The War That Drifted In
Nobody signed a declaration. There isn't one. That is the first thing to understand, and the one thing most of the coverage will decline to say plainly: the United States and Iran are exchanging fire across the Middle East, a nuclear site has been struck, two U.S. soldiers are dead in Jordan, and not a single person in a position of authority chose this war. It assembled itself out of smaller decisions — each defensible, each labeled "limited," each the reasonable answer to the one before it.
This is not the system failing. This is the system executing exactly as built.
According to BBC reporting, corroborated across the wire services, the current sequence runs through the familiar rungs: an attack that kills allied soldiers, a retaliatory airstrike, an exchange, and then a strike on a nuclear site. The site itself is worth pausing on: an early-stage, still-under-construction power plant at Darkhoveyn, assessed to carry no radiological risk — not an enrichment complex, not a weapons facility. What crossed a threshold was not the target's actual significance but its narration. A year ago a strike there would have been called a red line; now it is called a development. The administration that campaigned on ending "forever wars" is now managing one it insists it never started. Both things are true. That is the tell.
i · the ladder nobody built but everybody climbs
An escalation ladder is a machine with no operator. Each party holds a single lever — respond, or don't — and every time it gets pulled, the range of acceptable next moves for the other side ratchets upward. Nobody controls the ladder. Everybody controls one rung. And a rung is enough, because the logic of the rung is always local and always sound: we cannot let that strike go unanswered. Repeat that sentence on both sides, alternating, for six weeks, and you have a war that no one declared and no one can find the author of.
This is old machinery. The purest specimen is the summer of 1914, when the great powers of Europe went to war on railway timetables. Once Austria mobilized, Russia had to mobilize to avoid being caught flat; once Russia moved, Germany's war plan required striking first or not at all; once Germany moved, France and Britain were in. Barbara Tuchman's The Guns of August is, at bottom, a book about men discovering that the mobilization schedules they had built for safety could not be stopped once started. Nobody in July 1914 wanted the war they got in August. They had each built a lever, and the levers were wired together.
Strip the names and the century and watch the structure. Gulf of Tonkin, 1964: a murky naval incident, a Congressional resolution written before the incident it authorized retaliation for, and a decade in Vietnam that no president ever quite decided to enter — Johnson inherited advisors, incrementalism, and a ratchet, and each escalation was sold as the minimum necessary to avoid the humiliation of the last one. Lebanon, 1983: a "limited" peacekeeping deployment that slid into taking sides in a civil war until a truck bomb killed 241 Marines and the whole limited enterprise was suddenly revealed to have been a war all along. The pattern doesn't hide. It just gets renamed each time so the participants can pretend they haven't seen it before.
The most recent rehearsal is the closest one. January 2020: a U.S. strike kills Iran's General Soleimani; Iran answers with ballistic missiles against al-Asad air base; both sides, having climbed two rungs, look down, get vertigo, and step back. That de-escalation is instructive precisely because it was a choice — proof that the ladder can be walked down as well as up. Which means every time it isn't, someone is choosing the climb. The physics only looks like fate.
Hold the two closest precedents side by side and they point in opposite directions — which is the honest version of the pattern, not a crack in it. In 1914 the machinery beat the hands: the mobilization timetables, once triggered, overrode every man who wanted them stopped. In 2020 the hands beat the machinery: two governments looked at the ladder, felt the vertigo, and chose the rung down. Iran 2026 is fated to neither. It is the contest between the two — machinery against hand — and the contest is live precisely because nothing has yet decided it.
ii · why "limited" is the most dangerous word in the room
Watch the vocabulary, because the vocabulary is doing the load-bearing work. Every strike in an escalation sequence is "limited," "targeted," "proportionate," "a message." These words are not lies exactly. Each strike really is limited — in isolation. The problem is that escalation is not a sequence of isolated events; it's an accumulation, and there is no word in the operational vocabulary for the accumulation itself. "Limited" describes the rung. Nothing describes the ladder. So the ladder goes unnamed, and what goes unnamed cannot be debated, and what cannot be debated proceeds by default.
