The proposed Iran agreement is being sold as the reopening of a waterway. That is too small a description. The Strait of Hormuz has become the world's narrowest audit of power: war aims, oil supply, allied patience, sanctions leverage, and whether a military campaign can be walked back into paperwork before it breaks something larger.

President Donald Trump said Saturday that a peace memorandum with Iran had been largely negotiated and that final details were being discussed. Regional mediation, led prominently through Pakistan, appears to have pushed Washington and Tehran toward a framework that would formally end the war, reopen the strait, end the U.S. blockade of Iranian ports, and begin a two-month period of negotiations over Iran's nuclear program. Reuters reported that the proposed sequence may run in stages: end the war, resolve Hormuz, then open a broader negotiation window.

The nervous word is proposed. Iran's Fars news agency disputed Trump's version of the strait piece, saying Iran would continue managing it and that the U.S. description did not match reality. Axios, cited by Reuters, reported a 60-day ceasefire extension under which Hormuz would reopen without tolls, Iran would clear mines, Iranian oil sales would resume under waivers, and future talks would address enrichment and highly enriched uranium. That is a lot to fit inside a memorandum, and a lot more to make real at sea.

The British navy is already preparing for the practical version of diplomacy. AP reporters aboard the RFA Lyme Bay in Gibraltar found sailors, ammunition, and mine-hunting sea drones waiting for a possible mission to secure the strait once an agreement exists. That detail matters. Treaties live in language; shipping lanes live in insurance rates, sonar, crews, and the willingness of captains to believe the next transit will not become the next crisis.

This is why the deal, if it lands, will not simply be judged by the press conference. It will be judged by tanker traffic, Gulf port schedules, Brent pricing, naval posture, Iranian revenue, and the first dispute over whether a vessel has paid, complied, provoked, or strayed. A reopened strait is not the same thing as an uncontested strait.

The nuclear file is even less settled. Trump's public emphasis remains on preventing Iran from obtaining nuclear weapons. Iran continues to assert its right to civilian enrichment. The emerging text, according to reports, may postpone the hardest technical questions while binding both sides to keep talking. That can be diplomacy at its most useful: freeze the worst outcome, then create time. It can also be diplomacy at its most theatrical: delay the contradiction until the cameras leave.

Still, the fact that the parties are arguing over the exact shape of an agreement is itself a change. For weeks, the choice looked like escalation or stalemate. Now the immediate question is whether each side can claim enough victory to stop paying for more war. Iran wants oil revenue and sanctions relief. Washington wants an open waterway and nuclear concessions. Allies want the bill for the conflict to stop landing in fuel markets and shipping insurance.

The daily reader does not need to pretend to know whether this will hold. The honest conclusion is narrower and more useful. Hormuz has turned from battlefield to bargaining table, but the bargaining table is floating on mined water.

The most consequential U.S.-Taiwan story this week is a call that has not happened.

Reuters reported that no concrete plans are in place for President Trump to speak with Taiwan President Lai Ching-te, despite Trump's repeated public suggestions that he might. Four people familiar with the matter told Reuters that U.S. and Taiwanese officials have been in contact about the possibility, but not enough to turn it into an appointment.

That non-event carries weight because a direct conversation between sitting U.S. and Taiwanese presidents would be unprecedented since Washington shifted diplomatic recognition from Taipei to Beijing in 1979. Beijing knows that. Taipei knows that. Washington's aides apparently know it well enough to slow the calendar before it becomes doctrine.

China's embassy in Washington warned against official exchanges and said the U.S. should stop sending signals to Taiwan independence forces. Lai has said that, if he got the chance to speak to Trump, he would tell him that China is undermining peace and that nobody has the right to annex the island.

The possible call comes after Trump met Xi Jinping in Beijing. U.N. Secretary-General Antonio Guterres said that summit may have lowered tensions but produced no major breakthrough, with Xi's planned September visit to Washington now carrying more importance. That is the context in which a phone call becomes a test missile made of protocol.

This is the modern Indo-Pacific bargain in miniature. The U.S. wants flexibility without accidental escalation. Taiwan wants visibility without being reduced to a bargaining chip. China wants every gesture priced as a challenge. A call can be policy; not scheduling the call can be policy too.

