THE PACIFIC

Trump and Xi meet, and the world's two giants step back from the brink

After a year of escalating trade war, the American and Chinese presidents met in person and struck a truce: a pause in the punishing tariffs, a resumption of the rare-earths and soybeans each side had weaponised. The decoupling slowed, for now, and the world exhaled.

President Trump and China's Xi Jinping met in person in South Korea on October 30, on the margins of a regional summit, and produced the de-escalation a jittery world had hoped for: a truce in the trade war that had convulsed the global economy all year. The two agreed to pause or roll back the most punishing tariffs each had imposed, to restore the flows each had cut off as weapons, China's near-monopoly on the rare-earth minerals essential to modern technology, America's soybeans and advanced chips, and to a temporary understanding on the fate of TikTok. It was not a resolution of the fundamental rivalry, which remains the defining contest of the century, nor an end to the slow decoupling of the two economies, which continues. But it was a deliberate step back from the edge, a decision by both giants that the mutual damage of open economic warfare had become too great, and markets, which had feared far worse, rallied in relief.

The meeting was a reminder that even in an age of confrontation, the two economies remain profoundly entangled, each holding levers that can inflict enormous pain on the other, and that this mutual vulnerability is itself a kind of stability, a balance of economic terror that makes total rupture too costly for either to want. Whether the truce holds, or proves merely a pause in a longer decoupling, is the question that will shape the global economy for years. But for a fortnight, the two most powerful men on Earth chose de-escalation over confrontation, and a world exhausted by a year of trade war took the reprieve gratefully, however temporary it might prove.

TECHNOLOGY

A single cloud region fails, and a huge slice of the internet goes dark

A large part of the internet stopped working on October 20 when a major region of Amazon Web Services, the cloud-computing giant on whose infrastructure a staggering share of the world's websites and apps quietly run, suffered a prolonged failure. For hours, thousands of services, banks, airlines, government sites, popular apps, games, even smart-home devices, went dark or degraded, not because each had failed but because they all depended, invisibly, on the same underlying cloud infrastructure, and when it failed, they all failed together. It was CrowdStrike's cousin, another demonstration of how much of the modern world runs on a handful of concentrated points of failure, and how a single fault in one of them cascades instantly across everything built on top of it. Days later, Amazon announced it would cut some 30,000 corporate jobs, partly attributing the move to artificial intelligence.

IN BRIEF

Jewels from the Louvre; a hurricane razes Jamaica; a chainsaw endorsed

Thieves pulled off an audacious daylight robbery of the Louvre in Paris on October 19, using a lift to reach a first-floor gallery and making off in minutes with priceless crown jewels, a heist of almost cinematic brazenness that embarrassed France and gripped the world. Hurricane Melissa, one of the most powerful storms ever to strike the Caribbean, devastated Jamaica in late October, a Category 5 monster that flattened towns and underscored, again, the intensifying violence of a warming climate's storms. And in Argentina, voters handed the radical libertarian president Javier Milei a decisive midterm victory, endorsing his chainsaw approach to the state.

The Column

The monoculture underneath everything

A chunk of the internet went dark this fortnight because one cloud region failed. The outage is usually treated as a glitch. It is better understood as the visible symptom of a hidden bargain we have all made, trading resilience for efficiency, without ever quite deciding to.

By Daniel Reeve

A single cloud region failed this fortnight and took a startling swath of the internet down with it, banking apps, airline systems, smart-home devices, all dark at once because they all, unknown to their users and often to each other, depended on the same underlying infrastructure. The event is being filed as a glitch, a temporary outage, embarrassing but soon forgotten. It deserves to be understood instead as the visible symptom of a hidden bargain that the whole digital world has quietly made, trading resilience for efficiency at every level, without anyone ever quite deciding to, and arriving at a concentration whose fragility is invisible right up until the moment it fails.

The dynamic is one that biology understands better than computing, and the useful word comes from agriculture: monoculture. A monoculture, a vast field planted with a single genetically uniform crop, is enormously efficient, optimised, productive, cheaper per unit than the messy diversity it replaced, and it is also, for exactly the reason it is efficient, catastrophically fragile, because the uniformity that lowers the cost also means that a single pest or blight adapted to that one variety can destroy the entire field at once, where a diverse planting would have lost only part. Efficiency and resilience, the monoculture teaches, are not merely different goals but often opposing ones: the optimisation that strips out redundancy and diversity to lower cost is the very thing that removes the buffers a system needs to survive a shock. The cloud has become a monoculture in precisely this sense. The relentless economic logic of efficiency has concentrated the world's computing onto a handful of providers, a few regions, a small number of shared dependencies, because concentration is cheaper, and the same concentration that lowers the cost has removed the diversity that would have contained a failure, so that one region going down now blights the whole field.

