Why thirty years is easier to predict than three

31 August 2026 · The Law of the Board

The claim sounds backwards, and everyone hears it as a rhetorical trick. It is not. It is a statement about which variables survive at which timescales, and it is the argument the fourth volume of the series rests on.

Two kinds of variable

Some things that move outcomes are fast and unpredictable. Who wins an election. Whether a negotiation collapses over something said in a room. Whether a particular person has a heart attack. A misread radar return. These are real, they genuinely change events, and nothing lets you forecast them.

Other things are slow and already determined. Everyone who will be forty in 2060 has already been born; the number is known to within a percent. Most of the debt that will constrain the 2030s has already been issued, at known maturities. The geography of a chokepoint has not moved in ten thousand years. Where the minerals are is fixed. What a reactor costs to build and how long it takes is a matter of record.

Over three years, the first category dominates. It is loud, it is what coverage consists of, and it swamps the signal.

Over thirty, the first category averages toward zero — not because the events stop happening, but because their effects point in random directions and partially cancel. What remains is the second category, which has been pointing in the same direction the whole time.

The consequence for forecasting

A three-year forecast is a bet on personalities. It requires you to know who will hold office, what they will decide under pressure, and whether an accident intervenes. Nobody can do this, which does not stop it from being the dominant form of published prediction, because it is the form that is entertaining.

A thirty-year forecast can be a bet on arithmetic. Demographics, debt service, resource geography, the replacement rate of installed capital. These do not require you to know anybody's name.

That is why the honest long forecast looks less impressive than the dishonest short one. It says: this arrangement cannot continue, here is the mechanism by which it ends, here is the band of years, and here is what would tell you I am wrong. It does not say which year, and it cannot.

Where the method still fails

Three failure modes, and they should be stated as plainly as the claim.

Timing. Structure tells you that something is unsustainable. It does not tell you when the unsustainable thing ends, and "unsustainable" has a documented habit of lasting a generation longer than the analysis expects. Every forecast in this framework carries a band of years for exactly this reason, and the bands are wide because narrow ones would be dishonest.

Discontinuity. A genuinely new technology can change the payoff matrix rather than a value inside it. Structural forecasting handles change within a game well and change of the game badly. Anything that alters the cost of energy, the cost of computation, or the cost of moving goods can invalidate a chain of otherwise sound reasoning.

The unlisted player. The commonest error in practice is not bad reasoning about the players you identified. It is a player you never listed — a creditor, a domestic faction, a supplier — who turns out to hold a veto. Structural analysis is only as good as the roster, and the roster is the step everybody rushes.

How to use it

Take any long-running situation you care about and sort what you know into the two categories.

In the fast column: current officeholders, current tensions, the state of the negotiation, this quarter's numbers. In the slow column: the population pyramid, the debt schedule, who controls the physical routes, what the installed capital costs to replace and when it must be.

Now ask what the slow column implies if the fast column is noise. That answer is your forecast. Then ask which single item in the slow column, if wrong, would break it — that item is what you should actually be watching, and it is almost never what is being reported.

The honest version of the claim

Nobody is seeing the future. The prophet who works from structure is reading the present carefully — reading the parts of it that are already fixed and that most commentary skips because they are not new.

That is a much weaker claim than prophecy and a much more useful one, because it can be checked. A forecast built this way states in advance what would prove it wrong, which is the only property that separates a prediction from a horoscope.

Common questions

Why is a thirty-year forecast easier than a three-year one?

Over three years, personalities, elections and accidents dominate and cannot be predicted. Over thirty, those effects point in random directions and largely cancel, leaving the slow variables — demographics, debt schedules, resource geography — which are already fixed and can simply be read.

What can structural forecasting not do?

Timing, discontinuity and unlisted players. It will tell you an arrangement is unsustainable and give you the mechanism, but not the year. A genuinely new technology can change the payoff matrix rather than a value inside it. And the commonest practical failure is a player never listed who turns out to hold a veto.

How do you tell a real forecast from a horoscope?

A real one states in advance what would prove it wrong, names the indicator you would read, and gives a band of years in which to read it. If a prediction can be reinterpreted after the fact to fit whatever happened, it carried no information when it was made. The structural reading it rests on is the Law of the Board.