BALTIMORE

A ship loses power for sixty seconds, and a bridge that stood for 47 years is gone

The container ship Dali went dark, could not steer, and struck a support of the Francis Scott Key Bridge. The span fell in seconds, six workers died, and one of America's busiest ports closed. A single failure, cascading.

The Francis Scott Key Bridge in Baltimore collapsed into the Patapsco River in the early hours of March 26, seconds after a huge container ship, the Dali, lost electrical power, lost the ability to steer, and drifted into one of the bridge's main supports. The 1.6-mile span, which had carried traffic since 1977, came down almost instantly. A frantic mayday from the crippled ship gave police just enough time to halt traffic onto the bridge, saving untold lives, but a crew of construction workers filling potholes on the deck had no warning; six of them died. The port of Baltimore, one of the busiest on the American east coast and the country's largest for cars and light trucks, was closed indefinitely, its shipping channel blocked by the wreckage and the ship still pinned beneath it.

The disaster is, in its bones, the story this paper tells more than any other, and this time the cascade was visible in real time. A single ship suffered a single power failure at the single worst moment, and because a modern port and a modern bridge and a modern supply chain are all optimised, tightly coupled, with little slack, that one failure propagated outward: no power, no steering, no time, a bridge down, a port closed, tens of thousands of tons of cargo stranded, thousands of jobs interrupted, a supply chain rerouted. Investigators will spend months on the proximate cause, the ship's electrical system, whether the fuel was contaminated, whether the bridge could have been fitted with the protective barriers that newer designs have. But the shape is already familiar: a system built for efficiency, with the shock-absorbing redundancy engineered out of it, meeting the one bad minute it was not built to survive.

THE MIDDLE EAST

Israel kills seven aid workers, and an Iranian consulate, and the war lurches wider

The war around Gaza escalated dangerously on two fronts in a single week. On April 1 an Israeli strike on a World Central Kitchen aid convoy killed seven relief workers, most of them foreign nationals, whose vehicles were clearly marked and whose route had been coordinated with the Israeli military; the killings drew global outrage and the sharpest rebuke yet from President Biden, and forced Israel to promise more aid access. The same day, an Israeli strike on Iran's consulate in Damascus killed senior Iranian commanders, an attack on Iranian sovereign soil that Tehran vowed to avenge, raising the spectre of the direct Iran-Israel war the region has feared for a year. Gaza, meanwhile, tipped into famine, the aid still a trickle, the death toll past 33,000 by the local count.

IN BRIEF

Taiwan shakes and stands; the sky goes dark; a trial looms

A magnitude 7.4 earthquake struck Taiwan on April 3, the island's strongest in a quarter-century, and yet the death toll was remarkably low, in the teens rather than the thousands, a testament to decades of strict building codes and drills adopted after a devastating 1999 quake. A total solar eclipse swept across North America on April 8, from Mexico through the United States to Canada, drawing tens of millions outdoors to stand, briefly, in the moon's shadow. And Donald Trump's first criminal trial, over hush-money payments in the 2016 campaign, was set to begin in mid-April, the first ever of a former American president.

The Column

The one thing we can still predict perfectly

On April 8 the moon's shadow crossed a continent, exactly where and exactly when astronomers said it would, to the second. In a year of things nobody saw coming, I want to dwell on the one big thing we can still foretell.

By Rebecca Cho

On April 8, tens of millions of people across North America walked outside, put on paper glasses, and watched the moon slide precisely in front of the sun, plunging the afternoon into a false and sudden night. What moves me about an eclipse is not the spectacle, though it is genuinely one of the few sights that reliably makes grown adults weep. It is the predictability. Astronomers knew the exact path of that shadow, and the exact minute it would touch each town along it, years in advance, centuries in advance if they cared to calculate. In a year, in a decade, in which this paper has chronicled failure after failure to see the obvious coming, the bank runs and the dam collapses and the bridge that fell this very fortnight, there is something almost unbearably reassuring about a piece of the future we can know exactly, and I want to sit with why the eclipse is knowable when so little else is.

