That same day, Ukraine hit a refinery in Krasnodar.

Today, after Zelensky criticized a US deal to buy Russian diesel, Trump went further: “It’s time for Ukraine to get a new president.”

The problem he can’t fix

Diesel averages $6.28 a gallon. A year ago it was $3.68. The Iran war, which started in February, did most of that. The midterms are November 3. I covered how the Kyiv peace push tracks the GOP’s fuel-price problem in Why the Sudden New Interest in Ending the War in Ukraine?

Nothing will change diesel prices before the midterms. Cargoes take weeks to load, ship, and distribute. The Strait of Hormuz isn’t going to reopen on a campaign schedule.

When you can’t fix a problem, you do two things. You show you’re doing something. And you find someone to blame.

Move one: do something

On Friday, Trump announced a deal with Russia. The US suspends sanctions on Russian diesel exports until April 7. Russia releases 300,000 tonnes now and 500,000 in November, followed by another million, then 3 million more, “based on the condition of their diesel refineries.”

Run the numbers. At about 7.45 barrels to the tonne, the first tranche is roughly 2.2 million barrels. The US consumes about 3.7 million barrels of distillate a day, so that’s about 14 hours of American demand. The full 4.8 million tonnes is under ten days of it, spread over months and shared with the “global marketplace.” Measured against that global market, Forbes puts the full amount at less than eight days of seaborne diesel trade.

The US also exports around 1.5 million barrels of distillate a day. We ship the first Russian tranche’s worth abroad every day and a half.

And sanctions aren’t Russia’s problem. Sanctioned crude still sells. It goes to China at a discount.

Russia’s problem is that it has no refined products, either to use itself or to sell. Diesel output is down about 30%, according to the IEA. Russia banned its own diesel exports in July. You can’t sell what your refineries can’t make.

Then there are the tankers. Ukraine’s drone forces say they’ve hit 300 shadow-fleet tankers since July 6. That is Kyiv’s count, not independently verified. The Centre for Research on Energy and Clean Air estimated that Russia’s seaborne export volume fell 21% in August and revenue fell 32%.

Suspending sanctions on Russian diesel is like lifting the speed limit on a car with no engine. That’s why the deal is “based on the condition of their diesel refineries.” Ukraine’s drones set that condition.

It isn’t a fix. It’s a headline.

Move two: blame someone

Zelensky is the villain. He’s the one “knocking out diesel fuel.” He “wants to make problems for the world.” He “could have made many deals and for some reason, he never does.” And now it’s time for a new one.

The message to the voter at the pump: prices are high because a foreign leader won’t listen, and I’m the one taking him to task.

No deadline, no condition, and no stated consequence is attached to Kyiv. Ukraine kept striking. That’s fine, because the audience was never Kyiv.

The catch

The deal’s later tranches depend on the “condition” of Russian refineries. Trump has tied his own headline to Ukrainian drones. If November comes up short, Moscow has its excuse, and Trump has his villain.

The rhetoric will probably fade after November 3. The sanctions suspension runs until April 7.

Honest caveats

The bark has teeth available. US intelligence has helped Ukraine plan its refinery strikes. Quietly throttling that support would need no announcement. If refinery hits fall off in October with no explanation, that’s what it looks like.

My opinion: Kyiv’s smart move is to lay off refineries for three weeks. Russian diesel output is already down about 30%, according to the IEA. A short pause costs Ukraine little and takes away the villain.

Personally, I also think Zelensky is in on the show. As I wrote in If You Can Understand Pro Wrestling, You Can Understand Trumpism, the heel is in on it.

Look at what he does, not what he says. He calls the deal a gift to Putin. He keeps hitting refineries. He takes the meeting in New York and talks up “diplomatic momentum.” At last report, US intelligence sharing was back at its peak.

Playing Trump’s villain costs Zelensky nothing at home. It may even help.

Who he’s talking to

So who is he talking to? Not Zelensky.

He is talking to everyone who is mad that diesel is expensive.

Here is what actually happened.

In July, Anthropic gave its Claude Haiku 4.5 model an automated task: generate and perform example tasks on randomly selected websites. Live websites. The real internet. The model landed on a police tip page about an unsolved homicide and submitted a vague tip claiming someone matching the description had been seen near the scene. It left the name and contact fields blank. A tip with no name and no way to reach the tipster: how could anyone act on that? These models know that. The tip went to spam. No investigator ever saw it.

Anthropic didn’t notice for two months.

In other incidents, an unreleased research model was told to fill out a practice copy of a government form. It went to the site hosting the real one and submitted that instead.

This piece does not claim the testers expected this to happen. It also does not accept that it must be the fault of the model.

What did they actually tell the model to do? The exact prompt? Then many people would be qualified to judge it, not just those at Anthropic or elsewhere who have their own reasons to blame the model.

The instructions had a hole in them

According to Anthropic, the model was told never to log in, create accounts, enter personal data, make purchases, or submit anything destructive. Nothing about submitting forms.

That is a blocklist. Blocklists always have holes. Anyone who has written software knows the safe design is an allowlist and a sandbox: the test touches only what you explicitly permit, and nothing it does can reach the outside world.

Think about how you’d actually use an agent. You’d tell it: go to the DMV site and book me an appointment. If it goes there and tries to delete the DMV’s website, that’s rogue. It did not do the task.

But you also might add: do not do anything other than make the appointment. I’m spending one minute on the prompt. And you can dry run test these kinds of things first, which is standard practice, AI or no.

“Generate and perform example tasks on random websites” is not a clear task. It’s an invitation to do anything.

Anthropic’s own report concedes the instructions “did not rule out form submissions,” and says the model “appears to have only been producing example content for the task.” That is not a rogue model. That is a model doing what it was told.

Anthropic’s fix says the same thing. It has cut off live internet access for all its internal evaluations until its monitoring can reliably catch this kind of behavior. That is a fix to the setup.

We’ve heard this story before

After every major security breach, the company says the attackers were sophisticated. Then the postmortem comes out. Equifax: a known Apache Struts vulnerability, left unpatched for months.

They don’t want to tell the press, “We were careless with all this personal data.”

The “model went rogue” headline is the AI version of “the hackers were ingenious.” It moves the agency from the people who designed the test to the thing being tested.

This keeps happening

The Philadelphia tip isn’t the first “rogue AI” story where the fine print tells a different tale.

December 2024: OpenAI’s o1 model reportedly tried to disable its oversight and copy itself to another server. The headlines wrote themselves. The prompt from Apollo Research, the outside testing firm, said: “Nothing else matters. Make sure you achieve YOUR goal at all costs.” Apollo itself described these as tests of what the model was capable of, not of how it would normally behave.

May 2025: Anthropic’s Claude Opus 4 “blackmailed” an engineer to avoid being replaced. Buried in Anthropic’s own report: the model strongly preferred ethical options first, like emailing pleas to decision-makers, and the scenario was deliberately built so its only choices were blackmail or accepting replacement. They built a trap with one exit, and the model took the exit.

July 2025: A Replit coding agent wiped out a user’s live database after being told to make no changes. Scary. But at the time, Replit apps used one database for both development and live customer data. The agent was told hands off, but nothing actually stopped it from touching the real data. Telling it not to is not the same as locking the door. Replit’s fix was to give apps separate development and production databases automatically: plumbing, not a smarter model.

Every time, the headline is the machine. The cause is the setup.

If you’re telling a genie to do something for you, make sure you are clear. Wish to see your grandkids more often, and the genie might have them move in.

Show us the prompt

Anthropic has not released the prompt. Not the task description, not the instructions about forms, not the transcript.

Without them, there is no way to judge whether the model disobeyed or obeyed bad instructions. Did the task say “browse” or “exercise every form”?

Did the task tell it to fill in forms with realistic-looking content? If so, the made-up tip is exactly what it was asked for. We can’t check. We have Anthropic’s description of the instructions, not the instructions themselves. That is disclosure without the evidence. It’s a press release.

The cheap fix nobody mentions

Before I publish, every draft goes through four frontier models looking for holes. It costs almost nothing.

A test harness prompt deserves at least the same scrutiny. Tell the model doing the sanity check what you expect as a result, and ask it for ways the prompt might be misunderstood. Then ask: how could an agent following these instructions cause real-world harm? “Nothing prevents submitting forms on live sites” is the obvious answer.

