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.

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.

The leaderboard

Computed 21 August 2026 · 16 countries · 15 level + 4 rate indicators · weights v0.1.0, published as priors

Distance: margin. Longer bar is betterRate: margin being spent. Tick further right is worseM1 interaction fired* coverage below 70 per cent, provisional

1. Netherlands81.1
Rate 86.1Coverage 93%Rank 1–1
2. Norway74.7
Rate 50.1Coverage 100%Rank 2–5
3. Denmark73.7
Rate 59.5Coverage 100%Rank 2–5
4. Switzerland71.6
Rate 46.1Coverage 88%Rank 2–6
5. Germany70.0
Rate 52.1Coverage 93%Rank 2–7
6. Australia67.4
Rate 43.0Coverage 80%Rank 4–8
7. Sweden67.3
Rate 49.6Coverage 100%Rank 5–8
8. Ireland64.2
Rate 57.5Coverage 93%Rank 6–8
9. Poland52.9
Rate 36.1Coverage 100%Rank 9–13
10. New Zealand *52.1
Rate 28.9Coverage 65%Rank 9–12
11. United States *50.9
Rate 48.8Coverage 67%Rank 9–14
12. France49.2
Rate 31.4Coverage 93%Rank 9–14M4
13. Canada42.6
Rate 64.4Coverage 70%Rank 11–15M2
14. Italy42.0
Rate 23.1Coverage 100%Rank 12–15M4
15. Spain37.1
Rate 59.5Coverage 93%Rank 13–16M1
16. United Kingdom33.3
Rate 56.2Coverage 78%Rank 14–16M2

№	Country	Distance & rate	Rate	Interactions	Coverage	Rank range
1	Netherlands	81.1	86.1	—	93%	1–1
2	Norway	74.7	50.1	—	100%	2–5
3	Denmark	73.7	59.5	—	100%	2–5
4	Switzerland	71.6	46.1	—	88%	2–6
5	Germany	70.0	52.1	—	93%	2–7
6	Australia	67.4	43.0	—	80%	4–8
7	Sweden	67.3	49.6	—	100%	5–8
8	Ireland	64.2	57.5	—	93%	6–8
9	Poland	52.9	36.1	—	100%	9–13
10	New Zealand *	52.1	28.9	—	65%	9–12
11	United States *	50.9	48.8	—	67%	9–14
12	France	49.2	31.4	M4	93%	9–14
13	Canada	42.6	64.4	M2	70%	11–15
14	Italy	42.0	23.1	M4	100%	12–15
15	Spain	37.1	59.5	M1	93%	13–16
16	United Kingdom	33.3	56.2	M2	78%	14–16

Take the numbers: CSV · JSON

The two markers pull against each other by design: the bar is how much margin a society still holds, the tick is how fast it is spending it. A long bar with a far-right tick is a country living well today and paying for it tomorrow. Rank ranges come from 200 draws perturbing every family weight by up to 50 per cent either way. A country whose range spans much of the table has an ambiguous position, and we say so rather than citing its rank. Provisional rows sit below the coverage threshold: a data-collection to-do, not a finding.

Go deeper

The framework.

The thesis, the seven analytical moves, and what would count against them.

Read the framework

Methodology.

How the number is computed: the indicator panel, the weights, the sensitivity analysis, the caveats.

Read the methodology

Mechanisms.

The registered couplings between conditions, each with a pathway, a control case, and its falsification terms.

Read the register

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 22 August 2026