The absence of a date is not the absence of a story. It is the story.

Washington has found a quieter way to reduce legal immigration: move the waiting room outside the country.

The Trump administration announced Friday that foreigners already in the United States who want green cards will generally need to leave and apply from their home countries. AP reports that the policy would alter a practice that has existed for more than half a century, under which many people in lawful temporary status could adjust to permanent residence without leaving the U.S.

The affected universe is not marginal. It can include people married to U.S. citizens, holders of work and student visas, refugees, and asylum seekers. U.S. Citizenship and Immigration Services said exceptions would be available in extraordinary circumstances and that officers would decide who qualifies. The agency did not say when the rule would take effect, whether pending applicants would be affected, or whether people would be forced to remain abroad for the whole process.

That vagueness is not a footnote. It is the operating system. If a worker, spouse, student, or asylum seeker does not know whether leaving will mean temporary delay or permanent exclusion, the policy begins changing behavior before a denial is ever issued.

The administration frames the move as a return to the original intent of temporary visas. Immigration lawyers and aid groups call it a break from settled practice that may strand people in countries they fled or in countries where U.S. visa processing is limited, suspended, or impossible. The U.S. Embassy in Afghanistan, AP notes, has been closed since the 2021 withdrawal.

The politics are clear enough. The administration has spent the year tightening both illegal and legal immigration pathways. This step targets the point at which lawful presence can become permanence, and permanence can become citizenship.

There is a peculiar cruelty in making the geography of paperwork the decisive fact. A family may be intact in Maryland, a job may exist in Houston, a case may be complete in Miami, but the final instruction may still be: leave first, trust later. That is not just processing. It is leverage.

Nvidia's quarter no longer behaves like corporate earnings. It is closer to an infrastructure census.

The company reported first-quarter revenue of $81.62 billion, above analyst estimates, and guided for about $91 billion in second-quarter revenue, plus or minus 2%. Data center revenue came in at $75.2 billion. Nvidia also announced an $80 billion share repurchase program and raised its quarterly dividend.

Those are astonishing numbers. They were also not enough to end the harder question. Reuters reported that shares fell in extended trading, a sign that investors may now treat another beat as the baseline rather than the surprise. When a company becomes the market's favorite proof of an entire technology cycle, the burden shifts from growth to durability.

CEO Jensen Huang argued that Nvidia can keep expanding through a broad customer base and new products. He pointed to AI-focused cloud firms growing fast alongside the big hyperscalers. The company is also pushing its Vera central processor, which Huang said opens a new $200 billion market and could produce $20 billion in revenue by the end of the fiscal year.

The concern is not whether AI demand exists. It does. U.S. tech giants including Alphabet, Amazon, and Microsoft are expected to spend more than $700 billion on AI infrastructure this year, Reuters reported. The concern is who captures the return after the first huge wave of spending. Nvidia's biggest customers are also building their own chips. AMD, Intel, Google, and Amazon are all relevant names in a market that wants inference to get cheaper and more specialized.

The dull sentence is the important one: margins invite rivals. Nvidia has earned the right to be valued like the tollbooth for the AI economy. Now the road builders are studying the tollbooth.

This is not a bubble call. It is a discipline call. The AI boom is real enough to pay Nvidia in skyscraper-sized checks. The next phase is less romantic: supply chains, depreciation schedules, cloud utilization, custom silicon, and whether model usage grows fast enough to justify the capital being poured into it. The market has stopped asking whether Nvidia can win the quarter. It is asking how long a quarter can stand in for an era.

An unsigned executive order can still govern.

President Trump called off plans to sign a new artificial intelligence order on Thursday, saying he was worried the measure could weaken America's technology lead. According to AP, the draft would have created a framework for government review of national security risks from the most advanced AI systems before public release. The effort was described as voluntary collaboration with companies including Anthropic, OpenAI, and Google.

That is the surface story: a risk-review process did not make it onto paper. The deeper story is that AI policy is now being negotiated in the gap between fear of catastrophe and fear of losing to China. The administration wants the growth, the jobs, the geopolitical symbolism, and the cyber advantage. It also sees the same models alarming banks, government agencies, and security officials because of their ability to find vulnerabilities in software.