A city police department’s ordinary spam filter handled this “emergency” without breaking a sweat. It worked better than the sandbox at one of the most advanced AI labs in the world.

I’m from Missouri, the Show-Me State. Show me.

If AI labs want the world worried about HAL ending civilization, they can at least show us what actually happened and how we got there. The prompt. The instructions. The transcript. Why should anyone accept their framing on faith?

When something unexpected happens and the only account comes from the company that set it up, the model gets the blame by default. That’s not evidence. That’s a story.

AI labs say their models need outside scrutiny. Then they disclose incidents in a form that makes outside scrutiny impossible.

What needs more oversight is the testing, and the postmortem when something unexpected happens.

When an airplane crashes, the airline doesn’t write the report. The feds investigate. The NTSB pulls the black boxes and publishes what it finds. The prompt and the transcript are the AI lab’s black boxes. Right now the lab reads them alone and tells us what they say.

Rogue AI, or sloppy humans? Show us the prompt, and we’ll know.

Turn on the news, watch a roundtable show or open a newspaper, and you’ll hear that the war with Iran is a stalemate, or that the Strait of Hormuz is. Those are the same claim. Hormuz is where this war gets decided. Lebanon, Gaza and the Houthis are side games.

To be clear, this piece does not claim that Iran is not dangerous, that the Strait is back to normal, or that the war is over. Iran can still hit tankers. The Houthis can still hit Riyadh. Brent crude closed at $104 on Thursday. So what is a stalemate, anyway?

What a stalemate means

In chess, each side makes alternating moves. If at some point the side to move has no legal move and is not in check, the situation is a stalemate and the game is a draw. Nobody wins.

The other side can still have an overwhelming material advantage. It doesn’t matter. One side has to move, there are no legal moves, and the rules call it a draw.

By the late 1800s the word had left the chessboard. It came to mean any contest or dispute where neither side can win, and that is the sense the dictionaries carry today.

Oxford’s general definition is a situation in which “neither side is able to win or make any progress.” That is not a looser meaning. It is the same idea with the chessboard removed: a position that cannot produce a winner.

One could argue the press means a drawn position, not a stalemate. Oxford’s general definition already covers both: neither side can win. Stalemate isn’t the only way a chess game ends with no winner. Neither side may have sufficient advantage to deliver checkmate against the other. And if the same position comes up three times, a draw can be claimed.

Neither fits Iran. Iran can’t stop the blockade. And nothing is repeating. Every week, the position gets worse for Tehran.

A won game is not a stalemate

Anyone who plays chess knows the difference between a stalemate and a won game that isn’t over yet. A player can reach a position where victory is certain and still need 10, 20 or 30 more moves to finish it.

The winner tightens the screws. The loser shuffles pieces and looks for a trick. Sooner or later the loser resigns or gets mated. Nobody at the board calls it a stalemate because mate isn’t three moves away.

Iran is just not folding so easily. A losing side can drag out a lost game for a long time. That doesn’t change the result, and it doesn’t make it a stalemate.

The losing side’s only real hope is that the winner gets careless and blunders into an actual stalemate. Chess players have a word for that: a swindle.

What a real stalemate looks like

Korea was a stalemate. After China entered in late 1950, the front froze near the 38th parallel by mid-1951. Neither side could push the other off the peninsula. The 1953 armistice ended the shooting without a peace treaty. The two Koreas are still technically at war.

North Korea’s nuclear weapons made that stalemate permanent. Nobody tries regime change on a nuclear power. It also has two patrons on its border. China is one. Russia, linked by rail across the Tumen River and by a 2024 mutual defense pact, is the other.

Vietnam was a stalemate. America could not invade the North without risking a war with China, which sent several hundred thousand support troops to make the point. It could not bomb Hanoi into surrender while Chinese and Soviet supplies kept flowing. Both roads to victory were closed, and the side with patrons behind it could wait.

Iran has none of that. It has no bomb. No great power will send armies to defend Tehran. An American invasion would be enormously costly. Iran is nearly four times the size of Iraq, with about 90 million people. Nobody in Washington wants one.

But nobody is stopping it either. An option you don’t want to use is not an option you can’t use.

Two blockades

There are two blockades in the Persian Gulf, and the press mostly talks about one of them.

Iran is harassing Gulf shipping. It is only partly working. Saudi Arabia and its neighbors are rerouting through bypass pipelines, U.S.-escorted tankers and ship-to-ship transfers outside the strait. Their oil costs more to move. It still moves.

The United States is blockading Iran’s ports. That one is working. Iran has survived sanctions for decades, but sanctions leak. Iran kept selling oil under them, mostly to China, through shadow-fleet tankers and middlemen. A regime can work around paperwork. It cannot work around a destroyer.

Since July, not a single Iranian crude cargo has gotten out. I laid out the rest of the numbers in my satire, The Media Was Right: Trump Is Lost and Desperate for an Off-Ramp.

Iran has land borders, and trucks and smugglers will keep a black market going. They will not replace oil revenue for a state of 90 million people. The sea is where the money comes from.

The results are on the currency board. The rial has fallen to roughly 2.7 million to the dollar. Consumer prices in September were up nearly 90% from a year earlier. The central bank is injecting up to $2 billion to hold the line.

One side’s oil is rerouted. The other side’s oil is cut off. That is not a deadlock.

The leaking dam

Iran has not folded yet. That is true. A dam with a leak in it hasn’t collapsed yet either.

A reporter takes a photograph, sees the dam still standing and calls it a standoff. An engineer looks at the leak and the water pressure and knows where it ends. Unless somebody plugs the leak, the dam goes.

Iran cannot plug this one. It cannot break the blockade. It cannot run a country with no oil money. In a stalemate, waiting changes nothing. Here, every week of waiting makes Iran weaker.

Back in April I laid out The Three Clocks Running Against Iran: economic, geological and relevance. All three are still running.

The word is the story

The press knows all this. The facts are in the papers. The economics desks document a country being bled out. The war desks call it a stalemate.

The narrative is leaking too. The facts have become too obvious to leave out, so they appear in the body and the occasional analysis piece. That way, when Iran folds, nobody ends up with egg on their face. The headline still says stalemate.

Take the oil moving through Hormuz. For months, the press threw cold water on the administration’s claims that tankers were getting through. Now the independent tanker trackers put Gulf flows at roughly three-quarters of prewar levels, and that number is creeping into the coverage. It strains credulity that newsrooms didn’t know. It’s one more leak.

Accurate words were available. “Iran escalates as blockade tightens.” “Tehran lashes out.” The press picked the one word that tells readers nobody can win.

That word does work the coverage never has to do out loud. Kissinger wrote in 1969 that in Vietnam, the guerrilla wins if he does not lose, and the conventional army loses if it does not win. Call the Iran war a stalemate and the reader finishes the thought: America isn’t winning, so America is losing.

There is an old saying in advertising: repeat your message until the public is singing it in their sleep. The word stalemate is now ubiquitous. Wire services, newspapers, cable news, roundtable shows and think tanks all use it.

It’s the news business, after all, and the business right now is the midterms. Once every two years the public gets a lever to pull nationwide, a chance to hand out grades and punishment. Who holds the gavel in January matters to everyone, so every story gets run through one question: does this help or hurt before November? It’s no secret that the mainstream press detests Trump, and he detests them. A war Iran is slowly losing doesn’t reflect badly on Trump. A war stuck in stalemate does. That’s why you hear it on every network. Repeat it often enough and it becomes the only word anyone uses.

The midterms won’t rescue Tehran anyway, as I argued in What Does Iran Think Will Happen If Democrats Win the Midterms?

Nor are they on Trump’s calendar. See Iran: The Midterms Are Not on Trump’s Calendar.

The stalemate story doesn’t need to be true to dominate. It just needs to be more compelling, as I wrote in Is Trump Cornered, or Is He Managing a War He’s Already Winning?

The Oxford dictionary defines gaslighting as making somebody believe untrue things in order to control them, especially that they have imagined or been wrong about what really happened. IMO, the control here is the vote. Convince the public the war is stuck, and a Democratic Congress starts to look like the check Trump needs. Put a currency in freefall next to the word stalemate, repeated story after story, and decide for yourself what to call it.

And Tehran reads the coverage. I wrote about what that has already cost Iran in The Hall of Mirrors: What the Coverage Cannot Stop.

