Britain’s Quiet War On Democratic Public Scrutiny

The modern British state seldom needs to hide information in the dramatic sense. It achieves much the same result by making information harder to find, harder to compare, harder to retrieve, or easier to misstate without challenge. The pattern is not confined to one department or one party.

It runs across justice, health, education, migration, policing, regulation, and statistics, and it usually arrives dressed in procedure, caution, privacy, methodology, or administrative reform.

This matters because accountability is not a question of whether information exists somewhere inside the state. It is a one of whether journalists, researchers, and ordinary citizens can reach it in a form allowing real scrutiny.

Weaken such access and government gains room to present flattering accounts of itself which nobody can test in time for the testing to count. Which is, broadly speaking, the whole point if you are running out of days to cover up bad news.

There is rarely a single secret memo, no dramatic instruction to shred files, no minister at a podium refusing to answer. There are contractual disputes, data quality caveats, reclassifications, privacy objections, delayed responses, redesigned systems, and a regulator too swamped to enforce the public's rights.

The cumulative effect is the same. A system in which data is technically available but practically unusable shields power almost as effectively as one in which the data is formally withheld. The distinction matters legally.

It matters a great deal less to the citizen trying to find out what is being done in his name.

Blair Told Us The Truth By Accident

Tony Blair described himself in his memoirs as a naive, foolish, irresponsible nincompoop for introducing freedom of information, and complained the law was utterly undermining of sensible government.

He was not saying transparency had failed. He was saying it had worked, that internal discussions might later be exposed, and ministers had therefore grown cautious.

Papers released by the National Archives in December 2025 confirm the hostility was not merely retrospective regret. In February 2005, weeks after the Act came into force, Blair's chief of staff Jonathan Powell warned him the system was beginning to seize up, at a cost to Downing Street of nearly twenty thousand pounds a week.

Blair's reply was a request for a solution. The following year the government proposed regulations making it easier to reject requests, and abandoned them only under public pressure.

The admission is useful because it strips away the pretence governments object to scrutiny for neutral administrative reasons alone. Sometimes the objection is simpler. Scrutiny is uncomfortable. It limits message control. It exposes mistakes. It hands awkward facts to people who intend to publish them.

The Court Archive The State Deleted

The clearest current example is Courtsdesk, and it deserves telling properly because it has been widely muddled with CourtServe, which is a different company doing a different job.

Courtsdesk was a searchable database of prosecutions in England and Wales, built from 2020 under a pilot agreed with HM Courts and Tribunals Service and approved by the Lord Chancellor. Around 1,500 journalists across 39 news organisations used it.

Its value was simple. Court listings arrive in fragments, from different systems, at different times, with defendants on one list and charges on another. Courtsdesk stitched them together so a reporter could type a name and find a case, which is what open justice looks like in practice rather than in principle.

In the process it also produced an unflattering audit of the courts themselves. Courtsdesk found journalists received no advance notice of 1.6 million criminal hearings, listings were accurate on 4.2 per cent of sitting days, and half a million weekend cases were heard without any notification to the press. A lot of this dragged in embarrassing facts about grooming gangs.

In November 2025, HMCTS issued a cessation notice alleging unauthorised sharing of court data, after the company used a third-party artificial intelligence firm to test a feature. Courtsdesk says it wrote sixteen times seeking dialogue and asked for the matter to be referred to the Information Commissioner's Office.

No referral was made. Chris Philp, the former minister who approved the pilot, asked for the decision to be reversed. In February 2026 the government refused, and ordered the archive destroyed. The department's own account of the suspension went to the Justice Committee later that month.

Ministers told the Commons no one has deleted any court records, which is true in roughly the way burning a library's card index leaves every book on the shelves. The individual lists survive. The ability to search them does not.

HMCTS promised its own replacement service from the end of March 2026. Court reporters, who have watched HMCTS run its records for some years, greeted this with the enthusiasm one would expect.

Set aside the data protection argument, which may even be sound. The outcome is a department accused of poor record-keeping by a private database, using a compliance failure to have that database erased, and promising to build a better one itself.

Whatever else this is, it is not a good look.

The Habit Is Older Than This Government

None of this began in 2024. Between 2020 and 2022, openDemocracy exposed the Cabinet Office's Clearing House, a unit circulating to Whitehall departments a daily list naming journalists and campaigners, the requests they had made, and advice on how to handle them.

A tribunal found a profound lack of transparency around the operation, and suggested it might extend to ministers. The Public Administration and Constitutional Affairs Committee held an inquiry, reported that FOI administration across government appeared inconsistent with the spirit of the Act, and asked for an independent audit. The government rejected every recommendation.

The National Police Chiefs' Council maintains a Central Referral Unit advising forces on how to answer information requests. In a single quarter of 2024 it reviewed more than 1,700 of them.

Records obtained by the BBC and Democracy for Sale show the unit logging requesters' organisations, advising against release, discussing responses which would likely attract media attention, and in some cases instructing forces to withdraw disclosures already granted.

