The Metrics.
A society can be judged by one piece of machinery: whether effort still converts into a job, a home, a family, and a sense that the work is worth doing. We measure that, country by country, as the acceleration hits it.
Explore a country
Higher is better
The margin a society still holds.
Lower is better
How quickly that margin is being spent.
Scores are relative to this 16-country panel. They are not probabilities or a forecast of rupture.
Interactions flagged in Netherlands
None under the current thresholds.
Netherlands. Distance 81.1; Rate 86.1. Rank range 1.
16 of 16 countries
- 01NetherlandsDistance: 81.1Rate 86.1Rank range 193% coverage
- 02NorwayDistance: 74.7Rate 50.1Rank range 2–5100% coverage
- 03DenmarkDistance: 73.7Rate 59.5Rank range 2–5100% coverage
- 04SwitzerlandDistance: 71.6Rate 46.1Rank range 2–688% coverage
- 05GermanyDistance: 70.0Rate 52.1Rank range 2–793% coverage
- 06AustraliaDistance: 67.4Rate 43.0Rank range 4–880% coverage
- 07SwedenDistance: 67.3Rate 49.6Rank range 5–8100% coverage
- 08IrelandDistance: 64.2Rate 57.5Rank range 6–893% coverage
- 09PolandDistance: 52.9Rate 36.1Rank range 9–13100% coverage
- 10New ZealandDistance: 52.1Rate 28.9Rank range 9–1265% coverage Provisional
- 11United StatesDistance: 50.9Rate 48.8Rank range 9–1467% coverage Provisional
- 12FranceDistance: 49.2Rate 31.4Rank range 9–1493% coverage
- 13CanadaDistance: 42.6Rate 64.4Rank range 11–1570% coverage
- 14ItalyDistance: 42.0Rate 23.1Rank range 12–15100% coverage
- 15SpainDistance: 37.1Rate 59.5Rank range 13–1693% coverage
- 16United KingdomDistance: 33.3Rate 56.2Rank range 14–1678% coverage
Rank ranges come from 200 runs varying every family weight by up to 50% either way. A wide range means the position is uncertain. Provisional readings have less than 70% data coverage.
How to read the index.
AI is an accelerant, not a sole cause. It speeds up whatever trajectory a society is already on, and in many countries the machinery was failing before the technology arrived. Conditions like these multiply rather than add. So we measure the coordinates the debate treats separately, together: eight families of pressure, from work and housing to institutional trust and attention. We publish the workings, the weights and the uncertainty with every number.
Go deeper
The framework.
The thesis, the seven analytical moves, and what would count against them.
Methodology.
How the number is computed: the indicator panel, the weights, the sensitivity analysis, the caveats.
Mechanisms.
The registered couplings between conditions, each with a pathway, a control case, and its falsification terms.
Questions we are asked
What does The Metrics measure?
Whether effort still converts into standing: a degree into a job, savings into a home, a household into a family, work into meaning. Sixteen countries are scored on fifteen level indicators across eight families of pressure, plus four rate indicators. Scores are panel-relative: they compare these sixteen countries to each other and to nothing else.
What is the difference between Distance and Rate?
Distance is margin: how far a country stands from rupture, on a scale of 0 to 100, where higher is better. Rate is how fast that margin is being spent, 0 to 100, where higher is worse. Reading them together is the point. A country can hold a wide margin and still be spending it quickly.
Where do the numbers come from?
From published statistical series that have a methodology page: the World Bank, the ILO, Eurostat, the OECD, WID.world, the PopuList, DataReportal and the Reuters Institute. An indicator qualifies only if it recurs annually or better, is comparable across countries on one definition, couples to a registered mechanism, and carries a source status tag on every cell.
How are the weights chosen?
They are published priors, not findings. Nobody knows how these pressures truly combine, and anyone offering a fixed model of it is describing a religion. So every leaderboard carries a rank range drawn from 200 seeded draws, with each family weight multiplied by a factor between 0.5 and 1.5. Where the range is wide, the position is ambiguous, and we say so.
What does a wide rank range mean?
That the country's position depends on the weighting rather than on the data, so the rank should not be quoted on its own. Read the family scores and the registered mechanisms instead. A composite starts an argument; it does not settle one. The number is published with its workings so the argument can be had properly.
What is missing from the index?
Coverage and the share of cells drawn from non-primary sources are printed on every country page. Countries below the coverage threshold are flagged provisional rather than hidden. The health family rests on a five-country series ending in 2017 and should not be cited. Graduate underemployment is measured differently outside the EU, so those national variants are excluded from the composite.
What would change the index?
Four failure conditions are named in advance. A registered mechanism whose conditions co-move without the predicted effect is retired in public. If AI-exposed labour markets diverge favourably for new entrants over a sustained period, the accelerant term is wrong. If our forecasts resolve at or below chance, the framework is decoration. If rank ranges widen until the number carries no information, the index is retired.
Is this a forecast, or a ranking of countries?
Neither. It is not a forecast, not a comparison of cultures, and not stable under different weight priors, which is printed on it by design. It measures structural pressure, which is measurable, and declines to predict ignition, which is not. The registered mechanisms carry the claims. The number is an index of the pressure behind them.
Last reviewed