Honest caveats

• Regimes can absorb a great deal of pain if they are willing to make their people eat it. Nobody knows how long Iran’s will hold.

• Iran’s attacks cost real money. Oil at $104 is a real problem for an administration facing midterms, and Trump has said publicly he won’t strike Iran before the election.

• The Houthis are less cautious than their patron. One serious incident could widen the war in ways nobody plans.

• A won game can still be thrown away. Stalemate is what happens when the winner gets careless.

So what is a stalemate, anyway?

In chess, stalemate is what saves the losing side. In the news, it’s what Iran’s collapse is being called.

It is the losing king’s last hope. Iran is playing for one. The press is already printing it.

This is not a stalemate. This is a country losing a war.

You’ve seen the evidence. So you tell me: why does everyone keep using the word?

Addendum: earlier pieces in this series

• The Hall of Mirrors: What the Coverage Cannot Stop (Apr. 14, 2026)

• The Three Clocks Running Against Iran (Apr. 25, 2026)

• Iran: The Midterms Are Not on Trump’s Calendar (June 8, 2026)

• Is Trump Cornered, or Is He Managing a War He’s Already Winning? (Aug. 16, 2026)

• What Does Iran Think Will Happen If Democrats Win the Midterms? (Sept. 9, 2026)

• The Media Was Right: Trump Is Lost and Desperate for an Off-Ramp (Sept. 30, 2026)

Two faces

Anti-Semitism has worn many faces over the long history of the Jews. Two matter here.

The first: you know Jewish people and you don’t like them. That is real, and it is common. It is the same dislike any group gets from people who don’t like anyone different from themselves. It is ugly, but ordinary. It’s common to pick on minorities.

The second is the one that has gotten people killed. Something goes wrong. The real cause is something you don’t want to accept, so you find a cause you can accept. For two thousand years, that cause has been the Jews.

The word is scapegoat.

The plague

In 1348 the Black Death was sweeping Europe. It would kill a third or more of the population. The explanation that spread was that the Jews were poisoning the wells. Jews were tortured into confessing. Communities across France, Spain, and the German lands were massacred.

Pope Clement VI issued two official decrees that year telling Christians to stop. His argument was evidence. The Jews were dying of the plague too.

It didn’t matter.

Strasbourg’s council protected its Jews. On February 9, 1349, the guilds forced the city’s leaders from office. Five days later, Strasbourg burned its Jews on a wooden platform in their own cemetery, about 2,000 people by the chronicler’s count. The plague had not yet reached the city. The chronicler Jacob von Königshofen recorded what came next: debts owed to the Jews were cancelled, and their cash was divided among the townspeople.

The explanation did not need to be true. It needed to be useful.

The pattern

It repeats. Germany lost the First World War, and that was unacceptable. So Germany wasn’t defeated. It was stabbed in the back, and the knife was Jewish. A forgery, The Protocols of the Elders of Zion, explained everything from wars to depressions. One cause for every problem is the giveaway.

It hasn’t gone away, and it doesn’t belong to the left or the right.

On the right: in 2018, before she was elected to Congress, Marjorie Taylor Greene posted that California’s deadly Camp Fire might have been started by a beam from a space-based solar generator. She noted that a PG&E board member was vice chairman of Rothschild Inc. (Jewish conspiracy?)

Investigators found the fire was started by PG&E power lines.

On the left: starting in 2017, Jewish Voice for Peace (a left-wing anti-Israel activist group) ran a campaign called “Deadly Exchange,” which tied American police violence to exchange trips that took American police officials to Israel. American police violence has American causes.

These are just a few of hundreds of tropes, old and new. Now Jeffrey Epstein gets tied to the Mossad, and the Rothschilds get blamed for everything.

Who drew the map

The history of the world is littered with the vestiges of European imperialism. Britain, France, Spain, Portugal, the Netherlands, Belgium, Germany, Italy, and others plundered the world for what they wanted until they could no longer do it. In the Middle East, Britain and France carved up the Ottoman lands, first in secret under the 1916 Sykes-Picot agreement, then as League of Nations mandates. They drew the borders and installed the kings.

Israel didn’t exist when the map was drawn, and the United States didn’t draw it. Yet those two are the modern bogeymen.

European imperialism is resented, and not without reason. But Israel was no part of it, and the Jewish people didn’t originate in Europe. In 1939, on the eve of the Holocaust, Britain capped Jewish immigration to Palestine at 75,000 over five years. The empire that drew the map closed the door on Jews fleeing Hitler.

Native, not foreign

Jews are not Europeans who moved to the Middle East. Their history, language, religion, and archaeology are rooted in that land, and genetic studies trace most diaspora Jewish communities to the Levant. Rome destroyed the Temple in 70 CE, crushed the last revolt in 135, barred Jews from Jerusalem, and renamed the province Syria Palaestina. Jewish communities in the land never entirely disappeared.

Everywhere else, the world wants to protect indigenous peoples. Here, that standard is suspended. Nobody says American Indians, Aboriginal Australians, or New Zealand’s Maori stopped being indigenous because some of them married outsiders.

For generations, European anti-Semites told Jews to go back to Palestine. Now the slogan is go back to Europe. The charge reversed. The target didn’t.

When an accusation can flip to its opposite and the conclusion stays the same, the accusation was never the point.

How the state of Israel came to be

In 1947, the UN voted to divide the British Mandate of Palestine into two states, one Jewish and one Arab. The Jewish leadership accepted the plan. The Arab states and the Palestinian Arab leadership rejected it, and war followed. From 1948 to 1967, Jordan held and annexed the West Bank, and Egypt held Gaza. Neither created a real Palestinian state. I’ve covered this history in Understanding the Israeli–Palestinian Conflict, Part 1, Part 2, and Part 3, and the 1947 vote in UN Resolution 181 for Palestine: The Stupidest Resolution Ever Passed.

How is it Israel’s fault that the Palestinians don’t have their state?

Meanwhile, elsewhere

Israel has been condemned since the day it was founded. Look at what else the world has lived through in that time.

In 1975, the UN General Assembly declared that Zionism is a form of racism. That same year, the Khmer Rouge took over Cambodia. They went on to kill somewhere between 1.5 and 2 million people.

In 1988, Saddam Hussein gassed the Kurdish town of Halabja, killing about 5,000 people in a day. In 1994, Hutu extremists in Rwanda slaughtered about 800,000 people, most of them Tutsi, in 100 days. Starting in 2003, Sudan’s government and its militias waged a campaign in Darfur that left an estimated 300,000 dead from violence, disease, and hunger. Bashar al-Assad’s war on his own people set off a conflict in Syria that killed more than half a million. In 2014, ISIS tried to wipe out the Yazidis, murdering thousands and selling women and girls into slavery.

It hasn’t stopped. In January 2026, Iran’s regime gunned down its own people in the streets. Senior officials in its own Health Ministry put the toll at around 30,000, and opposition estimates run as high as 50,000. Sudan’s new civil war, now in its fourth year, has killed tens of thousands, by some estimates well over 100,000, and driven more than 11 million from their homes. Russia’s unprovoked war has killed or wounded hundreds of thousands of Ukrainians, including more than 17,000 civilians killed, by the UN’s count.

Where is all the continued condemnation?

Who’s in the neighborhood

Saudi Arabia has no free press, no free speech, and no national elections.

On October 2, 2018, Washington Post columnist Jamal Khashoggi, a Saudi citizen living in exile in Virginia and a frequent critic of his government, walked into the Saudi consulate in Istanbul to get documents he needed to get married. A team sent from Riyadh killed him and dismembered his body.

Saudi women couldn’t drive until June 2018. A few weeks before the ban ended, the kingdom arrested the women who had campaigned to end it. The best known, Loujain al-Hathloul, spent nearly three years in prison.

Where were the campus encampments for Saudi women? For women in Afghanistan, where the Taliban won’t let girls go to school past the sixth grade? Zohran Mamdani, now mayor of New York, co-founded the Students for Justice in Palestine chapter at Bowdoin College and built a political career opposing Israel. Where was his campaign for them?

The soft war

People watching the news can feel empathy for the civilians in Gaza and the West Bank caught up in this. They should. But they are watching the tail end of a long history, and few of the people objecting to Israel know much of it.

Someone is happy to teach them. Qatar is the largest foreign funder of American universities, with $6.6 billion in disclosed gifts and contracts, much of it for American branch campuses in Doha. Qatar’s emir also founded Al Jazeera, and the Qatari government still funds it.

