The Reliability Discount.

Empires rarely announce their own decline. There is no press release, no formal notice pinned to the gates of Rome. Instead there is a slower and more administrative process, in which allies begin doing something they would once have considered rude: hedging.

Britain did not lose its empire at Suez. It lost it in the years afterward, when countries that had once assumed London would turn up started making other arrangements first and informing London second. France did not stop being a great power on a Tuesday. It stopped being one gradually, in a hundred small meetings where someone quietly asked whether Paris still meant what it said the previous week.

Washington has spent eight decades enjoying an unusual privilege. Not just military weight or the size of its economy, but something harder to put a value on: the assumption that an American position, once stated, would still be the position when it actually mattered. That assumption has been having a difficult run.

In May last year, Wall Street settled on an acronym for the pattern. TACO, Trump Always Chickens Out, was coined by a Financial Times columnist to describe tariffs that arrive at ruinous levels, cause the market to fall over, and are then quietly reduced a fortnight later once the falling over has been noticed. The president, asked about it at a press conference, called it "the nastiest question" he had been asked and explained that it was, in fact, negotiation. Markets have continued trading as though it were TACO.

Iran has offered an even more compressed version of the same theatre. Strikes are announced with a specific evening attached. Gulf states intervene. A deal is said to be close. The strike is called off, sometimes with ten minutes to spare, sometimes the same day it was threatened. This has now happened often enough that news outlets keep a running tally, the way one might track a cricket score.

None of this requires a theory about motive. Historians can spend decades arguing about whether a given retreat was calculated brilliance or simple improvisation. Everyone else has moved on to a more practical question, which is what to do about it. The answer, it turns out, is insurance.

Australia is not abandoning its alliance with the United States. It is quietly diversifying who else it talks to. Japan continues to describe the alliance as its cornerstone, while quietly adding rather more scaffolding around it, in the unglamorous language of regional supply chains and defence pacts that do not require a phone call to Washington first. None of this is announced with any drama, because insurance rarely is.

This is the interesting part of the reliability discount. It does not require a single dramatic failure. It accumulates from ordinary Tuesdays, one reversed tariff and one cancelled airstrike at a time, until governments that once treated an American commitment as load-bearing start treating it as merely helpful. The commitment itself may never be formally withdrawn. It simply stops being relied upon quite so heavily.

Markets, characteristically, worked this out before most governments did. Traders now spend a portion of every morning trying to establish presidential temperament alongside the usual business of interest rates and employment figures, a discipline that would once have sounded closer to astrology than economics. It is now simply called risk management.

Perhaps unpredictability is a legitimate negotiating strategy. Perhaps it even produces better outcomes on the odd occasion. History, however, keeps a long and unglamorous memory for this sort of thing. It rarely records the day an empire lost a war. It records, in much smaller print, the day everyone else quietly stopped assuming it would still think the same thing on Thursday.

Rome kept minting coins with its own face on them long after the rest of the empire had quietly stopped believing they were worth the metal.