Neither Washington nor Tehran is fighting for territory or a declared objective. Both are fighting with probabilities, and the theory says that is exactly when the dice matter most.

Thomas Schelling had a name for what is happening in the Gulf. In The Strategy of Conflict, published in 1960, the economist described deterrence as "the threat that leaves something to chance". A state does not need to promise war to move an adversary. It needs only to create a risk of war that neither side fully controls. The threat works because the outcome is uncertain, and because both parties know it.

Schelling won a Nobel prize for the idea in 2005.

The Strait of Hormuz is now the purest laboratory his theory has ever had. Consider the past week. The US struck Iran at least 170 times in two days, with Central Command reporting some 90 targets hit in the latest wave, from air defences and missile storage to naval assets and coastal logistics. Washington then declined to confirm follow-on strikes reported around the perimeter of the Bushehr nuclear plant. The Islamic Revolutionary Guard Corps answered with attacks on US bases in Kuwait and Bahrain, and with 10 ballistic missiles fired at a base in northern Jordan, which Amman says it intercepted.

Neither country has declared war. Neither has ruled it out. We are between the lines. 

The Cold War escalation ladder does not describe this; forget about overlaying Soviet archetypes on this. On a ladder, every rung is visible and each side knows where the other stands. What the Gulf has instead is a probability distribution. Call it a stochastic war: a conflict in which each actor deliberately randomises to deny the other a readable pattern.

The underlying game theory is old, but it has rarely appeared in recent years. John von Neumann and Oskar Morgenstern showed in Theory of Games and Economic Behavior, published in 1944, that the optimal approach in many adversarial games is a mixed strategy, in which a player randomises between moves according to fixed probabilities. A predictable player can be exploited. A random one cannot. Poker players internalised the lesson long before defence planners did.

Iran's randomisation game is the doctrine of the weaker party, but as previously noted, it excels in this culturally, where Anglo-Saxon societies like predictability, Iranians have entire different interpretation of things, leading Trump to call them "weird". Stripped of its air defences in two campaigns and burying a supreme leader whose successor has yet to appear in public, Tehran has retreated to the grey zone. Sea mines laid by small boats. Deniable strikes on merchant shipping, which they would blame on Israeli assets. Under the memorandum of understanding signed in mid-June, Iran undertook to use its best efforts to ensure safe passage through the Strait of Hormuz for 60 days. It then demanded that all vessels coordinate with its authorities and transit a northern route hugging its coast, and this week attacked three ships that took the southern route near Oman instead. Safe passage became a service Iran administers, with violence as the tariff for non-compliance. Variance is the weapon of a state whose conventional options have been bombed away.

Washington's randomisation has a different pedigree. In 1969 Richard Nixon told his chief of staff H.R. Haldeman that he wanted Hanoi and Moscow to believe he might do anything to end the Vietnam war. Haldeman later recorded the label Nixon gave it: the madman theory. Calculated unpredictability as leverage. The current US posture is a desiccation rather than a copy. Officials describe deliberate strike-and-pause sequencing, hitting Iran hard, then standing down to let negotiators work, while maintaining a target list as open leverage. Trump declared the June MoU "over" on June 8, then allowed within hours that a new deal was conceivable while doubting it would stick. For the sane among us, this is this is unfuriating. 

Whether the unpredictability is performed or genuine is unfalsifiable, and honest analysis should say so. From Tehran's seat the distinction hardly matters. The expected cost of provoking Washington cannot be calculated either way. That is the design.

The trouble with Schelling's mechanism is contained in his own phrasing. A threat that leaves something to chance occasionally delivers the chance (see Trump being elected twice at improbable odds). Every week of this conflict draws again from the distribution: ballistic missiles crossing Kuwaiti airspace, ordnance landing near a nuclear plant's perimeter, tankers burning off Oman. Add the succession. Mojtaba Khamenei has not been seen in public since inheriting the leadership. A state whose ruler cannot be read even by his own officials contributes variance that nobody selected as strategy. 

Frank Knight drew the relevant distinction in Risk, Uncertainty and Profit, published in 1921. Risk can be priced. Uncertainty cannot. The markets are attempting the former while living with the latter. Brent settled at $78.55 on July 9, its highest since June 22, and war insurers are advising shipowners to pause Hormuz transits altogether. The insurance market has, in effect, declared the distribution untradeable.

Schelling's insight was that leaving something to chance can keep the peace, and for 60 years it broadly has.

Nothing in the theory says the chance will not come up.