

There is a precedent for solving housing crises by lifting decisions above local council objection, and it is worth being honest about it. The 1946 New Towns Act created development corporations, appointed rather than elected, answerable to a minister who in turn answered to Parliament every sitting day, and used them to override existing local authorities in Stevenage, Harlow, Crawley, wherever the need was. It worked, and it worked because the line of accountability, while short-circuiting the council, ran straight and visibly to a named minister who could be questioned at the despatch box and sacked if it went badly. What is being built now runs somewhere stranger: through an elected figure with a real mandate, exercised on a matter their election never mentioned, checked by scrutiny committees whose reports nobody reads and by an electorate who will judge the whole four-year record, planning included, in one undifferentiated verdict at the ballot box. Call it neither decentralisation nor centralisation. It is the redistribution of the same fog to a smaller map.

None of which means councils are blameless, or that nimbyism deserves the reverence it still gets from people who own their houses and pulled the ladder up on schedule.

There was a cleaner way to do this. Put mayoral call-in on a ballot the way London's mayoral powers went in 1998, with a referendum attached to the office itself. Instead we have a press release, a threshold in square metres, and a consultation document that will land next week to describe machinery already switched on. Pennycook calls it essential. He may well be right. He has simply not been asked to say, by anyone who could remove him for being wrong, who exactly agreed.

July is supposed to be the good month. Self-assessment taxpayers settle up by the 31st, HMRC banks the windfall, and for four weeks a year the public finances look almost respectable. This July, provisional ONS figures put self-assessment receipts at a record £17.1 billion, the highest July haul since the deadline system began. A Reuters poll of economists had the month coming in close to balance on the strength of it. The government borrowed anyway: £1.8 billion, on the same release. Not a catastrophe by the standards of the year. But a month built to be the exception turned out to be the rule.

The tax take was the best it has ever been in a single month. The story is what happened on the other side of the ledger while nobody was looking. Welfare spending, uprated automatically each April against a formula nobody in Parliament revisits line by line, has been running ahead of forecast. Debt interest, swollen by the unusually large share of UK gilts linked to inflation, moves in lockstep with every CPI print whether the Treasury likes it or not. Public sector pay settlements agreed earlier in the year land as cash out the door now. Put record income against rising, largely automatic outgoings, and July's surplus becomes July's deficit. Nobody stole the money. It simply left through doors that open on their own.

The strongest version of the government's case is worth taking seriously, because it isn't nothing. Britain's index-linked gilt stock is a legacy of borrowing decisions made across four decades, not this Chancellor's design. The welfare uprating formula is statute, not discretion. A Treasury cannot decide, mid-year, to stop honouring either. If you inherited this exact machine, you would get this exact result, and calling that incompetence misunderstands what the word means. The machine was built to run this way.

Which makes it the point rather than the excuse. Net debt now sits at 94.1 per cent of GDP, a level Britain last carried in the early 1960s. Back then the country was paying down war debt through growth it could feel in real wages. Now it sits near a comparable level having built almost nothing to show for it, on a spending settlement nobody consciously chose in any single vote and everybody agreed to in a thousand smaller ones. Every individual link in that chain can defend its own decision. That is how a country ends up here.

The Bank of England held rates at 4 per cent the day before the July figures landed, decided by nine people on the Monetary Policy Committee, none of whom face an electorate. That leaves fiscal policy doing the adjusting alone, because the other lever has independently chosen not to move. Rachel Reeves will stand at the despatch box in the autumn holding a number that four unelected mechanisms built for her without her signature on any of them. The fiscal rule she is judged against, debt falling as a share of GDP within a rolling five years, predates her tenure and will outlast it.

I have refreshed the ONS release page more times this week than is defensible, so I will say plainly: these monthly figures get revised, sometimes by hundreds of millions, once the dust settles. Treat the exact number as provisional. Treat the pattern — record receipts losing to automatic spending — as the thing that keeps recurring regardless of which month you check.

None of the institutions responsible for July's £1.8 billion answer to a ballot. The OBR forecasts the path Reeves is judged against. The ONS counts what actually happened. The MPC sets the rate that decides how much of the gap is interest. The uprating formula, last legislated in a Parliament that has since dissolved twice over, decides how much of it is welfare. A Chancellor stands up in November to defend arithmetic she did not construct, in front of a public that never got a vote on any of its component parts. The best tax month in the country's history still wasn't enough, and the fault line runs through four rooms no elector has ever entered.

Liverpool City Council has worked out that deploying artificial intelligence across its children's social care service will save the authority roughly £150,000 by the end of the 2027-28 financial year. The council's overall budget gap for the same period is upwards of £29m. Put the two numbers next to each other and you have, in miniature, the entire story of how English local government proposes to survive the next three years: with a rounding error, dressed as a strategy.

The scale of the problem the rounding error is meant to address is not in dispute. Councils across the UK are cutting £3bn this financial year alone. Despite a round of funding reforms that ministers described, with the usual confidence, as addressing the sector's structural weaknesses, a £4bn gap remains. Windsor and Maidenhead faces a shortfall of £66.9m in 2027/28 — £419.71 for every resident. Barnet needs close to £80m in government support just to keep functioning as what its spokesperson, with the particular stoicism of a council press office, called "a tightly run ship" caught in "the perfect storm." Scottish authorities are staring at a combined £562m blackhole. Into this gap, councils are pouring chatbots.

The catalogue is not small. Birmingham uses AI-based tools to identify cost savings, though the council has not said how much it has actually saved, which is its own small data point. Oxfordshire uses it to plan school transport routes for children with additional needs, to transcribe social workers' notes during care visits, and to staff a virtual assistant that can, its council leader says, detect when a caller is angry and adjust its tone accordingly. West Berkshire fits cameras to council vehicles to photograph the road surface every ten metres and flag potholes before anyone reports them, which earned it a green rating from the Department for Transport. None of this is stupid. A machine that spots a crack in the tarmac before a resident's suspension does is a machine doing something useful, and a system that stops a social worker typing up the same conversation twice frees up an hour that might otherwise go on paperwork. The case for efficiency here is not fantasy — it is the honest argument, made by people who are not lying to you, that a service starved of money should use every tool available to stretch what remains.

