The UK's productivity crisis was a statistical error: 0.7% became 1.3%
Fifteen years of Britain being cast as the worst performer on productivity among developed economies may have been an ONS miscalculation of hours worked — counting people on holiday as working 40 hours inflated the denominator and pushed productivity down.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
The UK productivity crisis was a calculation
After the financial crisis, UK productivity growth was estimated at 0.7% a year, the worst among developed economies, and it became the raw material for fifteen years of the "productivity puzzle." After the ONS changed how it measures hours worked, that number became 1.3%, nearly double. Output and GDP did not change; only the denominator did: Britons actually work far fewer hours than previously thought. Chris Giles says pre-crisis growth was 2%, and 2% means productivity doubles roughly every 35 years, while 0.7% means it takes a century to double; 1.3% falls in between. Britain is no longer an international outlier, but growth is still below the pre-crisis rate.
— Chris GilesAsking people their weekly hours cannot be trusted
The old method simply asked respondents how many hours they worked in a week and took the answer at face value. The problem came at follow-up: if no one answered the door in August, the ONS carried over the 40 hours that person had reported three months earlier, plugging in a 40 — when that person was very likely on holiday and had not worked a single hour. The OECD recommended back in 2018 that countries stop doing this; the UK took eight years to change. Giles says that over those eight years people wrote a batch of "UK productivity crisis" articles that no longer hold up. He admits it is infuriating, but the change is in the right direction.
— Chris GilesOne in ten Irish people as clergy is a misquote
The book said that in the 1960s one in every ten Irish children became a nun, priest or Christian Brother. The 1961 census shows 18,854 Catholic clergy and nuns, 2,125 Christian Brothers and other religious brothers, and 4,842 seminarians, for a total of 25,821 — 1/104 of the population. The 1951 ratio was 1/127. The original source is research by Mary Daly, who looked only at students sitting the 1966 leaving certificate exam, when fewer than 20% of the school-age population took that exam, about 12,000 people, of whom roughly one in ten entered religious life. That qualifier was lost in retelling and became "all children." The author has said subsequent editions will be corrected.
— LizzieGrid queue numbers are inflated by applications
Between November 2024 and June 2025, cumulative electricity demand in the UK grid connection queue jumped from about 41 GW to 125 GW, tripling in six months, with data centres accounting for 73 GW. But queuing is cheap, like booking a table at a popular restaurant months ahead, and many projects will never actually be built or will shrink. Ofgem estimates that if all this data centre demand materialised, the cost would be close to £700 billion, nearly a quarter of UK annual GDP. A 2025 government report projected 6 GW would be needed by 2030, while another queue analysis reckons about 20 GW has serious financial backing behind it. Johnny Gowdy of Regen says no one knows how much will ultimately be built.
— Nathan GowerThe data centre emissions comparison counts only direct emissions
Foxglove's analysis claims two planned data centres at full load would emit more carbon than ExxonMobil's UK operations. That comparison counts only ExxonMobil's own operational emissions, excluding the petrol and diesel it sells — which is the main source of emissions, something an ordinary reader may not spot. Foxglove also assumes the data centres run at 100% power year-round, while a scholar studying AI's carbon footprint uses 65%. More important is the power source: the planning documents for these two projects say that initially, because they cannot connect to the grid, they will use on-site gas generation. The UK grid as a whole is about half renewable and only 30% gas, so this batch of data centres amounts to stimulating demand for higher-carbon generation.
— Nathan GowerVinyl's physical limits mattered less, not more
The 1950s story goes: the closer a record gets to the centre, the lower the linear speed, and the less bass and brightness a groove can hold, so you put the quiet song at the end of a side. Mitchell Sterling's database has 2,284 albums and 27,000 songs spanning 1953 to the present. Measured by relative loudness, before 1982 closing tracks averaged about 0.65 dB quieter than earlier tracks; after 1982, when cassette sales overtook vinyl in the US and the constraint disappeared, the gap widened to about 1.5 dB. In the streaming era after 2010, the gap is still twice what it was in the vinyl era. In other words, the less the physical limit mattered, the quieter closing tracks became.
— Mitchell SterlingThe quiet ending went from constraint to artistic device
Sterling defines a quiet, acoustic closing track as the "typical quiet acoustic farewell." In the 1950s more than 60% of albums ended with a quiet, acoustic song; as recordings overall got louder, that absolute share slid all the way down to under one-sixth in the 1980s; by the 2000s and 2020s it had recovered to about one-third. Sterling's analogy is the save icon on a floppy disk: the constraint is long gone, the habit remains. He also found in streaming data that an album's last track averages longer, more acoustic and sadder — but that is an average, and hip-hop, heavy metal and electronic music are full of loud exceptions.
— Mitchell SterlingIn their own words · checked verbatim
We used to think that the amount of output per hour increased by 0.7% a year after about 2010. And now we think it's gone up to 1.3%.
Chris Giles2:12
It's become clear that when you ask people how many hours a week you work, you shouldn't believe it.
Chris Giles4:16
it's taken us eight years to introduce this change eight years where we've all written productivity crisis articles which are now not true.
Chris Giles5:19
So it's not one in 10 of all Irish children. It's one in 10 of the 12,000 children who achieved their leaving certificate.
Lizzie12:38
The problem is it's fairly cheap to apply for a place in the queue. It's a bit like booking a table at a popular restaurant months in advance.
Nathan Gower15:44
if there ever was a genuine attempt by studio engineers and producers to make sure that this was the the done thing in the 1950s then it's moved from being a technical limitation and preference to being an artistic lever to grab
Mitchell Sterling25:00
it's a bit like the floppy disk save icon it's just still there this idea that you end on a quieter note
Mitchell Sterling25:00
Figures
| UK post-crisis productivity growth (old estimate) | 0.7% a year | 2:12 |
| UK post-crisis productivity growth (new estimate) | 1.3% a year | 2:12 |
| UK pre-crisis productivity growth | about 2% a year | 2:12 |
| 1961 Irish population | 2,673,473 | 10:33 |
| 1961 Irish Catholic clergy and nuns | 18,854 | 10:33 |
| 1961 Irish clergy as a share of population | 1/104 | 10:33 |
| Change in cumulative UK grid queue demand | about 41 GW in November 2024, 125 GW in June 2025 | 14:41 |
| Data centre demand in the queue | 73 GW | 14:41 |
| Estimated cost if all data centre demand materialises | close to £700 billion | 15:44 |
| Relative loudness gap of album closers before 1982 | about 0.65 dB | 24:00 |
Glossary
- ONS
- The UK's Office for National Statistics, which produces official figures on GDP, hours worked and productivity.
- productivity puzzle
- The UK's unusually weak post-financial-crisis productivity growth, which long defied explanation.
- leaving certificate
- The Irish secondary school leaving exam, taken by only a minority of students in the 1960s.
- loudness wars
- The CD-era phenomenon of producers competing to push master volumes ever higher.
How to listen
Investors and researchers tracking the UK economy, the macro productivity narrative, or who need to cite UK official statistics; anyone investing in data centres and grid infrastructure should see the third segment.
The Irish clergy segment (7:26-13:38) is a pure fact-check, unrelated to the rest, and can be skipped.