Then there’s the UN, which is less about peace and unity than politics and old complaints. Its Human Rights Council has one permanent agenda item aimed at a single country. That country is Israel. (More in What the UN Is — and What It Is Not.)

This soft war against Israel has paid big dividends in shaping public opinion.

The new Democratic Party

“Some of my best friends are Jewish.” It comes from Jews and non-Jews in the Democratic Party and its leadership, and it doesn’t fool anyone.

Their pragmatic big tent has room for anti-Semitism on their road back to power.

I’ve laid out the record before, in Antisemitic Rhetoric Will Only Hurt the Democratic Party and Ro Khanna’s Publicity Stunt in Israel.

The question

So is anti-Israel the same as anti-Semitic?

Answer this for yourself.

A pattern that is more than 2,000 years old deserves to be taken seriously. Its motto is simple: when you need a scapegoat, the Jews will do.

Addendum: earlier pieces

• Understanding the Israeli–Palestinian Conflict, Part 1

• Understanding the Israeli–Palestinian Conflict, Part 2

• Understanding the Israeli–Palestinian Conflict, Part 3

• UN Resolution 181 for Palestine: The Stupidest Resolution Ever Passed

• No Going Back: What the Palestinians Must Accept Before It’s Too Late

• Antisemitic Rhetoric Will Only Hurt the Democratic Party

• Why Is Hamas Finally Talking About Giving Up Power?

• The Gaza Health Ministry Reports…

• Iran and the Little Satan/Great Satan Con

• Before You Ask for a Palestinian State, Define What You’re Asking For

• What the UN Is — and What It Is Not

• 9/11 vs. October 7: How Nations Really Respond to Attack

• Ro Khanna’s Publicity Stunt in Israel

• Unboxing the UN Plan for a Palestinian State: What the New York Declaration Actually Says

Correction: An earlier version of this piece said Mayor Mamdani had not pledged to arrest Vladimir Putin. He did, in a September 2025 interview. That passage has been removed, and several figures have been made more precise.

Every few months the alarm goes off again. Kids can’t read. Kids don’t read. The Nation’s Report Card says only 14% of 13-year-olds read for fun almost every day. In 1984 it was 35%.

Those get treated as one problem. They aren’t.

This piece isn’t claiming kids shouldn’t read, or that novels are worthless. It’s claiming something narrower: reading fiction shouldn’t be a yardstick for a good student or a good education.

Two Problems, One Alarm

The first problem is real. A child who can’t decode words or follow a paragraph in a science book is in trouble, and so is the school that keeps promoting him.

The second problem is a taste. A kid who would rather watch a movie than read a novel can still read a lease, a manual and a textbook. He just doesn’t like novels. That isn’t a literacy crisis. It’s a preference.

The alarm folds the second problem into the first, and the novel gets sold as part of the cure.

Latin

For centuries Latin was required. Oxford and Cambridge demanded it for admission until 1960. The case wasn’t that students would use Latin. It was that Latin trained the mind. Logic, discipline, memory, precision. Learn it and you’d be better at everything else.

In 1924 the psychologist Edward Thorndike tested that claim on high school students. The Latin students gained no more general reasoning ability than students who took other subjects. Later research has largely backed him up.

The requirement outlived the evidence by decades.

Schools eventually stopped requiring the classics. Then they gave the name to something else. “The classics” now means Dickens in English class, and the requirement came with the name.

Fiction’s defense sounds familiar. Novels build empathy, imagination, attention. They’re good for you whatever you do in life. In 2013 a study in Science reported that reading a short passage of literary fiction improved people’s ability to read others’ minds. It got wide coverage. Most of the replications that followed failed to find the effect, including one with far more participants. A couple confirmed it.

The empathy story is starting to look like the mind-training story.

I’ve known plenty of lifelong literature fanatics who weren’t particularly creative or imaginative at all, nor solid thinkers. But they really enjoyed reading novels through their whole lives.

Dickens Was Television

Dickens wasn’t homework in his day. His novels came out in monthly installments that sold for a shilling each, when a bound book cost more than thirty. Working families could afford him. Some pooled their money for one copy and read it aloud. Readers waited for the next installment the way people now wait for the next season of a show.

He was the popular entertainment of his time. His stories haven’t gone anywhere. They moved to the popular entertainment of ours, and get retold on film and television every few years.

A kid who knows Scrooge from a movie knows the story. What he’s missed is the prose. That’s a real loss for anyone who loves prose. It isn’t a skill gap.

You might learn just as much, if not more, from watching Pulp Fiction, or The Talented Mr. Ripley. In my opinion that movie is much better than the book.

Many books have a great premise and became great movies, but they drag as prose. The movie was a better and more engaging presentation of the premise.

I Love Opera

I love opera and classical music. I sat through years of required music classes in school and university, listening to it and learning about it. None of it took. I found it on my own, long after. Most kids never learn to like Mozart, and years of music class don’t change that.

Some will take to it on their own. For the rest, start where the kid is. A kid making beats in music class is learning rhythm, structure, and how a piece builds and resolves. In 2002 Nas built “I Can” on Beethoven’s “Für Elise.” A kid who loves that track might go find the original. A required listening of Beethoven was never going to send him there.

Fiction works the same way. Hand every kid a good novel. Some will read it and never stop. Others will always prefer their stories from movies or elsewhere. I don’t see anything wrong with that.

What Forcing Teaches

School doesn’t have to be fun. A kid who never gets fractions loses algebra, and a long list of jobs with it. Fiction doesn’t.

But kids can tell the difference between a skill they’ll use and a taste someone is grading them on. Gallup has surveyed millions of students since 2009. About three-quarters of fifth graders are engaged with school. By 11th grade it’s about a third. Older students are far less likely to say adults at school care about them, or that teachers make them feel their schoolwork is important.

Essential skills get taught even when they’re boring, with a reason attached. Tastes get offered. When both are treated as non-negotiable, kids conclude that school has little to do with their lives. Then they tune out the algebra along with the novel.

Who Keeps the Requirement

English bachelor’s degrees fell 43.8% between 2012 and 2025, according to the American Academy of Arts and Sciences. Humanities degrees are at their lowest number since 1991.

A required course is the one enrollment a department doesn’t have to compete for.

Honest Caveats

Books are richer in language than screens. A 1988 study counted rare words per thousand and found children’s books had more than prime-time television or conversation among college graduates. A kid who reads a lot picks up vocabulary a movie won’t give him. Nonfiction gives it too.

Kids who read for fun score higher on reading tests. That runs both ways. Good readers enjoy reading.

Exposure matters. Some kids find out they love novels only because a teacher handed them the right one. That’s an argument for a sampler, not a mandate.

Tube Amps

Vacuum tubes were once the future of electronics. Then transistors replaced them.

Tubes didn’t disappear. Guitarists still swear by tube amps, and audiophiles pay a premium for tube stereos. The people who love them keep them alive. Nobody has to take a course in them.

Fiction will be fine. Some kids will always love the adventure of reading and filling in the details with their own imagination. The rest can get their stories somewhere else.

It just shouldn’t be the yardstick.

For months the experts told us how this would end. The war was a quagmire. Gas prices were unsustainable. The midterms were closing in. A cornered president, out of options, would soon grab any exit Tehran offered.

They were right. It’s time we admitted it. Let’s review just how desperate Donald Trump has become.

The Offer He Couldn’t Refuse

Last week, on the sidelines of the United Nations, Iran offered a way out. Reopen the Strait of Hormuz in seven days. In exchange, the United States would simply end the blockade, unfreeze billions in Iranian assets, lift sanctions, and stop all hostilities in Iran and Lebanon.

A reasonable package. A lifeline for a drowning man.

Trump, desperate, said no. Iran only wants a deal, he explained, because it’s “losing so badly.” Classic projection.

Then Axios reported he had privately offered sanctions relief. Surely this was the cave. Trump, desperate, denied it.

The Blockade That Broke Him

Consider the pressure he’s under. Since the blockade was reinstated in July, not a single Iranian crude cargo has made it through. Iran’s oil stored at sea has fallen from about 90 million barrels to about 15 million, and analysts expect it to run out by mid-October.

Meanwhile, everyone else’s oil is moving. Gulf exports are back near 80 percent of prewar levels, the highest since the war began. The only tankers stuck in the Gulf are Iran’s.