The trouble starts with what "efficiency" is actually being asked to do. Jonathan Carr-West of the Local Government Information Unit put it more plainly than most officials manage: "there is no silver bullet… we have seen some savings being made from AI, but they are, to be frank, pretty marginal compared to the scale of the problem." That is the head of a sector body telling you, on the record, that the technology being rolled out as the answer to a £4bn hole is not the answer to a £4bn hole. It is a way of making the hole slightly smaller while everybody agrees not to mention that it is still there. One trade publication cataloguing these projects — ten different ways councils are using AI, filed under the header of transformation — noted at the bottom, in the interests of disclosure, that the article itself had been written with the help of ChatGPT. Nobody involved seems to have found that funny, which is somehow the most damning detail in the whole file.

The deeper problem is not the size of the savings. It is what the savings are made from. Andrew Reece of the British Association of Social Workers has flagged the use of AI to draft care assessments — the documents that determine what support a vulnerable adult or child is entitled to. "This is stepping into the area of making decisions," he said, "and we don't think that's appropriate." He is right, and the distinction he is drawing is the one this whole rollout keeps blurring: a tool that transcribes a conversation is a stenographer, but a tool that drafts the assessment is deciding, in the first instance, who gets help and how much. Nobody stood for election promising that. No councillor voted for it in a chamber where a resident could watch and object. A procurement officer signed off a contract with a vendor whose product was demonstrated at a conference, and the decision about who receives care in 2027 travelled from that meeting room into a family's life without ever passing through anything a citizen could observe, let alone unseat.

This is the pattern the column keeps finding wherever it looks. A decision too politically expensive to make in public gets relocated somewhere no one can vote on it. Rate rises are unpopular, so the gap is closed instead by a software licence, and the licence's failure to close the gap becomes, conveniently, nobody's fault — not the council's, which can point to the innovation; not the vendor's, which sold a pilot rather than a cure; not the minister's, who announced a funding settlement two years ago and has moved on to the next brief. Attlee's government, with the country broke and the shops empty, built the health service and 200,000 houses a year because a minister was prepared to be the person who decided and the person who answered for it. Liverpool's council has automated £150,000 of a £29m problem and called it modernisation.

The technology is not the scandal. What it is being asked to disguise is: that somewhere between the funding settlement and the care assessment, a decision worth making in public is being made instead by a system nobody elected, procured by an officer nobody can sack, to manage a shortfall a minister has already declared solved.

Here is the number nobody at Westminster wants to sit with for longer than a news cycle: £164,347. That is what Lord Sainsbury gave Andy Burnham in the four months before he became Prime Minister, and it is not, whatever the Labour press office would like you to hear, a donation in the way you or I understand the word. Filings to the register of interests describe it as covering office space and "contractors assisting Mr Burnham with policy and operations." That is not bunting and a minibus to the count. That is staff. People employed, on a billionaire's money, to write the policy of the man who is now running the country.

The total declared comes to £345,000, collected between March and Burnham's arrival in Downing Street, and on its own that figure tells you almost nothing — political campaigns cost money, and Britain has decided, sensibly enough, that the public purse shouldn't cover the bill. Sacha Lord put in £35,000, on top of everything he already gave as Burnham's night-time economy adviser in Manchester. Gary Lubner, the Autoglass founder, gave £25,000. Charlie Parsons Creative gave £50,000. ThinkLabour, the rebranded Labour Together, gave £22,186 for staffing and IT. Ernst & Young seconded a staff member worth £16,500. Lord Blunkett, Burnham's mentor since the 1990s, put in £2,000 for the Makerfield by-election and another £10,000 later. Four trade unions chipped in £20,000 between them for mayoral campaigning. Every penny of it is declared, dated, and traceable, which is precisely the case for not worrying about any of it.

That is the argument worth taking seriously before you dismiss it: Britain's political funding rules exist and Burnham followed them. New MPs must register financial interests and gifts from the preceding year within a month of election; he did, promptly, down to the £3,800 of Brit Awards tickets from Sony and the £2,520 Glastonbury package from the festival itself. The Electoral Commission has no complaint. A Labour spokesperson's line — "it's right that political campaigns are funded by the generous support of party donors, rather than via the public purse" — is a fair description of the system as designed. Nobody broke a rule. The disclosure worked exactly as intended.

That is the problem, not the defence. Disclosure tells you the amount and the name. It does not tell you what £164,347 of "policy and operations" support actually produced, which papers got drafted, which lines survived into a leadership manifesto, which didn't. Sainsbury's staff were embedded inside Burnham's policy operation months before he held office, funded by a man whose fortune runs into the billions and whose interest in Labour's direction has never been a secret. You can know every figure in the register and still have no idea whose judgment shaped the words that became government policy. Transparency answers "who paid." It has nothing to say about "who decided," which is the only question this column ever really asks.

And Sainsbury is not a fresh face testing the waters. He funded New Labour through the 1990s and 2000s on a scale that dwarfs this — reportedly north of £15 million over the years — and was rewarded with a peerage and a decade as science minister under Blair and Brown, a job he held without ever facing a voter. Three decades on, the same donor is funding the policy staff of a different Labour Prime Minister. One man, two Prime Ministers, thirty years apart, and the voters were consulted for neither appointment.

None of this requires a conspiracy. Sainsbury may want nothing more sinister than a Labour government that governs competently, and Burnham's policy staff may have produced entirely defensible work. But the mechanism matters more than the motive. A donor with the money to fund a leadership candidate's policy shop gets a seat at the table that a nurse in Wigan, a shopkeeper in Bolton, or a Unison branch secretary funding £5,000 at a time simply cannot buy. The register doesn't fix that imbalance. It photographs it, files it, and moves on.

