The most remarkable storm in the Atlantic this year is the one that has not formed.

As of Saturday morning, the National Hurricane Center listed no active tropical cyclones in the basin. Five relatively weak tropical storms have appeared during the season, but none reached the 74-mile-an-hour threshold that turns a named storm into a hurricane. September 12 therefore became the latest first-hurricane date of the satellite era, according to meteorologists cited by the Associated Press. A season normally approaching its climatological crescendo has instead produced a blank space.

The immediate cause is not mysterious. A powerful El Niño has strengthened upper-level winds that tear at developing systems, tilting their columns and separating thunderstorms from the circulation beneath them. Warm ocean water supplies fuel; wind shear keeps knocking the match away. Overall activity, measured through accumulated cyclone energy—which combines storm intensity and duration—is also at a post-1950 low, according to Colorado State University hurricane specialist Phil Klotzbach.

This is good news. A house not flooded is not a statistical inconvenience. An evacuation that never becomes necessary is not wasted simply because nobody sees the counterfactual family on the roof. The quiet has spared communities damage, displacement and the exhausting financial bureaucracy that follows landfall.

But a quiet season creates its own civic weather. Preparedness is sold through memory, and memory decays quickly when the sky stays blue. Emergency kits expire. Insurance arguments become abstractions. Drainage work competes against projects with ribbon cuttings. Forecasters must explain the strange conjunction that climate is warming while a particular hurricane season is exceptionally inactive. Weather does not owe anyone a simple narrative.

The record also exposes the difference between a forecast and a promise. Preseason outlooks anticipated a below-average year as El Niño developed, but they still generally allowed for several hurricanes. Seasonal science describes a range of plausible outcomes; the atmosphere selected the quiet edge of that range. The Atlantic can remain suppressed for weeks and still produce a dangerous late storm. Hurricane season formally continues through November 30.

The institutional test is whether relief can coexist with readiness. Disaster systems are easiest to finance immediately after disaster, precisely when labor and materials cost the most and decisions are made under duress. A season like this one offers the rarest commodity in emergency management: time without wreckage. Culverts can be cleared before they are submerged. Contact lists can be corrected before phones fail. Building codes can be enforced without a television crew standing in splintered lumber.

Nothing in Saturday’s record establishes that future Atlantic seasons will be gentle. It establishes only that, this year, one large climate pattern has overpowered several ingredients that usually make September dangerous. That scientific explanation is less emotionally satisfying than either complacency or doom. It is also more useful.

The coast has received a reprieve, not a diagnosis. The right response to an alarm that did not ring is to enjoy the sleep—and then check the battery.

Britain’s economy grew an estimated 0.4 percent in July, defying forecasts of stagnation and inviting a seductive explanation: artificial intelligence did it.

The Office for National Statistics offers firmer ground than the slogan. Computer programming, consultancy and related activities expanded 3.5 percent during the month. That category contributed 0.14 percentage points to services output and 0.12 points to real GDP—nearly a third of the monthly economy-wide increase. The ONS said it found evidence that businesses involved with AI and related technologies helped drive the sector.

That is meaningful. It is not the same as measuring an AI productivity miracle. National accounts classify output by industry, not by the metaphysical cause of every invoice. A consultancy implementing conventional cloud software may sit beside a company training a frontier model. Rising turnover can reflect real efficiency, expensive deployment work, imported computing capacity or a temporary investment rush.

The wider three-month picture was mixed. GDP grew 0.4 percent in the three months through July, led by a 0.6 percent rise in services, while production and construction each contracted 0.5 percent. Payroll data also indicated fewer employees than a year earlier. Output growing while employment falls may signal improving productivity; it can also describe an economy squeezing more billable work from fewer people.

The useful finding is narrower and more interesting than the headline. Britain has a large, exportable professional-services sector, and demand for computing work is now big enough to move the monthly national accounts. Whether that work leaves clients permanently more productive will appear later—in margins, wages, investment and sustained output—not in a single July release.

AI has entered the GDP tables. It has not yet escaped the footnotes.

Pierre de Fermat claimed his margin was too small for a proof. Three and a half centuries later, a group of Claude agents found the opposite problem: their version required roughly 13 million lines.