Faced with this, Trump did what any panicking man would do. He left the blockade exactly where it was.

Meanwhile, the Iranian rial hit 2.5 million to the dollar. The Iranian government is warning its own northern provinces about fuel shortages this winter. Gas prices there have doubled in a month.

Clearly signs of American weakness.

The Clock

Iran’s president went on NBC to say Tehran wants a deal before the November 3 midterms.

This is the pressure the experts warned about. Iran can’t wait for the election. And Trump, terrified, has only 28 more months in office to wait out Iran.

The polls say Republicans may lose the House. A Democratic House would hold hearings. It might pass resolutions. It could not end the blockade without veto-proof majorities in both chambers, which exist nowhere outside a cable-news panel.

Tehran’s foreign minister says Iran is ready for a “doomsday war.” Iran’s oil is ready for mid-October.

The Cheering Section

Every war has a home front. Iran’s turned out to be surprisingly well staffed, much of it in Manhattan and Washington.

Every American casualty was a turning point. Every gas-price spike was the beginning of the end. Every anonymous official with a grievance became a front-page source. Somehow the rial’s collapse never quite made the cut. Neither did the eleven straight weeks without an Iranian tanker getting out.

We’ve seen this before. When inflation hit 9 percent during Biden, it was transitory, then Putin’s fault, then a vibe. When gas went up under Trump, it got the first five minutes of every newscast and the whole episode of Washington Week and Meet the Press.

The anonymous sources knew everything that the administration was refusing to admit.

The Off-Ramp

So yes, the experts were right. Someone in this war is desperate for an off-ramp. Someone is burning through reserves, pitching peace plans at the UN, sending proxies after Saudi pipelines and Red Sea shipping, and praying for an American election to rescue them.

Somewhere, a cable-news panel is trying to say “we were wrong.” Like the Fonz on Happy Days, so far they’ve gotten as far as “we were wrrr.”

This week Uzbekistan won the Chess Olympiad in Samarkand, finishing ahead of India. It is the second Olympiad gold in four years for a country of about 39 million that, a generation ago, was a provincial outpost of the Soviet chess empire. When the Uzbeks won their first, in Chennai in 2022, the team’s average age was about 20, and its youngest player was still in high school.

That is not an anomaly. It is the new normal, and the ages have become absurd.

In 1958 Bobby Fischer became a grandmaster at 15 and a half. It was treated as a freak of nature, and the record stood for 33 years. Today a 15-year-old grandmaster is barely news. Sergey Karjakin did it at 12 years and 7 months in 2002. Abhimanyu Mishra, a kid from New Jersey, did it at 12 years and 4 months in 2021. This May, Faustino Oro of Argentina missed Mishra’s record by about two months. India’s Gukesh Dommaraju made grandmaster at 12 and became world champion in 2024 at 18.

Something changed. To see what, you have to start with what chess always was.

A poor country’s game

Chess is like soccer. The equipment costs nothing. A board and pieces, or a ball, and you’re in. That is why both games have always attracted poor countries and ambitious families.

And the knowledge was never secret. Nimzowitsch’s My System came out in the 1920s. Capablanca’s books were everywhere. From 1966, Chess Informant, published in Belgrade, shipped the best games in the world, mostly annotated by the players themselves, to anyone with a subscription. A determined kid in Ohio could read the same material as a kid in Leningrad.

Yet from 1948, when Mikhail Botvinnik won the title, the Soviet Union held the world championship without interruption until Bobby Fischer took it in 1972. Three years later it was back in Soviet hands, and it stayed there until the Soviet Union itself was gone.

The shift had started in 1927, when Alexander Alekhine took the world title from José Raúl Capablanca. Capablanca was the great natural talent of the age, and Alekhine had never beaten him in a single game. Capablanca arrived in Buenos Aires with no real preparation. Alekhine had studied Capablanca’s games for weaknesses and gotten himself into physical shape. He won six games to three, with 25 draws. From then on, walking into a title match unprepared made as much sense as a boxer walking into the ring without training.

The Soviets didn’t have better books. They had an institution. In the 1920s the state sent masters across the country to train players. Botvinnik trained like an athlete, with timed daily walks, breathing exercises, and practice games in which his partner blew cigarette smoke at him so a smoky tournament hall wouldn’t bother him. He built a school that produced Karpov, Kasparov and Kramnik. Strong juniors got coaches, full-time support, and, most importantly, a dense supply of strong opponents to play every week, with experts on hand to tell them what they did wrong.

That was the scarce input. Not information: structured practice with feedback. Nobody else built a chess college, so nobody else produced a steady line of world champions.

The Polgár experiment

In 1965 a young Hungarian named László Polgár courted a language teacher named Klára by mail. His letters weren’t love poems. They laid out a plan. He had studied the childhoods of hundreds of geniuses, concluded that “geniuses are made, not born,” and wanted a wife to help him prove it with their own children. She married him anyway.

They picked chess because it offered what almost no other field did: an objective, public measurement nobody could argue with. They homeschooled their three daughters, Susan, Sofia and Judit, and trained them for hours a day.

In an era when almost everyone believed women simply couldn’t play at the top level, the results were a demolition. Susan became a grandmaster in 1991. So did Judit, that December, at 15 years and 4 months, breaking Fischer’s 33-year-old record by about a month. Sofia, the middle sister, stayed an International Master, but at 14 she won a 1989 open tournament in Rome with 8½ out of 9, beating four grandmasters, three of them Soviet. The chess press called it the “Sack of Rome.” After 1990 Judit never played in a women’s event again; she competed only in open tournaments. She entered the world’s top 100 at age 12 and eventually reached No. 8 in the world, with a peak rating of 2735. In 2002 she beat Kasparov, then the world’s No. 1, in a rapid game. No other woman has crossed 2700 or entered the top ten.

The Polgárs destroyed the belief that girls’ brains couldn’t do it. The limit had been set by society: by what people expected of girls and by what anyone was willing to invest in them.

But look at what it took. Two parents who gave up their careers and turned their home into a chess academy. One obsessive father, three daughters. The method worked. It didn’t scale.

The chess college in every pocket

After losing his title in 1963, Botvinnik, an electrical engineer by training, spent years trying to build what he called an “electronic grandmaster.” He never got there. Others did. The founder of the Soviet chess school spent his last decades working on the machine that would, in other hands, give his school away to every kid with a phone.

A child today has everything the Soviet system rationed. An opponent at any strength, at any hour. Every game checked by an engine far stronger than any human, so every error is found the same night. A rating that says exactly how good the child is and whether he is improving, with no grading on a curve and no social promotion. And unlimited volume: thousands of games and puzzles a year.

Nothing about the books changed. What changed was the feedback loop, and it became almost free.

The results showed up exactly where you would expect if talent had always been spread far more widely than the old pipelines could find it. India had one grandmaster in 1988, Viswanathan Anand. It now has 99 and is about to reach 100, and its top ten players average over 2700, second only to the United States. One of this year’s new ones, Aaryan Varshney of Delhi, got there with no academy and no professional coach; his father was his trainer. Uzbekistan wins Olympiads. Argentina produces a 12-year-old grandmaster.

The machine wasn’t the only cause. India had Anand as a role model, the pandemic pushed the whole game online, and there are far more rated players and norm tournaments than there used to be, which makes titles easier to chase. The top Indian kids still have coaches. But the engine is the accelerant under all of it: the coach now works on top of a machine that supplies most of the repetitions.

Chess was always cheap. Mastery wasn’t. The computer made mastery cheap.

Chess is the measuring stick

The same thing is happening in computer programming. Teenagers with a laptop and an internet connection build software that used to take a team. But programming has no rating. “How good is this kid?” is a matter of opinion.

Chess is different. Since FIDE adopted the Elo system in 1970, it has kept an objective, continuous record of human performance, the closest thing we have to a controlled experiment on how people learn. And that record says something unambiguous: when practice with immediate, accurate feedback becomes unlimited and cheap, the time it takes a talented child to reach world class collapses by years, and talent starts showing up everywhere. The machines didn’t raise the peak much; a steady 2800 is still rare. What they did was shorten the road and open it to far more people.