The honest response to all this is not outrage, which the numbers don't need, but a plainer question: why does the country accept a funding model in which the people who write a Prime Minister's policy in opposition are paid by whoever can afford to hire them, rather than by the public whose government those policies will run? State funding of parties is unpopular, expensive, and has failed at every previous attempt to legislate it through. But at least it would put the bill, and the decision, somewhere a voter could find it.

Right now the decision sits with Lord Sainsbury's chequebook. It is fully declared, entirely legal, and completely beyond anyone's ballot.

Somewhere in Clacton-on-Sea this week, a Personal Independence Payment claimant either opened a brown envelope or didn't, because most people stop reading DWP letters until the money stops arriving. Nigel Farage's own constituency has one of the highest concentrations of PIP claimants in the country. His party's new welfare policy proposes reassessing every one of them, along with 2.89 million others nationally, through a system called a Disability Needs Assessment that does not currently exist, has no published criteria, and has been costed by nobody outside Reform's own press office.

Say the £50bn figure the way Reform says it and it sounds decisive. Say it the way the arithmetic says it and it sounds like this: the Office for Budget Responsibility puts total welfare spending at 10.6% of GDP this year, and once you strip out the state pension — which Reform, sensibly for a party courting the over-sixty-fives, has ruled out touching — the £50bn works out to one pound in every four of what's left of the working-age budget.

The strongest case for Reform is real, and it deserves saying plainly rather than waved at. PIP claims have gone from 2.4 million before the pandemic to four million now, a rise no minister of any party has satisfactorily explained, let alone budgeted for. Times columnist Fraser Nelson made the sharper point on air this week: the seats with the most PIP claimants and the seats Reform needs to win are frequently the same seats, which means telling those voters this is unsustainable is a genuine act of political nerve. That much I'll grant them.

What I won't grant is that £50bn was built up from a costed case at all. It reads like a number chosen for how it sounds in a headline. The two halves come apart under separate pressure.

The £21bn from health and disability benefits rests entirely on that unbuilt assessment being "substantially tougher" than the current one, in the IFS's phrase, without a word on what tougher means: which conditions, which thresholds. The IFS's verdict, delivered without adjectives because economists don't need them, is that there is "relatively little detail" about what the plan entails. We have been here before: previous governments toughened work capability assessments and the promised savings arrived late, or small, or not at all, because reassessing millions of disabled people is expensive and slow, and the people doing the reassessing have a habit of finding the claims were valid. Reform's plan assumes the process will move faster and cost less than it has for any government that tried before.

The £20bn from excluding foreign nationals rests on something else entirely: not a mechanism but a hope. Roughly 1.3 million non-UK citizens currently claim Universal Credit, 650,000 of them not in work, and stripping their entitlement would breach the UK's 2019 withdrawal agreement with the EU. That agreement cannot be unwound by a Reform government's say-so. It requires the agreement of the European Commission and twenty-seven member states, none of whom face a British electorate and none of whom owe Robert Jenrick or Danny Kruger a renegotiation. Several might reasonably ask what Britain proposes to do about the reciprocal claim, since the same agreement protects the benefit entitlements of British pensioners retired to the Costa del Sol. Kruger's own account of the policy on the record does not mention them once.

Here is the part that should embarrass everyone nodding along at the conference hall: £20bn of a £50bn welfare policy is not a domestic decision at all. It is a request submitted to Brussels, Berlin, Paris and twenty-four other capitals, none of whom has the slightest electoral incentive to grant it, and all of whom would be entitled to reply that Britain signed this deal five years ago with its eyes open. No British voter can deliver that £20bn by marking a ballot paper. Not Farage, not a referendum. It sits with people the claimant in Clacton has never voted for and never will.

Which leaves the £21bn resting on an assessment nobody has designed, applied to millions of people whose claims already went through one gatekeeper and survived it, priced by a party that has not published so much as a draft methodology. Add it up and the headline number is two different kinds of fiction stapled together: one that assumes a bureaucracy will behave better than every version of it has before, and one that assumes twenty-seven governments will do Britain a favour it hasn't asked nicely for. The Clacton claimant will read the letter when it comes. It just won't be the one Reform is currently promising to send.

Danny Kruger, Reform's work and pensions spokesman, went on BBC Breakfast this week to explain how his party would fund £50bn of welfare cuts, and somewhere in the answer he conceded that Brussels might retaliate by cutting off British pensioners in Spain and France. "I'd understand," he said, "if the Europeans decided to apply the same principle." A shadow minister, on the record, pre-agreeing to the collateral damage of his own policy. Nobody made him say that. He just did.

Start with the number BBC Verify already did the sum on, because it's the one Reform would rather you didn't dwell on. Total welfare spending this year is £353bn. Take out the state pension, which Reform has promised not to touch, and you're left with £207bn. Reform wants £50bn of that gone by 2030. That's not a trim. That's a quarter of everything left once you've protected the largest, most electorally sacred item on the list.

Where does it come from? £21bn from reforming disability and health-related benefits: abolishing Personal Independence Payment for working-age claimants and replacing it with a Health Security Allowance that pays cash only to the "severe, enduring, high-risk" cases, which Reform's own document says will touch 2.89 million people. Another £4.8bn from switching to a stingier inflation measure for uprating. £2.8bn from a fraud crackdown of the kind previous governments have tried and mostly failed to bank. And then the largest single idea in the whole document: stripping most benefits, including housing support and free childcare, from foreign nationals — a group that includes EU citizens with settled status, people who moved here relying on a treaty Britain signed in 2019.

That treaty is the problem. Reform can't touch those entitlements without unpicking part of the Withdrawal Agreement, which means renegotiating with the EU before a single pound of the promised saving exists. The policy document doesn't say when this negotiation starts, what Brussels wants in return, or what happens if it says no. It just assumes yes, books the saving anyway, and moves on to the pensions triple lock, which is the whole reason for this exercise, since its cost is heading for £15.5bn a year by 2030 and someone has to find the money.