A unit which tracks who is asking is no longer processing requests. It is managing scrutiny, which is a different activity with a different purpose.

A Data Watchdog Completely Under Water

All of this rests on the assumption a citizen refused information can appeal to someone with the time and appetite to make a department behave. That assumption is failing.

Freedom of information complaints to the Information Commissioner's Office rose by more than 40 per cent in 2025/26, taking the total past 10,000.

Data protection complaints reached almost 40,000 in 2025. Over six years the regulator received more than 220,000 complaints and issued an average of fewer than seven fines a year.

Its new triage framework published in February 2026 allows complaints assessed as low or moderate harm to be recorded for information only, without investigation.

The Good Law Project and the Open Rights Group have threatened legal action, describing the approach as a digital bin. The Commissioner himself resigned in June 2026.

A regulator this overloaded does not need to be captured. It only needs to be slow. Rights on paper stop being rights in practice when enforcement arrives long after the story has died.

It is worth adding a structural point which rarely gets attention. The Act binds public authorities, not the private companies increasingly delivering public services under contract. As delivery moves outward, the reach of the law does not follow.

Statistics Demoted Rather Than Deleted

Reclassification is subtler than deletion. The numbers still appear, but their status is softened, their certainty qualified, and their force in argument diluted.

The clearest case is the Labour Force Survey. Response rates fell during and after the pandemic, the sample was cut back in 2023 to free resources for a replacement survey, and the estimates due in October 2023 were suspended outright on quality grounds.

The Office for Statistics Regulation removed accreditation from the survey's estimates in November 2023.

By the time the dust settled, fourteen separate statistical outputs had lost accredited status, including employment statistics for the United Kingdom, the monthly labour market reports for Wales and Scotland, trade union membership figures, and business population estimates.

Britain spent the better part of two years arguing about employment, immigration, and worklessness while the survey underpinning the argument was formally not to be relied upon.

A note on terminology, since the label is often misused.

Experimental statistics ceased to exist as a category in September 2023. The current designation is official statistics in development, which sounds reassuringly like progress and means the figure has not met the standard for accreditation.

The Missing Children

Nowhere is the gap between official counts and reality wider than in school attendance. The Department for Education's own estimate was 117,100 children missing education at some point during 2022/23, itself up from 94,900 the year before, and drawn from local authority returns of varying quality.

In December 2024 the Education Policy Institute did what the department had not, comparing GP registrations against school rolls and home education records. Its estimate was up to 400,000 children in England not in school in 2023, of whom around 300,000 were unaccounted for entirely.

Both figures had grown by roughly half since 2017.

That is not a presentational dispute. It is a state which does not know where several hundred thousand children are, discovering the fact through an independent think tank using patient registrations at the doctor's surgery.

A register of children not in school is finally arriving. It is worth asking why it took a charity's arithmetic to force it.

Prisons Quietly Going Dark

Individual suspensions are always explicable. The Ministry of Justice has indefinitely suspended its table on first receptions of former armed services personnel into prison, on grounds of data quality, and now advises users to treat the existing published series with caution from the Covid period onwards.

Quarterly publication of tables on home detention curfew eligibility has also been suspended following a change in the eligibility rules. The Prison Reform Trust has separately reported the loss of national figures on how many prisoners are working in custody.

Each has a reason. Together they mean the veterans in our prisons, the prisoners released early, and the prisoners in work have all become harder to count in the same few years, at exactly the moment when early release, rehabilitation, and prison capacity dominate the politics of justice.

The Border And Willful Blindness

In May 2024 the Home Office began publishing daily figures for migrants prevented from crossing the Channel, alongside its daily arrivals count. Within a week, and after consultation with the French, it stopped.

The figures were described as estimates prone to measurement error. Before removal they showed 1,048 people stopped since 29 April. Arrivals data, which the government was already committed to publishing, continued.

The objection is not that estimates should escape qualification. It is convenient numbers appear rapidly while awkward ones turn out to be too fragile for daily release.

Note also the date. This was the previous government. The instinct is departmental, not partisan.

There is still no comprehensive official record of those who die or disappear attempting the crossing, despite the surveillance, interception, and operational awareness devoted to that stretch of water. Researchers and campaign groups have assembled those counts themselves.

On a matter successive ministers have called a national priority, the absence is difficult to read as an oversight.

The State In Court To Stop You Counting

In June 2025 the Centre for Migration Control asked the Ministry of Justice for convictions in England and Wales between 2018 and 2024, broken down by nationality and by specific offence, including rape, murder, robbery, weapons, and drugs.

The department refused.

The Information Commissioner ruled on 22 June 2026 that it must disclose the requested information, rejecting both the cost argument and the identification argument, and noting privacy risks could be handled by suppressing very small counts rather than withholding everything.

On 17 July 2026 the department appealed to the First-tier Tribunal. Three days later the government changed hands. The appeal, lodged by David Lammy in the last week of his tenure, now sits with a new administration and a new Lord Chancellor, and could be dropped tomorrow.

The department maintains there is no cover-up and points to the range of breakdowns it already publishes on foreign national offenders.