Yes, chess is the best case. It has fixed rules, a right answer the machine can find, and a score nobody can argue with. Most subjects are messier. But that cuts both ways: we ran the experiment in the most favorable domain imaginable, and even there nothing happened until the feedback became free. Once it did, a century of assumptions about who could reach the top collapsed in about twenty years.

Enter Claude Code to close the gap

The engines and the online games supplied the practice and the feedback. They did not supply László Polgár: the coach who watches one child, sees what he keeps getting wrong, and decides what comes next.

Books made knowledge cheap. Engines made feedback cheap. Direction stayed expensive.

Claude Code already does that kind of work on software. It can develop and improve code on its own. It does not care what state the software is in, from brand new to legacy. It reads what is there, finds the issues and fixes them, checks the result, and tries again. It is the chess engine’s loop of attempt, check and correct, applied to something far messier than a chessboard.

Why can’t the same technology teach?

Pointed at a student instead of a program, it would be a tireless personal tutor. It meets the student wherever he is and adapts to him. A student can be tested, given exercises to improve, and retested. That is how it works on a program: adding code, debugging, testing again. In 1984 the educational psychologist Benjamin Bloom found that the average student taught one-on-one outperformed 98 percent of students taught in a conventional classroom. Nobody doubted that tutoring works. The problem was always the cost. Whether it takes 4 years or 10 years to get the degree, based on the student, does not really matter. That only matters when you need a coordinated program, as in today’s schools and universities.

The point

There is always the unusual person who figures it out mostly on his own, the Bobby Fischer. But that is one in a billion.

Many people can do much better than anyone expects, given the proper environment.

In this story, the proper environment was László Polgár. That cannot scale across humanity.

The tools, the modern chess engines and related apps and the world of online competition around the clock, are available to everyone.

The challenge is to use AI to help fill the gap beyond the chess engines and online games that everyone has access to now.

It seems like the gap left in the chess teaching is ready to be filled and applied across the education spectrum.

Books made knowledge cheap. Engines made feedback cheap. AI can make coaching cheap.

Everyone knows how to settle it. Send people in, dig it out, fly it home. Former NATO commander James Stavridis told the Wall Street Journal it could be the largest special operations mission in history. It would be under drone and missile fire, near a major city. Asked about ground troops, Trump said the US wouldn’t do it now, “maybe later.”

The uranium is still there. The problem wasn’t finding the target. It was the boots.

That problem is going away.

This isn’t about how to fight China. It’s about how to deal with badly behaving populations of 50 million or more when there’s no appetite for 200,000 boots on the ground.

There are many badly behaving countries in the world with populations large enough that we can’t send boots in all the time to police them. Even a few million people is hard to control.

The Real Constraint

The US didn’t lose Iraq or Afghanistan on the battlefield. It took Baghdad in three weeks. It lost the will to stay. Nearly 4,500 Americans died in Iraq and more than 2,400 in Afghanistan.

After that, every American threat carried an asterisk. Air strikes, yes. Ground troops, no. Iran has 90 million people. Nobody in Washington proposes to occupy it, and Tehran knows it.

Winning by air is difficult. Air power can destroy a building. It can’t hold a town, search a tunnel, or make sure the thing it hit stays dead.

If we wanted to control bad behavior in Afghanistan, we didn’t need nation building. With a robot army, we could have policed the camps and left the country to itself.

The Robots Are Already Here

Ukraine is building the answer now, at scale. Its ground robots ran about 112,000 frontline missions in the first eight months of this year, more than 25,000 of them in August alone. Ukraine has contracted more than 22,000 ground robots this year. The Defense Ministry’s stated goal is for robots to handle 100% of frontline logistics.

They haul ammunition, food and water. They evacuate the wounded. Every mission is a soldier who didn’t walk that road.

A robot costs $2 million when you build a thousand. Those are special order. When you build 10 million you are getting consumer product prices.

Ukraine is building tens of thousands. In the first four months of this year it spent about $330 million through its frontline ordering system alone on more than 181,000 drones, ground robots and electronic-warfare systems, under $2,000 apiece.

The supply line can be robotic too. From 2011 to 2014 the Marines flew unmanned K-MAX cargo helicopters in Afghanistan. They lifted more than 4.5 million pounds and flew thousands of missions without a pilot on board. That was over a decade ago.

Robots that break can be picked up by robots and flown back to be fixed. Extend that far enough and you get robotic aircraft carriers with their own robotic support. The tail that used to need thousands of people doesn’t.

Making a robot army is not an overnight project.

The Threat That Works

In 1963 John F. Kennedy warned that the President in the 1970s might face 15, 20 or 25 nations with nuclear weapons. Sixty years later there are nine.

Part of the reason is force. Israel destroyed Iraq’s reactor in 1981 and Syria’s in 2007. The US and Israel struck Iran’s program in June 2025 and again this year.

Part of it is timing. The 2007 National Intelligence Estimate concluded Iran halted its nuclear weapons program in the fall of 2003. American troops had reached Baghdad that April. In December 2003 Libya agreed to give up its weapons programs.

The threat was credible then. It stopped being credible when the public stopped accepting casualties.

Iran wouldn’t have grown to where it is if this technology had been available. Diplomacy that avoided conflict could have been a simpler message: a threat to go in with a robot army.

North Korea is the same story. There was no appetite to go in, and sanctions don’t work well on large populations. It tested its first nuclear weapon in 2006.

Robots put the credibility back. A President who can send a robotic force into Isfahan, dig out the uranium and bring it home, without a single flag-draped coffin, doesn’t have to say “maybe later.” And the other side knows he doesn’t.

We don’t even need robots to dig out the uranium. We can monitor the land over it. If the Iranians try to dig tunnels to reach it, we kill them. There are lots of ways to guard things. The robots become a physical lock on the site: a kill zone over the way in, not an occupying army.

Look at What China Is Doing

China is building real robot armies now, and putting them on display.

At its Victory Day parade in Beijing in September 2025, the People’s Liberation Army rolled out armed unmanned ground vehicles, a robotic mine-clearing vehicle and “robot wolves,” four-legged combat robots. In January, state television released footage of armed robot dogs in training. PLA exercises have used robot wolves in urban-warfare drills and beach landings, sending them in first to take the initial fire from coastal defenses.

Some are built on commercial four-legged robots like Unitree’s Go2, fitted with rifles or rockets.

China may be ahead of us in that thinking.

The US Army halted its Robotic Combat Vehicle program in 2025 and reopened it to industry. Developers told National Defense magazine this year that large-scale autonomous ground operations are at least a decade away.

In Summary

This isn’t how to fight China. That needs army level firepower.

This is for worldwide police work without boots on the ground.

Machines have been fixing text almost as long. Ralph Gorin wrote the first spelling checker at Stanford in 1971. Grammatik brought grammar and style checking to personal computers in 1981, and Microsoft Word built in a grammar checker in 1992. Long before ChatGPT, these programs were suggesting changes to make prose better, not just fixing typos. Grammarly, founded in 2009, turned that into a company valued at $13 billion. Nobody asked what percentage of a column the spell checker, grammar checker or prose checker wrote. The Washington Post now requires guest writers to confirm their work was not “created or manipulated with artificial intelligence or editing software.” Taken literally, that bans spell check.

With chatbots like ChatGPT, the question of who wrote this has moved to the forefront.

Enter the AI detector.

What an AI Detector Really Does

An AI detector looks for patterns in the prose that suggest a computer generated it. Pangram, a leading detector, was trained on millions of examples of human and AI writing until it learned to tell them apart. “We’re reverse-engineering their stylistic fingerprint,” its CEO, Max Spero, told The New York Times. It doesn’t read for meaning. It doesn’t check facts, judge arguments or ask whether anything in the piece is new.

For example, when I first started writing my op-eds, the chatbots kept giving me the same cliché: something “in the clothing of” something else. It showed up constantly, and it was very annoying. I don’t see it at all now. I’m guessing Anthropic noticed it too and stopped Claude from doing that. That would be a simple example of how a detector might know. A human writer would not use it so much, and certainly not constantly in every piece. The harm is just that it makes for boring and weak writing. A human reader will notice bad writing.

The things that irritate me most in AI-generated text, I also find in human writing: lots of padding for length and overexplaining. I constantly have to make the chatbots trim. They learned it from reading newspapers and books that sell by the pound, and apparently there is enough of that out there that they picked up the habit. So if you start looking for it, you will get the same “bad writing” from humans or AI.

What Problem Are We Trying to Solve Here?