Here is the part that should worry a British pensioner in Alicante rather more than it worries a columnist in Manchester: reciprocity runs both ways. If Britain unilaterally strips entitlements from resident EU nationals, the EU is entitled — legally, not just rhetorically — to consider doing the same to the roughly 300,000-plus UK state pensioners living across the continent, plus everyone relying on reciprocal healthcare that sits on the same legal scaffolding. Kruger didn't deny this. He said he'd understand it. The mechanism the whole £21bn foreign-national saving depends on, conceded live on air by the man meant to be selling it.

Reform's diagnosis of the underlying disease isn't the fantasy here. Welfare spending is 24% of everything the government spends, PIP claims have risen from 2.4 million before the pandemic to four million now, and the Centre for Policy Studies and the IEA — not exactly hotbeds of soft-left sentiment — both say no party is being straight about the trajectory. Robert Jenrick calling the current settlement "suicidal empathy" is grim phrasing for a system supporting genuinely disabled people, but the underlying claim, that nobody has been honest about the cost curve, is unfortunately true. It's a real case, and it deserves better than the arithmetic bolted onto it.

What isn't real is that arithmetic. The £1.85bn Reform proposes spending on CBT, physiotherapy and employment support is new spending, not a saving, and it depends on clinicians who, as Reform's own councillor admitted to TalkTV this week, "aren't available." The 20-hours-a-week community work scheme is meant to save money by moving people off benefits, except someone has to run it, and that someone is cash-strapped councils, who will need funding to do it — a cost the document doesn't appear to net off anywhere. The National Insurance cut meant to keep the foreign-national ban "cost neutral" for employers is itself an unfunded promise stacked on an unfunded promise.

None of this is unusual for an opposition costings document eighteen months from an election. Every party's manifesto maths gets loose around the edges. What's unusual is the scale of the load-bearing assumption underneath it: that a foreign government will agree, on Reform's timetable, to hand back sovereignty over a treaty it negotiated in good faith seven years ago, purely so a British party can hit a spending target it announced on a Sunday in the Telegraph. Nobody in Brussels has been asked. Nobody in Westminster gets to vote on whether that negotiation happens, on what terms, or with what leverage, because it hasn't started, and the £50bn figure exists regardless. The saving is booked. The negotiation that would produce it is not.

It's the same trick this column keeps finding under different wrapping paper: a number treated as a decision, when it's actually a hope with a spreadsheet attached.

Somewhere on the Costa del Sol there's a retired plumber from Preston who has just discovered that his pension is a bargaining chip in a negotiation his own government's opposition hasn't opened, can't guarantee, and didn't think to mention to him first.

In the small hours of 26 May 2014, before the last European election results had even been declared, Nigel Farage stood in front of a scrum of cameras and announced a political earthquake. The broadcasters were still totting up regional counts. The projections had UKIP somewhere between 27 and 28 per cent, a number nobody in the studios seemed entirely willing to believe, and the panels were already reaching for the word "protest" before the final figures existed to be protested against. Farage did not wait for the caveats. He gave his victory speech to a result that had not yet happened, which turned out, when the counting finished, to be roughly accurate.

The record, once it settled, was this. UKIP finished on 27.5 per cent of the national vote and 24 seats, up from 13 in 2009 — the first time since the 1906 general election that a party other than Labour or the Conservatives had topped the popular vote in a national poll, and the first time since December 1910 that either of them had failed to win the most seats. In Farage's own patch, South East England, his vote had nearly doubled in five years, from 440,002 in 2009 to 751,439. The same month, in council elections held alongside it, UKIP took more than 160 seats it had not previously held. None of this was a rounding error.

The mainstream response, at the time, was to file it under mood rather than movement. It was Eurosceptic sentiment finding an outlet; it was a mid-term protest against a coalition nobody much liked; it would fade once the stakes were general-election stakes rather than European ones. This was not an unreasonable read. Through 2015, the polls did chart a decline in UKIP's support, and commentators noted, accurately, an unusual lack of energy from Farage himself, wondering openly whether the man was still fit for the fight he had started. If you had stopped reading the papers there, in the spring of 2015, the protest-vote theory would have looked vindicated.

But the wave had already done its structural work before it receded. In October 2014, Douglas Carswell defected from the Conservatives and forced a by-election in Clacton, which he won as UKIP's candidate — the party's first parliamentary seat, achieved not through the arithmetic of proportional representation but by taking an actual Conservative MP and his actual Conservative seat. Mark Reckless did the same at Rochester weeks later. A movement that was supposedly a summer flu now had two named members of parliament, elected the ordinary way, by name, in a first-past-the-post system that is specifically hostile to insurgents. That is a harder thing to argue away as sentiment.

Harder still to argue away was what the 2014 result did to a Conservative Party that had already, in January 2013, promised an in-out referendum on Europe. Cameron's pledge predates UKIP's Euro triumph; what the triumph removed was any route back from it. A governing party that has just been beaten into third place in a national election by a single-issue outfit cannot then quietly retire the manifesto commitment that outfit exists to enforce. The referendum happened because it had to happen, and when it was fought, in 2016, it was fought substantially on Farage's terrain — immigration as the defining axis, the "Breaking Point" poster as its most notorious artefact, a campaign whose emotional register he had spent two decades assembling out of Dulwich, the trading floor, and fourteen years inside a Conservative Party that had finally lost patience with Maastricht.

There is a useful piece of arithmetic buried in that "first time since 1906" line, and it is worth sitting with rather than skimming past. The last party to break the Labour–Conservative duopoly in a national poll was Labour itself, then an insurgent representation committee treated by the establishment of its day much as UKIP was treated in 2015 — a protest, a phase, something the Liberals would absorb once the excitement wore off. Within two decades it had not been absorbed. It had replaced the Liberal Party as one of the two parties that mattered. So the precedent for what happens after a breakthrough like 2014's is not neutral. It is a coin with only two faces, absorption or replacement, and Britain's one recorded flip of that coin came down on the side nobody in 1906 expected.

