Five engines.
Five half-lives.

That is the whole insight, and almost every confused argument about national decline comes from ignoring it. Cohesion cracks in a decade and produces the headlines — polarisation, protest, institutional distrust, a surprising election. It is also the most recoverable. Demography fails over half a century, produces almost no headlines at all, and is effectively irreversible on any policy timescale.

So a country whose demographic engine has already seized — China, Russia, Italy, Korea, Japan — is on a fundamentally different road from one whose engines are merely misfiring, however similar the news coverage looks this year. The United States has a cohesion score of 34, which is genuinely bad and genuinely alarming. It also has a demographic score of 62, which is why it sits in the holding quadrant and Italy does not.

Paul Kennedy — imperial overstretch (1987)

Great powers decline when the military commitments accumulated at their peak outlive the economic base that paid for them. The test is the ratio of security spending to productive investment.

The critique, which is devastating and should be quoted openly: Kennedy called American decline at the precise moment America was about to have its best fifteen years. The Soviet Union collapsed, the productivity boom arrived, and the debt ratio fell. It is the single best argument for humility in this entire genre, and any site making Kennedy’s argument owes the reader that disclosure on the first page.

Why the idea survives the bad timing: the mechanism has arrived by a route Kennedy did not anticipate. The United States now spends more servicing its debt than defending itself — not because the garrisons got expensive, but because the borrowing that paid for them did. That is overstretch with a financial transmission rather than a military one.

Robert Gilpin — the cost of order (1981)

Hegemony is a business. The hegemon supplies public goods — sea lanes, a reserve currency, a settlement system, extended deterrence — and collects rents: seigniorage, standard-setting, alliance leverage. Decline begins when the cost of supplying the goods rises faster than the rents.

The observable is beautifully simple: is the leading power adding or shedding commitments? France’s African base network collapsing is a Gilpin ledger entry. Russia’s Africa Corps evacuating Kidal is another. The American debate over war powers is a third. Every one of them is a power discovering that a commitment costs more than it returns and quietly closing the position.

Organski and Kugler — power transition (1958/1980)

War is most likely not when a challenger is weak, nor when it is dominant, but during crossover — and only when the challenger is dissatisfied with the existing order. Graham Allison popularised this as the Thucydides Trap in 2017.

The critiques are strong and this site carries them in full on the conflict page: determinism, non-uniform measurement, the difficulty of operationalising “satisfaction,” the peaceful counterexamples, and the self-fulfilling-prophecy risk. The position here is that power transition identifies a hazard rate rather than a destiny, and that the dissatisfaction variable — not the power variable — is the one worth watching.

Joseph Tainter and Peter Turchin — complexity and elite overproduction

Tainter (1988): societies solve problems by adding complexity; complexity has declining marginal returns; eventually the society cannot afford the apparatus that keeps it functioning, and it simplifies — sometimes abruptly.

Turchin adds the quantitative machinery and two measurable drivers: popular immiseration and elite overproduction — too many credentialed aspirants competing for too few elite positions, reliably producing counter-elites and factional conflict.

The clearest live case in this index is China: youth unemployment at 16.9% among non-students in March 2026, produced by a graduate pipeline built for a decade of 8% growth meeting an employment mix that no longer generates those jobs. It is elite overproduction with Chinese characteristics, it has no electoral safety valve, and it is the most under-priced risk in the Chinese story.

The demographic school

Populations are the slowest-moving and most reliable of all the variables. A birth cohort that does not exist in 2026 cannot be drafted, taxed or employed in 2046. This is the index’s anchor, because it is the domain in which forecasting is least dishonest — the twenty-year-olds of 2046 are already a fact.

One caution stated explicitly. Glubb’s “fate of empires” — the ten-generation cycle ending in an “age of decadence” — circulates widely in this subject area and is not scholarship. It is a retired general’s pattern-matching, unfalsifiable and heavily moralised. This site uses the demographic evidence, which is excellent, and does not adopt the moral frame, which is not evidence at all.

Decline is not a slope but a sequence, and the stages are not interchangeable. The index places every nation on a six-stage track.

Two stages deserve explanation because they are the model’s original contributions.

Apex is scored as more dangerous than any decline stage. A power at or just past its peak has maximum capability and a deteriorating forecast — the exact combination that produces window logic. China and South Korea sit here for very different reasons.

Second Life describes a power that completed a full decline, executed a genuine rebound, and is now inside a second and structurally worse descent. Russia is the only case in the index, and the classification carries a specific claim: the first decline consumed the asset a second recovery would have required. In Russia’s case that asset was people. This is why Russia scores the steepest velocity in the model at −5.6 despite a power level of only 31, and why the index treats it as the most volatile rather than the most powerful actor.

What this framework refuses to do

• It does not produce a single decline number. A composite that averages Nigeria’s demographic 82 with its institutional 26 tells you nothing about either. The engines are published separately for exactly this reason.
• It does not treat decline as moral failure. A low engine score is a measurement, not a judgement. Europe trading global weight for stability and longevity may well be a rational choice.
• It does not forecast collapse. Every nation in this index will still exist in 2050. Relative decline is about share, and share is a ratio.
• It does not hide its weights. Every indicator, every weight and every source is on the method page. If you disagree with a score, that page tells you which number to argue with.

Continue
The method, weights and indicator library  ·  Where decline becomes fighting