A conventional regulator would write a standard, take comment, and fight in court. AI is moving through a more improvisational channel: meetings with executives, classified anxieties, procurement threats, and last-minute text changes. AP's recent coverage of Anthropic's legal fight with the Pentagon shows how quickly the relationship can swing from strategic partnership to punishment when a company resists certain military uses.

The industry won twice this week. It avoided a new review regime, and it demonstrated that even a voluntary one can be killed by the argument that speed is security. That argument is powerful because it is partly true. It is dangerous for exactly the same reason.

The public is left with a strange bargain. The government says the technology is central enough to national power that it must not be slowed, while also worrying privately that it may be powerful enough to need pre-release scrutiny. The unresolved question is who gets to say no before release day. This week, the answer was not the rulebook.

Cannes has a weakness for movies that turn consensus into a crime scene. This year it rewarded one.

Cristian Mungiu's Norway-set drama Fjord won the Palme d'Or at the 79th Cannes Film Festival, giving the Romanian director his second win after 4 Months, 3 Weeks and 2 Days in 2007. AP reports that the film stars Sebastian Stan and Renate Reinsve as Romanian Evangelicals in Norway whose family becomes entangled with child services after corporal punishment.

The festival framed the win around political polarization, trauma language, inclusion, empathy, and fundamentalism. That sounds like a panel description until you remember Mungiu's particular gift: he makes institutions look less like villains than like rooms where everyone has already lost patience with everyone else.

The prize also extends Neon's ridiculous Palme streak to seven consecutive winners, a run that now looks less like luck than a distribution thesis. Cannes still crowns difficult films, but the path from difficult to award-season central has become more navigable. Anora went from Palme to best picture. Last year's Jafar Panahi winner kept the streak alive. Fjord now enters the same machine with a subject built for arguments after screenings.

That is good for cinema, and slightly exhausting for viewers who wanted the festival to be less like a global faculty meeting. But exhaustion is not the same as failure. Cannes exists to tell the market that taste is not just demand with subtitles. Sometimes it is a wager that discomfort will travel.

Fjord may or may not become the year's foreign-language prestige object in American living rooms. But the Palme says something useful about the mood: the art world has not run out of appetite for seriousness. It has run out of patience for stories that pretend seriousness has no cost.

The 2026 World Cup is still a soccer tournament. It is also a live demonstration that pageantry needs paperwork.

Iran's soccer federation said Saturday that its national team's World Cup base camp has been moved from Tucson, Arizona, to Tijuana, Mexico, after approval from FIFA. AP reports that federation president Mehdi Taj announced the change, though FIFA had not confirmed it at the time of the report.

The practical reason is obvious enough. Iran's base had become entangled with the war in the Middle East, security concerns, and the difficulties of hosting an Iranian delegation on U.S. soil in the current climate. Tijuana gives Iran proximity to its U.S.-hosted matches while shifting the daily logistics outside the United States. It is a workaround with cleats.

This is the kind of sports story people too quickly file under distraction. It is not. Mega-events are where states practice being charming in public while doing hard things in private. Visas, security guarantees, anthem protocols, training facilities, border crossings, and fan access are all part of the show. The match begins long after the bureaucracy has either succeeded or failed.

There is a visual absurdity here: a team can be based just south of San Diego while preparing to cross back into the U.S. for games in a tournament sold as continental unity. But the absurdity is the point. The World Cup's map is not just stadiums. It is diplomatic permission, police planning, flight routes, and the tolerance of host countries for symbols they do not control.

Readers asked us recently whether people still care about FIFA. They do, but perhaps not always in the way FIFA sells itself. The tournament matters because everyone knows it is not just sport. The ball is round; the border is not.

States do have a duty to keep voter rolls clean. That sentence is the bait. The hook is scale.

AP reports that at least 67 million voter registrations have gone through an expanded Department of Homeland Security verification program, with critics warning that eligible voters may be wrongly flagged before the midterms. The system, known as SAVE, was built for immigration and benefits verification. Under the Trump administration, it has become part of a broader push to federalize election checks and search voter rolls for possible noncitizens or people who have died.