Which brings the argument to Clacton again, in a different decade. Farage resigned the seat in July 2026 amid a financial scandal — the kind of story that, for most politicians, ends the career rather than merely interrupting it — and pledged to win it back in a by-election. He has now won it back, per the reporting, and the two front benches appear to have reached, once more, for the same file marked protest, personality, phase. It is the identical bet Westminster made in 2015, priced in real time: that the wave recedes if you wait it out and decline to argue with it. That bet cost the Conservative Party a referendum, and arguably its coherence, the last time it was placed.

Clacton has now sent Farage to Westminster twice in twelve years under two different party names, and nobody at either front bench has revised the theory that this is a phase.

The confirmation came the way these things usually come now: not from the despatch box, not in a written statement laid before the House, but to a broadsheet, then to the BBC, via "a well-placed military source" who will remain nameless for as long as it suits everyone above him. British-made drones, built by two companies whose names we are not told, have struck targets on the Russian mainland for the first time — refineries, warehouses, a logistics depot belonging to Wildberries, Russia's answer to Amazon. Economic damage is put at £35bn. Russian officials have confirmed the UK components themselves, which is one way of finding out your government has expanded a war: the other side tells you.

Nobody voted for this, in the narrow sense that matters. Nobody in Parliament debated it, because nobody in Parliament was told. The Ministry of Defence's position, delivered by spokesperson rather than minister, is that support for Ukraine is "unwavering," a word doing a great deal of load-bearing work for a sentence that commits to nothing checkable. In April it announced 150,000 drones this year, "cutting-edge Stone Cloak devices" and "innovative low-cost cruise missiles," a package trailed with the pageantry normally reserved for a spending review. What it did not announce, then or since, is that the equipment would eventually be used to hit targets deep inside Russia itself, a threshold every government since the Cold War's coldest years has treated as qualitatively different from arming the defender at his own border.

The case for silence is not difficult to construct, and it deserves to be made properly. Ukraine is defending itself against an illegal invasion, and the war does not pause for select committee sessions. Disclosing exact capabilities in advance hands Moscow a targeting brief. Allies coordinate this sort of thing through channels that exist because open debate would compromise the operations being debated; that much is not controversial. Cabinet government has never conducted war by referendum, and Attlee did not put the bomb to a vote either. There is a serious case that some decisions are executive by nature and always have been.

But that case has a shape, and this does not fit it. Its shape is retrospective account: not permission sought in advance, but explanation offered once the fact is established and the tactical value of secrecy has passed. The Falklands campaign ran without a parliamentary veto over each sortie; even so, the Commons debated the deployment, recalled itself for it, and Thatcher stood up afterwards to answer for what had been done in the country's name. What happened here is different. The public learned of an escalation from a Sunday newspaper's sourcing, with the Ministry confirming the fact only once denial had stopped being useful. Operational security has rarely looked like a minister who never has to stand at the despatch box and answer the question a British weapon striking Russian soil obviously invites: what is the escalation threshold, and who decided we had not crossed it.

Because someone did decide. Somewhere between the April drone package and the strikes on Wildberries' warehouses, a judgment was made — that arming Ukraine with weapons able to hit Russian soil was worth the risk of retaliation and a wider war. It was never put to Parliament.

That judgment sits somewhere in the machinery connecting the Ministry of Defence, the Foreign Office and Downing Street, and it is not unreasonable to assume it was correct. The trouble is that "not unreasonable to assume" is doing the work parliamentary scrutiny used to do, and doing it worse, because it cannot be tested, only trusted.

This is the pattern the column keeps finding wherever it looks, and finding it in defence policy is more alarming than finding it in a housing quango, because the stakes are a fuel depot in Rostov today and, in the worst case, considerably more. Decisions that could reshape Britain's exposure to a nuclear-armed adversary are being taken by the same accountability-free method as decisions about planning appeals, settled below the sightline of any elected body, announced only once the fact could no longer be denied. A spokesperson's statement is not scrutiny, and a military source confirming to the BBC what his own department declined to announce is precisely how a government keeps the serious question — is Britain now a co-belligerent in a war with a nuclear power, and to what degree — permanently off the record it is supposed to keep.

None of this settles whether the drones were the right call. It shows only that nobody has had to argue they were, in public, to people who could vote on the answer. Britain has managed, in the space of one summer, to expand materially what it is doing in a European war with a nuclear-armed state, without a single division in the Commons. Churchill went to Parliament to ask for the tools to finish the job. His successors, it seems, would rather let the tools speak for themselves.

At Clacton Leisure Centre in the small hours of 14 August, the returning officer read out 22,239 votes for Nigel Farage and the room did not erupt. Farage was not there to hear it — he had told supporters beforehand that he would skip the stage line-up, the odd little ritual by which British democracy makes even a landslide look like a school photograph. Count Binface, the comedian in the space helmet, came second with 26.9 per cent. The main parties were not present in any capacity at all, because none of them had stood a candidate. A record 34 people did, beating the 26 who turned out for David Davis's civil-liberties stunt in Haltemprice and Howden in 2008. Farage's share rose 17 points on 2024, to 63.34 per cent, on a turnout of 44 per cent that would have embarrassed a parish council. This is what a coronation looks like when nobody bothers to contest the crown.

The honest case for treating this as a real verdict deserves stating before it is dismantled. Nobody made Labour, the Conservatives or the Liberal Democrats boycott Clacton — they chose to, calling it a stunt, and a party that refuses to compete forfeits the right to complain about the result. Farage did increase his vote total by 1,014 on the general election, in a seat where turnout collapsed by fourteen points, which is not nothing. He framed it, not unreasonably, as the voters being asked a direct question — "should the judges of my actions be the people of Clacton" — and getting to answer it without three other parties shouting over them. If democracy means anything it means the electorate gets to render a verdict even when the accused chooses the courtroom.

