Picture someone doing the most ordinary thing a citizen can do. They want to know what the Ministry of Justice spent their money on. There is a rule for this: central government publishes every transaction over £25,000. So they go to data.gov.uk, find the Ministry of Justice spending file, and look at the date on it.

12 August 2013.

That is not a typo, and it is not a one-off. Of the sixteen departmental spending datasets tracked in the £25k transparency summary, eleven are behind and only five are genuinely up to date (Source: data.gov.uk, spending-25k-summary). The stalest of them cluster in one extraordinary fortnight of neglect: two Department for Education files and one Ministry of Justice file all frozen on 12 August 2013, a Department for Work and Pensions file left on 5 September 2013, and another Education file abandoned on 21 January 2014.

Set that against the freshest end of the list and the whole thing gets stranger. The DWP updated a spending dataset on 31 July 2026. Transport updated one on 30 July 2026. Education, Defence and Justice all pushed updates on 8 July 2026. So these are not departments that have forgotten how to publish. They are departments publishing perfectly promptly in one place while leaving another file to rot for roughly thirteen years.

That is the bit worth getting annoyed about. A department that never publishes is failing openly. A department that publishes some files on time and quietly abandons others is doing something more corrosive: it makes the whole catalogue untrustworthy. You cannot tell, at a glance, whether the file you have opened is a live record or a fossil. Education, Work and Pensions and Justice each appear at both ends of the list. Same organisation, same obligation, wildly different outcomes depending on which link you happen to click.

And the consequences are not abstract. Spending-over-£25k files are the raw material for journalists chasing consultancy bills, for councils benchmarking contract prices, for suppliers checking who actually gets paid, for anyone testing whether a department's press release matches its bank statement. Every dead file is a hole in that record. Eleven behind out of sixteen means the majority of the catalogue cannot be relied on without checking the timestamp first, and most people never check the timestamp.

There is an easy fix and an honest one. The easy fix is to update the files. The honest one, if a dataset has genuinely been superseded, is to retire it properly: mark it as closed, point to the replacement, and stop it sitting there looking like current information. Leaving a 2013 file live on a government portal in 2026 is not transparency. It is transparency-shaped clutter.

Open data was sold as a discipline, not a favour. The promise was that anyone could follow the money without asking permission. Five out of sixteen files honouring that promise is not a functioning system. It is a system that works when someone remembers, and the record shows plenty of people stopped remembering more than a decade ago.