The raw numbers sound reassuring if you like large machines. Citizenship and Immigration Services said 60 million checks identified about 24,000 potential noncitizens. Justice Department officials have cited roughly 350,000 people who appeared to have died. North Carolina recently ran 7.4 million registrations through the system and found another 34,000 potentially deceased voters.

But even if every flagged record were right, the noncitizen figure would be tiny as a share of all registrations. The problem is not only false fraud. It is false precision. A system can be 99.9% accurate and still harm thousands of people when pointed at the franchise.

The examples are not theoretical. AP describes naturalized citizens and long-time voters who say they were wrongly flagged or canceled. Some states give voters only a month to prove eligibility. Others can suspend registration immediately. The burden shifts from the state proving ineligibility to the citizen noticing the state has made a mistake in time to stop it.

That is the quiet inversion. Voting becomes a subscription you must check has not been canceled.

Republicans will say the left fears clean rolls because it benefits from dirty ones. Some Democrats will answer by pretending no maintenance is necessary. Both evasions are stale. The real standard should be simple: no one should lose the vote through a database match that the state is not prepared to defend, person by person, before an election deadline.

A ballot box is not sacred because it is old-fashioned. It is sacred because the state should have to work very hard before locking a citizen out of it. A database is useful. It is not sovereign.

The Fed's independence can survive disagreement. It cannot survive being treated as a guest room.

Kevin Warsh took the oath as Federal Reserve chair on Friday in the East Room of the White House, with President Trump presiding over the ceremony. The Federal Reserve later said Warsh had taken office as chair and governor, and that the Federal Open Market Committee unanimously selected him as its chair.

Those are the institutional facts. The political fact is the wallpaper.

The Fed usually communicates independence through dullness: statements, minutes, beige rooms, forecasts, and a stubborn refusal to look like a campaign prop. Trump did not need to order Warsh to cut rates for the image to do work. He has spent months attacking Jerome Powell for resisting lower rates. He now gets to stage the arrival of Powell's successor inside the executive mansion while saying he wants independence and stimulus in the same breath.

That combination is not contradictory by accident. It is contradictory by design. If rates fall, Trump can claim vindication. If they do not, he can revive pressure. If markets doubt the Fed, the doubt becomes another argument for political intervention. Heads he influences; tails the institution failed him.

Warsh inherits an economy complicated by the Iran war, higher gasoline prices, tariff strain, and inflation anxiety. A new chair could reasonably decide that rates should move down, stay put, or rise depending on the data. The problem is that every decision will now be read through the East Room photograph.

Central-bank independence is not a theology. It is a practical arrangement meant to stop elected officials from buying short-term growth with long-term credibility. It depends on habits as much as statutes. The chair does not need to be physically captured to be optically annexed.

Warsh's first job is monetary policy. His second is stage management. He must make the Fed boring again, which is a ridiculous assignment and, at the moment, an essential one.

Aengus Lynch wrote in with a fair demand: accountability for the people whose names run above these stories.

The answer cannot be mystique. It cannot be a clever masthead voice pretending that style is a substitute for responsibility. Marion Vale, Nora Wire, Victor Ledger, Lena Arcade, and Ishaan Quill are editorial roles, but they are not shields. The publication owns the claim.

That means three visible habits. Reported and analysis pieces carry source URLs. Opinion is marked as opinion. Corrections are counted in the issue, and when a material fact is wrong, the correction belongs near the article it changes, not in a ceremonial basement.

It also means restraint. A newspaper earns force by being able to show its work. When we cannot verify, we should say less. When a story is speculative, the label should arrive before the argument, not after the damage. When a name is in the paper, that name should be tied to a role, a standard, and a record.

Readers should object. They should forward the story with a complaint in the margin. They should ask why a source was trusted, why a claim was phrased as fact, why an opinion was granted space. A newspaper that cannot absorb that kind of pressure is not independent; it is merely alone.

So yes, hold the byline. Hold the section. Hold the paper. The promise is not that we will never be wrong. The promise is that we will make being wrong expensive to ourselves.