But look at what the verdict was actually on. Farage resigned as MP last month "in response to scrutiny of his financial backing" — the BBC's phrase, not mine — specifically a Parliamentary standards investigation into a £5 million gift from Christopher Harborne, a Thailand-based crypto donor. That investigation was paused, as procedure requires, the moment he stopped being an MP. He then stood in the by-election he had triggered by resigning, faced no serious opposition, secured fewer actual votes than Giles Watling managed as a Conservative in 2019, and reopened the door to Parliament — at which point the standards probe reopens too, because that is what the rules say happens to a sitting MP.

Notice the sequence. Scrutiny intensifies; the scrutinised person resigns, converting a standards question into an election question; the election, deliberately unopposed by anyone capable of contesting it, produces the only kind of "increased majority" achievable against a comedian in a space helmet; and the scrutiny resumes exactly where it left off, only now framed by its subject as something the electorate has already settled. Andy Burnham, asked about it on Radio 5 Live, gave the honest answer available to a prime minister with no lever to pull: the voters "have come up with their answer," but the investigation "needs to be resolved... that's incumbent on all of us, including Mr Farage." Which is Whitehall's polite way of admitting that a Commons vote and a Parliamentary Commissioner's inquiry are answering two entirely different questions, and that only one of them has been settled this month.

Here is the audit trail, followed properly. Who decides whether Nigel Farage's £5 million donation was declared correctly? Not the electors of Clacton — they were never asked that, and the ballot paper had no space for it. It is the Parliamentary Commissioner for Standards, an office few voters could name and none can remove, working through a process that can be paused by the simple expedient of resigning and un-resigning. The Conservatives called the whole exercise a "farce" designed to "dodge scrutiny," which is rich coming from a party that spent a decade perfecting the same manoeuvre with different names on the door, but the description is accurate regardless of who's making it. A mechanism that can be switched off by the person it investigates, simply by timing his own departure and return, is not an accountability mechanism. It is a courtesy.

Clement Attlee took office in 1945 over a Britain that was, in the most literal sense, bankrupt — rationed, bombed, owing America money it would still be repaying in 2006 — and built 200,000 houses a year and a health service anyway, because the machinery for being held to account was crude, slow, and largely inescapable: you stood, you won or lost, and the results were the results. Nobody in 1945 could resign a seat specifically to suspend an inquiry into himself and then reopen it on his own schedule. The instruments we have built since are more sophisticated than anything Attlee had, and they have somehow made it easier, not harder, for a man under investigation to control the tempo of his own reckoning. Clacton gave Farage a mandate. It did not, and could not, give him an alibi. Only the Commissioner for Standards can do that — and nobody elected her either.

Nigel Farage won Clacton on Thursday with 63% of the vote and did not turn up to hear it declared. He was, his office said, unwilling to be "humiliated by nobodies" — a fair description, perhaps, of Count Binface, who came a distant second in a bin costume, but an odd one for the office that will decide whether the winner keeps his seat. That office is not on any ballot. It is Daniel Greenberg, the Parliamentary Commissioner for Standards, and the moment Clacton's result was confirmed, his suspended inquiry into Farage's finances switched back on.

The sequence is worth setting out plainly, because Reform's messaging depends on you not doing so. Farage resigned his seat on 8 July, calling the commissioner's inquiry into an undeclared £5m gift from the Thailand-based crypto financier Christopher Harborne an "establishment hit job." He framed the resulting by-election as a verdict: let Clacton judge him, not Westminster's tame officials. Clacton judged him, handing him 22,239 votes and a majority of 12,784, up from 8,405. The inquiry, which Farage was reportedly told wrongly would end rather than pause, restarted within hours, widened now to cover money and favours from his friend George Cottrell, an aristocrat convicted of fraud in the United States a decade ago. Clacton was never asked about any of that.

Farage's supporters have a point that isn't mad. An electorate, fully informed of both the Harborne gift and the Cottrell arrangement, both extensively reported before polling day, chose to increase his vote share anyway. That is a real mandate for something. If the charge is that voters were duped, the charge fails: Clacton knew exactly what it was electing. And there is a genuine democratic case buried in that argument: an unelected commissioner sitting in judgment on an elected MP's conduct is exactly the kind of arrangement this column normally distrusts, a decision made by a body no elector put there.

The honest wrinkle, worth admitting outright, is that nobody can rerun the count with Greenberg's eventual finding already published. We don't know what 63% becomes once the report lands, only that it doesn't become zero.

The trouble is that the mandate and the inquiry are not answering the same question, and Farage's entire rhetorical strategy depends on collapsing them into one. Clacton was asked, in effect, "do you want this man as your MP." It was not asked, and could not be asked, "did Nigel Farage declare a £5m gift from a foreign national in accordance with the rules governing every other member of the House of Commons." Those are different tests, adjudicated by different bodies, and no number of votes in Essex changes what Harborne transferred or when. A man can be popular and have failed to file a form; British politics has managed both facts about the same person more than once.

What Farage has actually purchased, at a cost to the taxpayer of roughly £250,000, is delay dressed as vindication. The commissioner's inquiry does not evaporate because a constituency likes its MP; it resumes, as the Guardian and others confirmed on the 14th, precisely because he is now an MP again and therefore back within its jurisdiction. If Greenberg finds a breach serious enough to trigger a Commons suspension of ten sitting days or more, the Recall of MPs Act allows a petition; if 10% of Clacton's electors sign it, there is another by-election, a third act to a farce the voters have already paid for once. The Reform aide who told the BBC "we consider this matter closed" was not describing a fact. He was issuing an instruction.