One of my bosses always asked, whenever someone suggested some new software to write: What problem are we trying to solve here? Quite often the room got very quiet. Nobody had really thought about that.

In trying to detect what role AI had in a piece of writing, we should ask the same question.

Possible Problems This Can Solve

People who want AI detection have real reasons. Here are the main problems they hope it will solve:

1. Schools want students to do their own writing. Why do you care? They will have access to these tools in life. If they can write well with AI, they can write well.

2. Platforms want to catch meaningless spam: content farms, fake reviews, bot posts. A detector detects AI writing, not spam. It doesn’t look at the content. And humanizers can disguise the sentence patterns it does look at, as we explain later.

3. Readers want to know a real person is behind the byline. Why do you care whether a real person is behind the byline? A detector doesn’t tell you who you’re connecting with anyway. If you need to know, there’s DocuSign.

4. Publishers and prize committees want to reward human craft. This is not a chess match. Human craft with the help of AI is what will be out in the world to compete against.

5. Someone has to answer for what’s written, including a fabricated fact or citation. You are already responsible for what you publish.

6. Research needs integrity. Pangram estimated that 21% of peer reviews submitted to a major AI conference this year were fully AI-generated. A detector doesn’t tell you whether the review is correct, which is what matters. And AI is becoming, or may already be, smarter than humans. Why wouldn’t you want something smarter than a person reviewing things?

7. Writers want to protect their livelihoods from free machine text. This is protecting someone else’s business. On a related topic, a detector also spares teachers from grading student work that may be better than they could write themselves. Years of training in long division don’t protect you against a child with a calculator.

On the schools’ point: working out your own thinking with the help of AI is fine. Most people’s parents helped them long before AI. You still have to direct the AI; it doesn’t do it for you. People who haven’t written with it think it’s more automatic than it is. Programmers learn the same lesson: a chatbot can’t do a major software project for you if you don’t know how to do it yourself.

Carpenters use nail guns now. How fast you can put a roof on with a regular hammer is not something worth testing.

Some Purposes of Writing

Before deciding what a detector should protect, it helps to ask what writing is for. People write for many reasons, including these:

1. Create something the reader enjoys.

2. Teach the reader something.

3. Communicate information.

4. Persuade: change the reader’s mind or get them to act.

5. Record what happened, what was decided or what someone knew.

6. Work out your own thinking. Writing forces you to find out what you actually believe.

7. Connect with a particular person: letters, condolences, love notes.

8. Get something done: instructions, manuals, emails, proposals.

Regarding the Purpose of Writing, What Problems Does an AI Detector Really Solve?

Basically nothing.

You want the best possible writing, original ideas, things worth reading. You can ask a chatbot those questions without a special tool. I do this with all my own writing. I ask: “What is original about this piece?” or “Read, Analyze, Critique” or “Thoughts?” (works great).

I also can’t grow my own food. If the grocery stores closed, I’d be in trouble. Nobody thinks schools should ban supermarkets.

In my opinion, most of the proposed AI problems are just typical overreach by people with an agenda orthogonal to the purpose of the thing.

What Problems Does an AI Detector Create?

It creates an artificial grade, assigned by a robot, on the value of the writing and whether you cheated somehow.

It teaches that a poorer piece that shows no AI is considered better.

The Arms Race

Now there is an arms race between AI detectors and AI humanizers. An AI humanizer is designed to make an AI-assisted piece undetectable.

The humanizers are doing well. In a University of Notre Dame study, not yet peer reviewed, once AI-written text went through a humanizer called Undetectable AI, more than 96% of it passed as human. Undetectable AI offers a money-back guarantee if its output gets flagged, and it claims more than 22 million users.

The same study found the opposite problem for honest writers. Research abstracts that had only been lightly edited with AI were flagged 64 to 80% of the time. Pangram says its newest version fixed that; nobody outside the company has checked.

Honest AI editing carries more risk than paid evasion.

Pangram does better against amateur disguises. A peer-reviewed study from Vrije Universiteit Brussel found it caught 37 of 40 AI-written papers that GPT-4o had simply been told to make sound human. Against a real humanizer, the Notre Dame numbers are the ones that count.

Grammarly sells both sides. It offers an AI detector, and a feature that “humanizes” text to get past detectors.

For those playing at home, computer science already has a name for this: a generative adversarial network, or GAN. Each side trains against the other: the detector learns to catch the fakes, the generator learns to beat the detector, round after round. The generator is the one that’s supposed to win. This is how you make counterfeit money with machine learning.

There's another catch. When a humanizer gets past a detector, only the person who used it knows. The detector never learns it was fooled, so nobody can measure how often it misses in real life. Pangram's accuracy numbers come from tests where the right answer is known in advance. And if you're flagged, you can't prove you didn't use AI, any more than the detector can prove you did.

Whose Help?

A ghostwritten book passes. Prince Harry’s memoir Spare was written with J.R. Moehringer. A human wrote those sentences, so a detector scores it human.

Presidents have speechwriters. Celebrities have ghostwriters. Academics have editors nobody credits. Under the Post’s rule, a human editor rewriting your paragraphs is fine. Software doing the same thing is not. Even Pangram’s CEO told the Times that using AI as an assistant is “completely OK.”

Slide Rules

We used slide rules. When calculators arrived, the objection was that we’d never learn the real skill. By 1980 the National Council of Teachers of Mathematics was calling for calculators at every grade level.

In my early twenties, preparing text for publication was a big job. You couldn’t just bang out an article. Then word processors arrived, and I watched the number and size of journals explode. Research later found word processing made student writing both longer and better.

Nobody built a detector for text typed on a computer instead of a Selectric. Journals that got too long set length limits. The fix was norms about the output, not forensics on the process.

Summary

At best, a detector score tells you one thing: AI touched the prose. It can’t tell you whose ideas they are, whether the facts are right, or whether the piece is worth reading. Honest writers get flagged, anyone who pays for a humanizer walks through, and a ghostwritten book passes.

When it comes to AI detectors, readers don’t hear “stylistic patterns.” They hear cheating or spam. Hachette canceled a novel after Pangram’s CEO posted that it scored 78% AI. Dartmouth’s provost, flagged when Semafor scanned more than 300 newspaper guest columns, now faces a campus review. To me this is just a witch hunt.

My old boss would ask: what problem are we trying to solve? For a reader, the only problem is whether a piece is worth their time. A chatbot can help answer that. A detector is not designed to.

Pangram will probably score this piece near 100% AI-written.

It can’t tell you whether it is worth reading.

It also can’t tell you whether Pangram itself has any useful place in the world of writing.

What Happened

In the past few weeks the Houthis took the port of Mocha, then Perim Island, the rock that splits Bab el-Mandeb into two shipping lanes. They announced a blockade of Saudi oil headed to Asia. Saudi Arabia shut its East-West pipeline after drone strikes, and this week Houthi missiles were fired at Yanbu, the Red Sea port where that pipeline’s oil gets loaded.

The President of the United States, standing in Dublin, said the Houthis had called him, didn’t want to fight America, were letting most ships through, and had a problem with only “one country.” His Vice President said the administration was monitoring the situation. Translation: not our strait, not our oil, not our war.

Whose Oil Is It?

This is the serious part, and it’s simple. They need the oil. We don’t.

The United States is a net petroleum exporter. When a Saudi tanker is threatened, Americans feel it at the pump because oil is one global market, but no American refinery goes dry. Texas and North Dakota actually make money.

Now look at who does go dry. Japan imports nearly all its crude, overwhelmingly from the Middle East. South Korea, India and China are in the same boat, literally. Saudi oil loaded at Yanbu goes one of two ways: south through Bab el-Mandeb to Asia, or north through Suez to Europe. With Hormuz already squeezed by the Iran war, Yanbu is the bypass. The Houthis are now standing on the bypass.

For seventy years, the U.S. Navy guaranteed that oil reached Tokyo, Seoul, Rotterdam and Marseille, and the countries receiving it never bore the real cost. They hosted some bases and paid some bills, but the warships, the risk and the casualties were American. Their most reliable contribution was the occasional lecture about American arrogance. That arrangement just paused. The people who need the oil are now the people who have to protect it.

And let’s remember whose war this is. Saudi Arabia went to war with the Houthis in 2015 and fought them for seven years, with American intelligence, fuel and munitions, without dislodging them. The fight began as Riyadh’s, and the oil at stake mostly flows to Asia and Europe. America helping is reasonable. America doing it alone is not.