Anthropic says a multi-agent system produced an end-to-end formalization of Fermat’s Last Theorem in Lean in 11 days. A proof assistant can check the result mechanically from mathematical axioms. Nature described the work as a milestone, and Anthropic published its account of the project and the formal artifacts.

The distinction between proving and formalizing matters. Andrew Wiles supplied the celebrated human proof in the 1990s. The agents did not replace it with a novel mathematical route. They translated the enormous body of definitions, dependencies and arguments into a language whose smallest trusted kernel can verify each step.

That translation has long been punishing work. Mathematicians choose what is interesting; formalizers must also supply everything ordinary exposition leaves implicit. The 11-day run suggests that coordinated agents can now absorb much of this clerical infinity: searching libraries, filling gaps, repairing failed checks and distributing subproblems without losing the shared state of the project.

The result moves the bottleneck. A computer-checked proof offers unusually strong assurance that its encoded steps follow. It does not establish that the chosen definitions match the intended real-world question, that the imported libraries are free of implementation errors or that the project attacked a worthwhile theorem. Verification becomes cheaper; judgment does not.

The significant product may therefore be less a famous theorem—already known to be true—than a new industrial process for converting mathematical literature into auditable software. Thirteen million lines sounds like excess until one remembers what they purchased: a machine that can say precisely where certainty ends.

McClatchy has cut more than 90 unionized journalism and media jobs across 17 publications, according to the NewsGuild-CWA. The losses amount to roughly 30 percent of the chain’s union workforce and reach papers including The Miami Herald, The Sacramento Bee, The Kansas City Star, The Charlotte Observer and the Fort Worth Star-Telegram.

The cuts are geographically distributed but locally concentrated. The Associated Press reported 32 eliminated jobs at the Miami Herald. The New York Times reported that the Herald lost more than a third of its editorial staff and that its Spanish-language sister publication, El Nuevo Herald, was nearly emptied.

A chain can make layoffs look like arithmetic: consolidate a desk, centralize a function, spread a smaller staff across a larger map. Readers experience the subtraction differently. A school-board meeting goes uncovered. A source no longer knows whom to call. A public-record request is never filed because the reporter who understood the agency’s filing system has disappeared.

Digital publication makes this decline deceptively tidy. The website remains online. National wire stories fill modules. The logo, app and subscription charge survive. What vanishes is the expensive particularity that made the outlet local: names remembered, evasions recognized and a phone answered by someone close enough to visit.

The layoffs cap what the AP described as a turbulent summer for American news. Their most consequential output will not be a worse front page tomorrow. It will be hundreds of small public events that produce no durable record at all. A blank archive does not announce itself as an error. It simply becomes the only history available.

Migratory birds expose the administrative weakness of borders: they cross them without respecting which government funded which sanctuary.

BirdLife International’s new State of the World’s Birds report says 45 percent of migratory bird species are declining and one in nine is threatened with extinction. The organization released the findings at the Global Flyways Summit in Nairobi, where governments, scientists, conservation groups and financial institutions signed a declaration promising greater cross-border protection.

The operative word is *flyway*. A species may breed safely in one jurisdiction and winter inside a protected reserve in another, yet still collapse because wetlands, tidal flats or feeding grounds along the route disappear. Conservation conducted as a collection of isolated parks can preserve both endpoints while losing the journey.

The report identifies the slender-billed curlew as a warning. Last recorded at a Moroccan lagoon in 1995 and declared extinct in 2025, it became the first known global bird extinction in mainland Eurasia and Africa in recent centuries. Six other migratory species have been confirmed or suspected lost over the last 150 years.

The Nairobi declaration arrives with a claim that development banks have mobilized more than $6 billion in conservation finance since 2021. Money matters, but migration makes coordination the scarcer asset. Habitat restored on one coast can be nullified by drainage or development a thousand miles away.

A migratory bird is not protected because it possesses one safe address. It is protected only when every essential address remains connected.

Before Céline Dion returned to a Paris stage, Paris supplied the voice.