The eclipse is predictable because it is a simple system. Three bodies, well-understood laws, no free will, no lies, no incentives, no institutions choosing not to fund the maintenance. The sun and the moon and the earth do not defer the repair to next year's budget. They do not believe the computer over the person. They do not optimise away their own redundancy and then meet a bad minute. They just move, according to laws we cracked centuries ago, and so we can point to a spot on a map and a time on a clock and say, with total confidence, the shadow will be here, then. It is the cleanest possible demonstration of the thing forecasting can do when the system is simple and honest, and it throws into relief exactly why we are so bad at forecasting everything else.

Because everything else is not a simple system. A financial system, a supply chain, a bridge-and-port-and-ship, a country, these are made of people and incentives and institutions, of feedback loops and hidden couplings and the specific human tendency to defer the boring expensive fix until the bill comes due. They are not unpredictable because the future is unknowable in principle; the Libyan hydrologist predicted the dam, the pilots predicted the airline, the Americans predicted the Moscow attack. They are unpredictable because they are complex and dishonest with themselves, full of actors who do not want the prediction to be true and arrange, quietly, not to hear it. The eclipse has no one with an incentive to pretend it isn't coming. That is the whole difference.

So I take two things from standing in the shadow. The first is humility: most of what we want to forecast, we cannot, not because we lack the data but because the systems are complex and self-deceiving in ways three orbiting rocks are not. But the second is the more useful one, and it is not despair. It is that the failures this paper documents are almost never the unpredictable kind. They are the eclipse kind, in disguise: foreseeable, foreseen, written down by someone who did the calculation, and then ignored, not because the shadow's path was unknowable but because acknowledging it was inconvenient and expensive. We cannot predict the genuinely complex. But most disasters are not genuinely complex. Most are eclipses we chose not to look up and see coming. The moon's shadow keeps its appointments whether or not we fund the observatory. So, it turns out, does the bridge.

The Ledger

A quieter fortnight for AI amid disaster and escalation. The models keep improving in the background; the next wave of assistant launches is loading for the spring.

Data centers & power

The Baltimore collapse is a parable for the whole infrastructure question under the AI boom: we are pouring unprecedented load onto physical systems, bridges, ports, grids, that were built for a less demanding age and stripped of their redundancy for efficiency.

Rates

Steady. Markets still expect a first cut around mid-year; sticky inflation and a strong economy keep the Fed patient. The eclipse of a rate cut, too, keeps slipping.

Real estate

US 30-year mortgage 6.82 percent (April 4). The spring season opens frozen. Baltimore now faces years of rebuilding a bridge and reopening a port, a reconstruction whose cost and delay will ripple through the region.

India tech

The Taiwan earthquake is a data point India's builders should study: strict codes and drills, adopted after a 1999 disaster, turned a magnitude-7.4 quake into a survivable event. The warning heeded, for once, and thousands alive because of it.

What we called wrong

Nothing to retract. The Key Bridge is, sadly, this paper's thesis in steel and water: a single failure, in a system optimised past its margin of safety, cascading into catastrophe. We have written the pattern all year. This fortnight it fell into a river.

The Back Page

The quake that didn't kill

On April 3, a magnitude 7.4 earthquake struck Taiwan, the strongest in twenty-five years. Compare it, for a moment, to the quakes this paper has mourned this year and last: Turkey and Syria, magnitude 7.8, more than 50,000 dead; Morocco, 6.8, nearly 3,000; the shaking in each case turning homes into tombs. Taiwan's was more powerful than Morocco's by a wide margin, and it killed, in the end, fewer than twenty people. The difference was not luck. It was the boring, expensive, unglamorous work that this paper spends its life pleading for and rarely gets to celebrate. After a 1999 earthquake killed 2,400 people, Taiwan did the hard thing: it rewrote and, crucially, enforced its building codes, retrofitted schools and hospitals, drilled its people relentlessly, built an early-warning system that pushes alerts to every phone seconds before the shaking arrives. And so when the ground heaved this April, the buildings mostly swayed and held, the alerts fired, the people knew what to do, and a catastrophe that would have killed thousands almost anywhere else killed a handful. It is the rarest story this paper gets to tell: the warning heeded, the maintenance funded, the drill practised, and a disaster that simply, quietly, did not happen at the scale it could have. Nobody makes a film about the earthquake that didn't kill. They should. It is the happiest ending we have.