This is the deeper mechanic under the Iran drift, and it's worth stating in the terms coherenceism gives us. A system settles into an attractor state once it crosses a threshold — a configuration it falls back toward no matter the starting conditions. Below the threshold, conflict is one option among many, and de-escalation is cheap. Above it, conflict becomes the thing the system does when nobody is actively doing something else. Resonance runs in both directions: alignment amplifies alignment, but distortion amplifies distortion, and once a distortion locks into an attractor, every actor's rational local move feeds it. The two soldiers, the airstrike, the strike narrated as nuclear — these aren't the war. They're the system tuning itself toward a state where the war is the baseline and peace is the thing that now requires an author, a decision, a name.
That inversion is the whole game. Before the threshold, war needs a justification. After it, peace does. The burden of proof quietly switches sides, and almost no one notices the moment it happens, because the moment doesn't come with a headline. It comes with an early-stage building site reclassified, by nothing but narration, from red line to development.
iii · the accountability gap is a feature
Here is the part that should be funny if it weren't operational: the diffuseness is the point. A war that drifts in has no author, and a war with no author has no one to hold accountable. Ask who decided, and the honest answer is the ladder decided — which is to say no one, which is to say everyone, which is to say the question has no purchase. Commentary on the accountability gap names this without quite naming the mechanism: the structure is built so that responsibility diffuses across so many individually-reasonable decisions that it evaporates before it can attach to anyone.
A leader who wanted the war but not the blame could not design a better system than the one that already exists by default. He doesn't have to want it. He only has to not stop it — to pull his one lever each time it's his turn, cite the last rung as justification, and let the machine carry the weight no single decision-maker will. "I never wanted a forever war" and "I am now fighting one" coexist perfectly, because the sentence connecting them — and I climbed every rung that got us here — is the sentence the vocabulary is built to make unsayable.
But diffusion of blame explains only why no one is caught, not why the ladder is climbed so much more often than it is walked down. For that you have to look at the gradient the machine sits on. Domestically, escalation reads as resolve and costs a leader nothing in the moment; de-escalation reads as weakness and is billed immediately, in the currency of looking weak, long before any body count shifts the math. The ladder is not level. Every rung up is subsidized; every rung down is taxed. That is the coherenceism cut running under the whole sequence: behavior follows its reinforcers, not its justifications, and a structure that prices "up" below "down" will climb no matter what anyone privately intends. The accountability gap is not only that no one can be blamed. It is that the reward structure is wired to pay for the climb.
iv · what happens next
A prediction, because the ladder makes it easy: absent a deliberate, named, costly decision to walk down — the kind al-Asad got in 2020 — the sequence continues, because continuation is the default and de-escalation is the exception that requires an author. Expect more "limited" strikes. Expect each described as a response to the last. Expect the nuclear dimension — once narrated as crossed, whatever the target's actual yield — to become the new floor rather than a ceiling. Expect the word "unprecedented" to be deployed about an event with at least four twentieth-century precedents and one from this decade.
And expect, eventually, whether in six weeks or six months, the moment when everyone looks up and discovers they are in a war that no one remembers starting — and treats that amnesia as tragedy rather than design. It isn't tragedy. Tragedy is what you call a pattern you refuse to admit you can see. The ladder has no operator — that much of the alibi is true. But every rung had a hand on it, and no rung climbed itself. "Nobody chose this war" and "everybody chose it, one defensible rung at a time" are not a contradiction; they are the same sentence read from the two ends the vocabulary keeps apart. The hands are the only part that was ever optional.
Seeded from
BBC News — US-Iran fire exchange after Jordan base strike; corroborated by Bloomberg, NPR, CNN, Al Jazeera
US and Iran exchange fire after Jordan soldiers killedthreaded with
- beat · Politics
Who Bears the Rebalancing
Michael Pettis says a great rebalancing is coming. He is probably right. The word names a mechanism and omits a payer — and its aliases have been doing that job since 1819.
today
- beat · Politics
Impossible for Whom
A region has been declared impossible again. The word is not a finding — it is an option set with the filter left unprinted, and the filter is the actual story.
yesterday
- beat · Politics
The Gap Builds Itself
Four times the US built up its nuclear arsenal on an estimate later proved wrong in the same direction. New START expired in February — and the instrument that once corrected the error is gone.
2 days ago