The counterargument is serious and it is the reason the concentration happened: that the efficiency is not a vice but an enormous good, that the centralised cloud is more reliable, more secure, and vastly cheaper than the fragmented alternative it replaced, that the average user experiences far less downtime now than in the era of everyone running their own servers, and that the occasional spectacular outage is a real but acceptable price for a system that is, on average, more robust than what came before. This is true, and it must be held honestly: the concentration delivered real reliability gains most of the time, and a nostalgic call to return to a fully fragmented internet would trade a rare large failure for frequent small ones. But the monoculture frame exposes what the average-reliability defence hides, which is the changed shape of the risk: concentration does not merely lower the average failure rate, it changes the distribution, trading many small independent failures for rare, correlated, catastrophic ones, and a system optimised for the average case is precisely a system that has removed its defences against the tail. The fields are more productive than ever, and one blight can now take all of them at once, and the second fact is not refuted by the first.

And that is the bargain worth making visible, because we have made it everywhere, not just in the cloud, and mostly without noticing. The same logic that concentrated computing has concentrated supply chains onto single suppliers, finance onto single clearing systems, software onto single dependencies that thousands of applications silently share, everywhere trading the inefficiency of diversity and redundancy for the efficiency of concentration, and everywhere accumulating the same hidden fragility, the single point of failure whose existence is invisible until it fails. Each individual choice to concentrate was rational, cheaper, better on average; the aggregate result is a civilisation of monocultures, optimised systems that work beautifully until the one blight adapted to the one uniformity brings down the whole field at once. The outage this fortnight was small and brief, and it will be forgotten, which is exactly the danger, because the lesson it briefly illuminated is the one we keep declining to learn: that efficiency and resilience trade against each other, that we have been buying the first by selling the second for decades, and that a world of monocultures runs more cheaply than ever right up until the day it doesn't, all at once, together, for the same reason it ran so cheaply in the first place.

The Ledger

The AWS outage and Amazon's AI-attributed job cuts are two faces of the same concentration: the cloud and AI infrastructure the world increasingly depends on is held by a shrinking handful of giants, whose failures cascade and whose efficiencies eliminate the human roles they replace.

Data centers & power

The outage is a stark reminder that the AI build-out concentrates computing into ever fewer, ever larger data centres, deepening exactly the single-point-of-failure fragility this fortnight exposed. More concentration, more power in fewer places, more catastrophic when one fails.

Rates

The Trump-Xi truce eased trade-war fears; the Fed continues its cautious easing. Markets, relieved, rallied on the de-escalation, though the underlying decoupling and its inflationary pressures persist.

Real estate

US 30-year mortgage 6.22 percent (October 30), near the year's low, as easing continues. The market's one reliable comfort in a turbulent year.

India tech

The rare-earths truce matters for India's technology and manufacturing ambitions, dependent on the same critical minerals China dominates; the fragility of a supply chain concentrated in one country is a lesson India, courting its own semiconductor future, is absorbing.

What we called wrong

Nothing to retract. The AWS outage confirms this paper's long argument about concentration and single points of failure, now embedded invisibly in the cloud infrastructure of daily life, and getting worse as the AI build-out concentrates computing further.

The Back Page

Seven minutes at the Louvre

There is something almost restful, amid the grim news, about a good old-fashioned jewel heist, and the world seized on the Louvre robbery this fortnight with a guilty delight that says something about how heavy everything else has become. The facts were absurd and cinematic: in broad daylight, on a Sunday morning, with the museum open, thieves drove a furniture lift up to a first-floor window of the world's most famous museum, cut through the glass, smashed the cases in the gallery of the crown jewels, grabbed a fortune in historic gems, and were gone, the whole thing over in about seven minutes, before anyone quite understood what was happening. It was a robbery from another century, a caper, the kind of thing that happens in films and heist novels, not in the actual Louvre in the actual present, and its sheer brazen improbability made it, for a few days, the most-discussed story in the world, a welcome absurdity in a year of catastrophes. France was mortified; the investigation churns; the jewels, priceless and unsellable in any legitimate market, may be broken up and lost forever, which is the real tragedy under the caper. But for a fortnight the world got to think about something that was merely astonishing rather than tragic, a heist worthy of the movies pulled off at the temple of Western art in the time it takes to make coffee, and there was, in the collective delight at the audacity of it, a small reminder that not everything has to be heavy, and that sometimes the news can simply be that someone did something impossible and got away with it, at least for now.