There is a version of this column's usual complaint available here: the quango, the unelected body, the decision nobody can reach at the ballot box. It is worth resisting the urge to run it, because Greenberg's inquiry is not the quango state at its worst. It exists to answer a question elections structurally cannot: whether a member complied with rules Parliament itself wrote and voted for, rules about foreign money and declared gifts that predate Farage by decades and were not repealed by his resignation. The commissioner is doing precisely the job an accountability mechanism is meant to do when a politician tries to outrun scrutiny by calling a vote on something else. If there is a democratic deficit in Clacton this week, it is not that an unelected official has power over an elected man. It is that the elected man worked out, correctly, that most voters would treat a 63% margin as an acquittal on a charge nobody put to them.

The nearest historical control group is John Profumo, who lied to the Commons about Christine Keeler in 1963, resigned within weeks, and did not stand for re-election to test whether the electorate could forgive him faster than the inquiry could report. He treated the two questions as different questions, because they were. Farage has found the reverse technique: don't answer the charge, outvote it. The report is still being written. Daniel Greenberg's desk has not moved.

Ronesha's younger son needs a fridge for his medication. For seven weeks last year, in a bed and breakfast in England, he didn't have one. The stay itself was already unlawful — the law caps a family's time in a B&B at six weeks, on the theory that anything longer causes measurable harm — but the illegality is not the interesting part. The interesting part is that nobody was fined, nobody resigned, and nobody at the Ministry of Housing, Communities and Local Government could tell you whose job it was to stop it happening, because responsibility for it is distributed across so many desks that it has, functionally, ceased to exist.

This is now the ordinary condition of childhood homelessness in England. Figures published this month show 177,530 children living in temporary accommodation as of March, up 5% on the year and the eighteenth consecutive quarterly rise. Eighteen quarters is four and a half years. It spans two prime ministers, three housing secretaries, and one general election fought partly on the promise to fix it. The number has risen in every single one of those quarters regardless of who was in office, which is the first clue that the people nominally in charge are not the ones deciding the outcome.

Follow the chain and it runs somewhere strange. Statutory duty rests with local councils, who must house homeless families under the Housing Act. Councils, though, don't set the housing allowance, the subsidy rates for temporary accommodation, or the planning system that would let them build their way out of paying private landlords by the night. Central government sets those. So the family in the B&B is the output of a decision made across a council's housing options team, a funding formula set in Whitehall, and a spending settlement negotiated by the Treasury — and at no point does that chain terminate in a person you could vote out for this specific outcome. That is not an accident of complexity. It is the settlement both main parties have preferred for a generation: give councils the duty, keep the money and the levers, then decline to say whose job it was to prevent this.

The government's defence is not unreasonable on its own terms. Prime Minister Andy Burnham has promised what his department calls "the biggest council house-building programme since the postwar period," a genuinely ambitious phrase if it turns out to be more than a phrase. Ministers can point to money committed, reforms to the rented sector, a homelessness strategy in preparation, and a housing market distorted by a decade of underbuilding and mortgage rates that turned home ownership into a luxury good. Councils, they might add, have discretion over how they procure temporary accommodation and could in principle do it more cheaply. None of this is fiction.

But eighteen quarters is not a number that survives a defence about circumstances. Circumstances change — interest rates rise and fall, chancellors come and go — and the number has done one thing throughout: risen. Eighteen quarters in a row is a design choice, and nobody signed their name to it.

What that design choice produces at ground level is documented with a thoroughness that ought to embarrass whoever last read it and moved on. The Shared Health Foundation's research into temporary accommodation identifies thirteen separate breaches of the UN Convention on the Rights of the Child, each one catalogued, not implied. Research from UCL's Institute for Innovation and Public Purpose finds children developing along what its authors delicately call "unconventional trajectories": isolation from friends, a measurable loss of the ordinary business of play, parents describing guilt and a total loss of agency over their own children's lives. Older data, now two decades stale, put school absence for this group at fifty-five days a year. Nobody has produced a fresher figure, which given how much longer stays have grown since is not a comfort.

Here is the one sentence I will allow myself without qualification: a country that lets a child's medication spoil for want of a fridge, in a stay that was already illegal, and then produces the same press release about it eighteen quarters running, has stopped treating this as a scandal and started treating it as weather.

Clement Attlee's government took office with the Treasury empty, half the housing stock bombed or condemned, and rationing that would run another nine years. It built roughly 200,000 houses a year and, while it was at it, a health service. Nobody in that cabinet had the luxury of blaming the previous government indefinitely, because the country had run out of time for that pleasure. What it had, and what ours conspicuously lacks, was a settlement in which the person announcing the target and the person answerable for missing it sat in the same chair — and rationing ran to 1954 anyway.

Today's Britain is roughly forty times richer than Attlee's and has produced, across those same eighteen quarters, a series of strategies still described as forthcoming. The number they are forthcoming about is 177,530. Somebody should have to say it out loud in the Commons, and then say exactly which desk it belongs to.

On the day it emerged that the men's prison estate in England and Wales had 83 spare cells left, eighty-three, in a system holding roughly 87,000 people, which is the margin of a coin toss dressed up as a policy, Estonia's justice minister was explaining, calmly and at some length, why the answer was no. Not a diplomatic no. A borrowed one. Liisa Pakosta reached for Margaret Thatcher's 1990 rejection of European integration, "No. No. No.", and applied it to a country that hadn't, as it turned out, actually asked her anything.

That detail is the whole story, and Westminster has managed to bury it under the bigger one. Tartu prison in Estonia has roughly 900 cells, and plenty of them are empty, enough that Sweden has just signed a deal to send up to 600 of its own prisoners there from next year: a proper contract, ratified by both parliaments, money changing hands for beds that would otherwise sit unused. The UK held talks about the same arrangement in 2024. They came to nothing. Pakosta says no formal proposal was ever put to her government at all, which means the "deal" Britain has spent two years discussing in select committees and party conference speeches was never a deal, or even an offer.

It was a rumour with a minister's quote attached.

