The Aiomic Age

One number per country, published with its workings.

Higher Distance means more margin. The number exists to force the argument about weights into the open, and it is always published with its sensitivity range.

Design constraints

  1. Weights are priors, published and perturbable. Every leaderboard carries the rank range under weight perturbation of 50 per cent either way.
  2. Level and rate never mix. The Distance ranks position. Velocity is reported beside it, because they are different questions.
  3. Interactions are explicit and few. Only registered mechanisms appear, as threshold multipliers, visible and capped.
  4. Symmetry. The index must be able to improve. Nothing in the construction ratchets.
  5. Data honesty. Coverage and the share of non-primary cells are printed. Low-coverage countries are flagged, not hidden.

Computation

  1. Load every sourced cell. Primary status wins over secondary, then the latest year.
  2. Normalise each indicator across the countries observed to a 0 to 100 pressure score, inverted where the dictionary marks pressure as lower. Normalisation is panel-relative.
  3. Family scores: the weighted mean of each family's observed indicators. A family with no data is missing, not zero.
  4. Level pressure: the weighted mean of observed family scores.
  5. Interactions: if the named family scores clear their thresholds, level pressure is multiplied. Total multiplication is capped and applied multipliers are listed.
  6. Distance = 100 minus the adjusted level pressure, floored at zero.
  7. Rate score: the rate indicators, combined the same way, reported separately.
  8. Sensitivity: 200 seeded draws, every family weight multiplied by a uniform factor between 0.5 and 1.5. The leaderboard prints each country's rank range across the draws.

The indicator panel

Fifteen level indicators across eight families, plus four rate indicators. An indicator must be published by an institution with a methodology page, recur annually or better, be comparable across countries on one definition, couple to at least one registered mechanism, and carry a status tag on every cell.

Conversion

  • Youth unemployment, ages 15 to 24 (World Bank, ILO modelled).
  • Youth not in employment, education or training, 15 to 24 (World Bank).
  • Graduate underemployment: tertiary-educated 25 to 34 in jobs not requiring tertiary education (Eurostat EU-LFS). National variants for the UK, US, Canada, Australia and New Zealand measure different things and are excluded from the composite.

Shelter

  • Young people spending more than 40 per cent of disposable income on housing (Eurostat, national).
  • 25-to-34-year-olds living with their parents (Eurostat, national).

Wealth structure

  • Gini index of disposable income (World Bank).
  • Net household wealth share of the top 10 per cent (WID.world, OECD).

Demography and binds

  • Total fertility rate (World Bank). Lower means more pressure.

Valves

  • Reciprocal labour-market access abroad: a constructed ordinal from full bloc access to none. Lower means more pressure.

Legitimacy

  • Trust in national government (OECD Trust Survey).
  • Distinct heads of government in the trailing ten years (public record).
  • Vote share of anti-establishment parties at the latest lower-house election (PopuList).

Attention

  • Daily time on social platforms, internet users 16 and over (DataReportal, GWI). The source redefined this measure in 2024, so the series restarts at 2025 and the break is carried in every row note.
  • Trust in most news most of the time (Reuters Institute Digital News Report).

Health

  • Life-expectancy gap at 30 between highest and lowest education (Eurostat). Available for five of sixteen countries on a series ending in 2017. The absence elsewhere is itself a finding. This column should not be cited until coverage improves.

Rate indicators

  • House price-to-income ratio, change (OECD).
  • Cumulative food-price inflation, latest three years (OECD, Eurostat, national).
  • Employment share in occupations most exposed to generative AI (ILO, IMF, OECD).
  • Enterprises using at least one AI technology (Eurostat, national). Contestable as pressure, so it enters the rate view only.

Modifying the index

The weights file is the editable surface: family weights, indicator weights, directions, interaction thresholds, the cap, the sensitivity parameters. Argue with them. That argument is the index doing its job. Every change is versioned with its reason, and if a published piece cited the old ranking, the next edition notes the revision.

What this index is not

Not a forecast. Not a comparison of cultures. Not stable under different weight priors, which is printed on it by design. And not the framework: the moves and the mechanisms are the framework.

Changelog

  • v0.1.0 (2 August 2026): initial construction. 15 level indicators across 8 families, 4 rate indicators, 3 indexable mechanisms, panel-relative normalisation, sensitivity at 200 draws. Weights await ratification.
  • Data refresh (3 August 2026): attention family filled from the Reuters Institute report and DataReportal; health family entered with the caveat above; World Bank series re-fetched. Re-verified 21 August 2026.

Last updated