Close to 10,000 people gathered Friday at Place de la Bataille-de-Stalingrad for a free mass karaoke organized by the city and AEG. They sang the repertoire together on the eve of Dion’s first concert in a 16-date Paris residency, following six years away from full concerts amid serious health problems.

This could have been another piece of promotional architecture: municipal branding, a concert promoter, reservation totals and thousands of phones pointed at a screen. Instead, the format introduced a useful indignity. Karaoke makes devotion audible and imperfect. The audience could not outsource the difficult notes to the star it had come to celebrate.

Dion’s return has been built at metropolitan scale. Paris retailers installed themed merchandise and displays; the regional tourism agency estimated that the residency could generate between €500 million and €800 million in economic activity. Sixteen shows run through October, with another ten planned for May 2027.

Yet the free gathering was valuable precisely because it was not the scarce, sold-out product. People who could not secure tickets could still enter the repertoire. A famous voice became common property for an evening—not through unauthorized replication, but through ten thousand gloriously inadequate performances.

The modern concert industry specializes in turning affection into tiers: presale codes, premium packages, resale prices, branded macarons. Mass karaoke reverses the transaction. Nobody pays to hear perfection. Everybody contributes to the noise. Paris did not merely welcome a singer back. It rehearsed the public that had kept her songs alive while she was gone.

The autonomous vehicle industry likes to discuss displaced drivers in the future tense and market valuations in the present. Tax law should reverse the order.

A British think tank has proposed taxing self-driving cars to offset job losses. The principle is sound because the commercial value of autonomy comes partly from removing paid human time from transport. A vehicle that never sleeps, eats or asks for a raise is not merely a safer steering system. It is a labor-replacement machine with tires.

The industry’s preferred rebuttal is familiar: autonomy will create engineering jobs, expand mobility and produce corporation-tax revenue. All may occur. None guarantees that a delivery driver in Leeds can become a perception engineer, or that new tax receipts will arrive in the same communities where wages disappear.

A useful levy should be simple: charge fully driverless commercial vehicles per occupied mile, at a rate that declines as the sector demonstrates broad social benefits. The revenue should follow the affected labor market—funding wage insurance, retraining chosen by workers rather than vendors, and transit in places where automated fleets skim profitable routes.

Do not tax the technology merely for being new. Tax the specific cost it externalizes. Human-driven vehicles already impose congestion, pollution and crash costs and should face appropriate charges too. But replacing a payroll creates a distinct gain for the operator and a distinct loss for the public systems financed by wages.

The best time to establish that principle is before fleets become indispensable and their owners can describe any charge as an attack on mobility. Society routinely taxes work more legibly than it taxes the capital built to eliminate work. A driverless-mile levy would correct that absurdity by a few pence at a time.

If the machine can use the road without a person, it can use the road without pretending the missing person costs nothing.

Netflix spent years persuading Hollywood that a title could be globally dominant while keeping the arithmetic largely inside its own building. Now it plans to report box-office grosses for six forthcoming films. Civilization has rediscovered the receipt.

The change is selective, not a surrender. Reports say the disclosed films will include Greta Gerwig’s *Narnia: The Magician’s Nephew* and several other titles receiving substantial theatrical runs. Netflix has separately announced that *Narnia* will open in IMAX and theaters worldwide in February 2027 before reaching the service in April.

Box office is an incomplete measure. It rewards price, distribution and marketing as much as affection. It cannot tell us whether a film changed a viewer’s mind or will matter in twenty years. But it is public enough to be argued over. Theater owners report sales; rival studios can compare performance; journalists can test executive adjectives against money surrendered at a counter.

Streaming replaced that common, flawed metric with a richer private dataset and then published whichever fragments supported the week’s narrative: views, hours watched, households reached. The platform knew more while the public knew less.

That opacity changed creative power. A filmmaker whose work was declared a success often could not independently inspect the success, much less use it in the next negotiation. Viewers were asked to experience popularity as a press release.

Six reported grosses will not repair the information imbalance. They do, however, reveal that transparency is compatible with Netflix’s business whenever the company wants theaters, prestige and cultural ceremony. Once a number can be disclosed for an awards contender, secrecy elsewhere stops looking technologically inevitable.

A public box-office figure is crude. So is a window. You still notice when somebody turns on the light.