The rule is simple and it has been in place for years. Central government departments publish every item of spending over £25,000, so anyone can see where public money goes. Consultants, catering, software licences, agency staff. All of it, line by line, month by month.

That is the theory. Here is the practice. Of the 16 departmental spending datasets catalogued on data.gov.uk, just five are up to date. Eleven are behind. That is more than two thirds of the trail going cold. (Source: data.gov.uk, spending-25k-summary)

And "behind" is doing a lot of work in that sentence. The three stalest files in the collection were last updated on 12 August 2013. One belongs to the Ministry of Justice. Two belong to the Department for Education. A fourth, from the Department for Work and Pensions, stopped on 5 September 2013. A fifth, another Education file, ran on until 21 January 2014 and then stopped too. That is close to thirteen years of silence on files that were set up specifically so the public would not have to ask.

Now look at the other end of the table, because this is the part that changes the story. The most recently refreshed dataset is from the Department for Work and Pensions, updated 31 July 2026. Then Transport on 30 July 2026. Then Education, Defence and Justice, all on 8 July 2026.

Spot the overlap. Education, Justice and Work and Pensions appear in both lists. The same departments that are keeping one spending file current are sitting on others that have not moved since 2013. This is not a capability problem. Nobody who can publish a July 2026 file is incapable of publishing a 2014 one. It is a housekeeping problem, and housekeeping is exactly where transparency quietly dies.

The honest alternative reading, and it matters: some of those 2013 files may be deliberately retired. Departments restructure, datasets get superseded, and a spending series that ended in 2013 might have been replaced by a newer one that is perfectly current. The catalogue records when an entry was last updated, which is not always the same as when a department last published anything. Sixteen datasets is also a small denominator, so one reorganisation can swing the ratio hard. None of that is nothing.

But it does not get anyone off the hook, because the fix for a retired dataset is to mark it retired. Leaving a dead file sitting in a public catalogue with no explanation is worse than deleting it. It looks like data. It behaves like data. It is a decade-old husk, and the only way to find out is to click it and lose an afternoon.

My view, plainly: spending transparency is only worth anything when money is tight, and money is tight. A £25,000 threshold catches most of the discretionary spending that ever ends up in an argument. If two thirds of those files are lagging, then the standard answer of "the data is published, go and look" is not really an answer at all. Five out of sixteen is not a transparency regime. It is a rota nobody is checking.