Those breakdowns show foreign nationals accounted for 14.1 per cent of sex offence convictions in 2025. They exclude naturalised citizens and offer no split for rape, grooming, or child sexual abuse.

Denmark, Germany, and Sweden publish this class of data as a matter of routine. The Police National Computer already holds it. By the Centre's account, some of the 2018 to 2024 figures have already been compiled in the course of the dispute.

This is precisely the subject on which anonymised, structured, official publication matters most, because in its absence the public debate runs on force-by-force leaks, campaign group estimates, and rumour.

A government fighting through the tribunal to prevent publication is not protecting anyone's privacy. It is protecting ambiguity, and it should not be surprised when the public draws the obvious conclusion.

Waiting Lists And Cooking The Books

Health offers the sharpest demonstration of how opacity feeds directly into misleading claims. From April 2025, NHS England paid trusts to validate their elective waiting lists, at £33 for each clock stop delivered through validation. Between April and September 2025 alone, £18.8 million was allocated and roughly 567,000 patients were removed in six months.

Shrewsbury and Telford took more than £460,000 for removing 14,148 patients. Princess Alexandra Hospital Trust received close to £1.2 million against a list of fewer than 40,000 cases. In January 2026, 268,283 people came off the lists in a single month, and the backlog fell by around 44,000.

The March 2026 target of treating 65 per cent of elective patients within eighteen weeks was met, with validation playing a substantial part.

Lists should not carry duplicates, or people who no longer need care, or the dead. The difficulty is that NHS England has not published a breakdown of removals by reason, so the public cannot separate patients treated from patients processed. That is the whole ballgame. It is the difference between healthcare delivered and a spreadsheet tidied.

Reporting by Hyphen found 79,130 names removed from the lists of 127 acute trusts between September 2024 and August 2025 because the patient had died before reaching the front of the queue. Of those, 28,908 had waited beyond the eighteen-week standard and 7,737 had waited more than a year.

NHS England correctly notes the data does not record cause of death, and many of these deaths will have nothing to do with the awaited treatment. Fairness requires saying so. Fairness also requires noticing the previous year's equivalent figure was higher, at 30,593.

But the caveat is the point.

Nobody can size the effect, because the composition of the fall is not published. Ministers get to announce the headline. The public gets to guess what is inside it.

Less Data But Bolder Claims

This is where the habit turns self-defeating. When a government reduces access to robust, challengeable information, it does not merely weaken scrutiny. It removes the discipline good data imposes on its own statements. The temptation to overclaim grows precisely as the means of checking shrinks.

Full Fact and evidence to parliamentary committees have catalogued repeated ministerial claims resting on unpublished or thin evidence. The absurd Office for Statistics Regulation has intervened often enough over the misleading use of official figures to have published considered work on the subject.

The waiting list story shows the mechanism at its purest.

Once public argument is driven by one headline number, and the composition of the number is obscure, ministers gain room to boast.

The same obscurity makes the boast brittle. When independent scrutiny catches up, as it did here within months, the claim collapses and takes some credibility with it.

Incompetence, Self-Protection, Or Design

Some of this is overload rather than conspiracy. Departments face real legal constraints, privacy duties, methodological failures, and administrative chaos. Not every degraded dataset conceals a motive.

The overload explanation cannot carry the full weight, though. Full-time equivalent civil service numbers reached 524,000 in March 2026, the highest since 2005, up 8,000 in twelve months, and up by more than a third since 2016.

The Chancellor announced in March 2025 an intention to cut administrative running costs by 15 per cent, implying a reduction of perhaps 40,000 posts. The number went the other way. If more staff coincides with less accessible data, the plea of being stretched stops being an explanation and starts being a convenience.

The honest answer is mixed motive.

Some secrecy is institutional self-protection. Some is bureaucratic caution. Some is genuine concern about privacy or quality. Some is political calculation.

For the citizen the outcome is identical. Less visibility, weaker challenge, and more room for official claims to outrun the evidence.

Real Accountability Means Real Action

A serious response would not consist of another commitment to openness, which costs nothing and means less. It would restore discoverability and comparability, which are the only forms of transparency worth having.

That means:

1. A searchable national record of criminal listings and outcomes, built to a standard journalists can actually use, and not merely promised.
2. Centralised, searchable regulatory findings, rather than the current arrangement at bodies such as the Nursing and Midwifery Council, where the most recent three months of outcomes appear on the hearings pages and older decisions must be hunted down as individual PDFs by name or registration number.
3. An information regulator resourced to enforce within the lifetime of a story.
4. Routine anonymised publication of politically sensitive datasets, with small counts suppressed, rather than a legal fight to keep them in a drawer.
5. The Freedom of Information Act following public money into private hands.
6. Official claims separating, in terms, service delivered from data cleansed.

Transparency is not meaningful if it exists only for those with inside knowledge, spare time, legal persistence, and the money to keep asking.

A country is not open because the information sits somewhere in the system. It is open when the public can find out what is being done in its name, without needing a lawyer, a tribunal, and eighteen months.