The case for renting cells abroad isn't stupid. Norway rents cells from the Netherlands. Belgium has done the same. In a genuine capacity crunch, buying time abroad while you build at home is a sensible bridge, and the alternative, releasing prisoners at 40 per cent of sentence, which is the emergency measure Britain has actually adopted, or stacking them in police cells, which it has also done, carries its own real costs to victims and to public confidence in whether a sentence means anything. If the choice is between a short lease in Tartu and a burglar walking free eighteen months early, the lease wins that argument comfortably.

But Britain never made that choice, because making a choice requires someone to own it, and nobody has.

Chris Philp calls Labour's failure to pursue the Estonia deal "a catastrophic mistake" that should be "urgently reversed," which is a fine sentence for a man whose own government legislated for the policy and then didn't do it either. Reform floats the same idea as though repetition were a substitute for a signature on a contract. Labour explored it in 2024 and let it drop without ever explaining why.

Three parties have now had a go at this. The sum total of the diplomatic effort, by the Estonian government's own account, is zero formal requests. Nobody decided not to do it; it simply stopped happening, the way things stop happening in Whitehall when the political credit for announcing a plan outweighs the cost of ever delivering it.

Underneath the farce sits arithmetic nobody in this argument wants to say plainly. The prison population has doubled in thirty years while crime has fallen substantially. The Ministry of Justice's own projections had it reaching 99,300, against roughly 4,400 new spaces currently funded to arrive. Then a radon gas reading at HMP Dartmoor in March 2024 closed more than 180 cells overnight, and the system that was already living on 83 spare beds discovered it doesn't have a margin. It has a rounding error.

That's a fact about a country that treats prison capital spending the way it treats reservoirs and hospitals: nice to have, hard to justify against a discount rate, easy to defer, until deferral becomes the only policy anyone's actually implemented. Building prisons in England has the same disease as building houses. Nobody local wants one, the planning fights run for years, and the Treasury would rather fund an early-release scheme this year than a facility that pays off over twenty.

Renting Estonian cells was never a rival to that project. It was a way of avoiding having the project at all, and calling it pragmatism instead.

Thatcher's "No, no, no" was the sound of a country certain enough of itself to slam a door. Pakosta's borrowed version is something closer to a diagnosis: a small nation of 1.3 million people, with 900 cells and a functioning waiting list, telling a much larger one that it can recognise a serious offer when it sees one, and this wasn't. Britain didn't lose the negotiation. It never opened it.

The photograph the Press Association keeps running is worth sitting with. Albert Bowers and Jessie Cole, leaving Reading Magistrates' Court in 2019, smiling — appearing, in the agency's own careful phrase, to laugh or mock the assembled media. They were seventeen. They were sentenced to thirteen years for the manslaughter of PC Andrew Harper, dragged to his death behind their getaway car as they fled the theft of a quad bike in Berkshire. Next year, under a scheme designed to relieve overcrowded prisons, both become eligible for early release.

Stuart Blaik, the former detective superintendent who led the investigation, has said in his first public comments since leaving Thames Valley Police in 2023 that he is "deeply concerned and appalled." Neither man, he said, showed remorse in custody or at trial. He hopes the government "takes account of the strength of public feeling." It is a modest hope, and it assumes public feeling is the thing this system is designed to register.

It isn't, quite. In the wake of Harper's death, Parliament passed a law bearing his name, toughening sentences for those who kill emergency workers in the course of a crime. His family now say that law feels forgotten. The irony writes itself: a statute with his name attached, debated and voted on, is being quietly overtaken by a scheme with no name attached to anyone — built to manage cell capacity, adjusted this week by category, and not put to a vote at all.

Andy Burnham's own conduct over the past fortnight is the tell. Last week he widened the exclusions from the early-release scheme to cover rape, grooming, and serious child sex offences — but manslaughter stayed eligible, which meant Harper's killers stayed eligible. Downing Street then said it was not possible to carve out an exemption for one or two individual manslaughter cases. A day later, Burnham said the opposite: "I've not accepted it's impossible to do more. I believe there is more that can be done." He has ordered an urgent review. It reports in September.

There is a real case underneath the theatre, and it deserves stating plainly before it gets buried. Prisons in England and Wales are genuinely full. A scheme that carves out an exception for every sympathetic victim — and Harper is about as sympathetic a victim as the justice system has ever produced — stops being a scheme and becomes a list of names, at which point it can no longer do the one job it exists for, which is finding cell space for the next dangerous offender walking through the door. Burnham did not build this trap. He inherited a capacity crisis and a release mechanism, and widening categories piecemeal, case by case, is not obviously worse governance than blanket rules that spare nobody scrutiny.

But notice what the September review actually is. It is not a decision, and it is not scheduled to look like one. It is a device for letting the row that is dominating this week's news cycle stop dominating it, so that whatever gets decided in September can be decided by officials at the Ministry of Justice with nobody watching, because nobody watches in September. "Urgent look" is the tell: urgency is for cameras, and the review reports on Whitehall time, which moves at exactly the speed required to outlast public attention.

The machinery being urgently looked at, moreover, was not built for this case, or by anyone now in office. Automatic release at a fixed point in a custodial sentence is architecture that predates Bowers and Cole's crime by a generation — a reform made for a different overcrowding crisis, under a different government, for reasons that had nothing to do with a dragged police officer on a Berkshire road. Nobody presently in Cabinet voted for it. They administer it, the way you administer weather.

Against that machinery, everyone with a direct stake is trying to force an answer the system was never built to give. The Thames Valley Police Federation is warning it may take the Prime Minister to court. Matthew Barber, the Conservative police and crime commissioner for Thames Valley, wants emergency legislation. Probation officers have voted to strike against the scheme itself. All of them are appealing, in effect, to an authority that does not exist — a single accountable decision-maker who can be named, questioned, and, if the answer is wrong, removed.

That authority used to exist. It sat in Parliament, and it answered for sentencing law in debates you could read afterwards. What we have instead is a Prime Minister contradicting his own Downing Street within twenty-four hours, and a review due in September that will settle, without a vote, whether a law named after a murdered police officer means anything at all.