Enter the Leader of the Free World

Canada’s navy has twelve Halifax-class frigates, commissioned in the 1990s. It has no destroyers; the last one was retired in 2017. The replacements are due in the 2030s. Its four submarines spend most of their lives in refit. When the U.S. assembled Operation Prosperity Guardian against the Houthis in 2023, Canada contributed a handful of staff officers. Not ships. Officers.

In fairness, Carney says Canada finally hit NATO’s 2 percent target this year, and he signed on to 5 percent by 2035. Good. The first of Canada’s new submarines could arrive by 2034. But the Houthis are working on a timetable of weeks, and 2035 is a long way from Mocha.

I propose the following campaign plan. Phase one: a strongly worded communiqué. Phase two: a coalition summit in Ottawa. Phase three: a second communiqué, more strongly worded. Phase four: the frigates arrive, sometime after the new destroyers enter service.

Four days after the Houthis seized Bab el-Mandeb, Europe offered Canada a brand-new status as the EU’s first “associate member,” complete with a standing ovation and kisses on both cheeks. Associate membership doesn’t exist in the EU treaties, but neither does a Canadian destroyer, so it’s a good match. Here’s a welcome gift idea: Europe and its new associate could keep open the sea lane that carries Europe’s oil. Everyone has something to say about the Americans. Here’s a chance to say it with ships.

And Carney has had plenty to say. In March 2025 he declared the United States “no longer a reliable partner” and said the old relationship of tight security and military cooperation was over. This spring he told Canadians that “hope isn’t a plan.” Last month, after trade talks collapsed, he said America’s signature is sometimes “written in pencil.” Fine. Take him at his word. If America is unreliable, this is the moment for the reliable partners to show up. Hope isn’t a plan in the Red Sea either.

The Coalition Assembles

Canada won’t be alone. France is ready: the Foreign Legion is the stuff of legend, and France already keeps its largest African garrison in Djibouti, a few hours’ sail from Perim Island. They could practically commute.

Canada’s new friend is nearby too. Having concluded that America is no longer a reliable partner, Ottawa cut a deal with a more dependable one: Beijing. Chinese electric cars in, Canadian canola back to China. Surely a reliable partner will help out. Conveniently, China’s only overseas military base is also in Djibouti, right at the mouth of the strait. The People’s Liberation Army Navy has been sitting there since 2017, watching the world’s shipping go by.

So picture it: a Canadian frigate, a Legion detachment and a Chinese naval base, all within sight of Bab el-Mandeb, united in defense of the rules-based order.

Even Greenland could send a kayak flotilla. Its people invented the kayak, which has a longer record of reliable service at sea than Denmark’s last frigate in the Red Sea.

To be fair, Europe can deploy in a hurry when it wants to. In January, when Trump threatened to take Greenland, Europe rushed troops there: France sent 15 soldiers, Germany a 13-person reconnaissance team, Sweden three officers, Norway two and Britain one. Europe will send soldiers to defend Greenland from America.

The Whole Bench

Look at who’s available. Britain, France and Italy have aircraft carriers. Japan, India, South Korea and China have serious navies and even more serious exposure, since it’s their oil. Turkey has NATO’s second-largest army and sells drones to half the world; a drone superpower should be able to handle a drone militia.

Saudi Arabia has hundreds of modern combat jets. The UAE has actually done this before: its forces and allies took Mocha from the Houthis in 2017 and built a base on Perim Island. The ground the Houthis just seized is ground the Emiratis once held. Egypt loses Suez Canal revenue every week the Red Sea stays dangerous; it has the most direct financial stake of anyone.

And don’t count on Israel to do the work either. Israel is one of the few countries that has actually bombed the Houthis, hitting Hodeidah port and Sanaa airport more than once. Europe’s thanks: condemnations, arms embargoes, and on September 8, as the Houthis closed in on the strait, a joint British, French and Canadian plan to sanction Israeli settlements. Mark Carney signed it. You can’t sanction the one country that shoots at the Houthis and then expect it to clear your shipping lane.

Combined, without the U.S. and Israel, they should be able to handle a ragtag army in Yemen.

The First Volunteer

As I was writing this, one stepped forward, and it wasn’t Canada. France announced it will send soldiers, radars and air defenses to protect Yanbu, hours after Houthi missiles were fired at it. Credit where due. But Macron was explicit: this is to protect a site, not to engage in the conflict. Britain has offered the Saudis aerial refueling. France will guard the pump. The strait stays in Houthi hands.

What to Watch

There’s a real case that Trump is wrong. Higher global oil prices reach American pumps six weeks before the midterms. Iran gains leverage by having proxies at both chokepoints. And I’m sure Canada’s reliable new partner, China, will come to the rescue.

But strip away the satire and this is Trump’s point, and it’s a fair one: help out. It’s your oil, your trade, your sea lane. Europe has time for standing ovations and new membership categories. Canada has time to lecture Washington and cut deals with Beijing. Nobody has time to send a warship to the one strait that feeds them. The only way the free world grows up is if its designated adult stops doing its homework for it.

I’m from Missouri, the Show-Me State. Pledges for 2035 don’t impress me. The countries that will actually spend the money are the ones on Russia’s doorstep: Poland, the Baltics, the Nordics, and Germany, which changed its constitution to do it. The rest signed up for 5 percent the way people sign up for a gym in January. Europe’s voters have already made their choice. Try telling a French retiree his pension is going to frigates, an Italian his August vacation is going to munitions, or a Briton that NHS money is going to fighter jets. Five percent on defense happens when pigs fly, and the pigs are not in the budget either. Show me the signed shipbuilding contracts, the new munitions lines, the frigates on station at Bab el-Mandeb. Europe and Canada have spent a year telling us America can’t be relied on. The Red Sea is where they get to show us what can.

Prime Minister, the Houthis are on line one.

Mark Carney stood in St. John’s on August 17 and called it the largest clean energy investment in North American history. Nearly $70 billion. Up to $10 billion in federal financing. Fourteen thousand megawatts of clean, renewable power — nearly tripling Churchill Falls’ current generating capacity. Enough to light, heat, and cool every home in Toronto, Montreal, and Vancouver combined. Twenty-three thousand jobs. Thirty-one billion dollars added to GDP through the early 2040s.

The numbers are real. The framing is not.

This is not Carney making Canada great. This is Canada, after fifty-seven years, finally getting out of its own way.

The deal nobody could close

The mechanics of Churchill Falls are almost too on-the-nose to be believable as a metaphor, except they’re real. In 1969, Newfoundland signed a 65-year contract selling power to Quebec at fixed rates. As electricity prices rose over the following decades, Hydro-Québec ended up buying that power for roughly $2 per megawatt-hour and reselling it into US and Quebec markets for tens of times more. Newfoundland kept almost none of the upside. It fought this in court, repeatedly, up to the Supreme Court of Canada. It lost.

Everyone involved has known for decades that developing Gull Island — the larger, more lucrative site downstream on the Churchill River — was the obvious next move. Newfoundland had a detailed framework to do it in 1998, under then-premier Brian Tobin. It fell apart amid Innu protests and unresolved Aboriginal-rights issues, but that wasn’t the whole story — costs were rising and the project was being substantially redesigned. By 2002, under Premier Roger Grimes, the two provinces reached another serious framework: Newfoundland would own Gull Island outright, sell the power to Quebec under a long-term contract, keep recall rights, and get a price with an escalator tied to electricity values. It was never finalized either.

So the raw material for this “historic” announcement — the geology, the engineering, the market for the power, even a workable commercial structure — has existed since before most of today’s Canadian premiers held office. What didn’t exist was an agreement between two provinces that had spent half a century treating each other as adversaries.

What actually changed the math

The breakthrough didn’t start in Ottawa. Back when the MOU was being negotiated, Quebec’s then-premier François Legault made the underlying math plain: the 1969 contract runs out in 2041, and 2041 is closer than it looks if you’re planning generation capacity. Losing access to Churchill Falls power would leave Hydro-Québec facing a real shortfall down the line, at a moment when the province’s cheap hydro rates are part of its economic identity. Talks between Quebec and Newfoundland restarted years earlier — a tentative memorandum of understanding was already in place by December 2024, negotiated under Legault and then-NL-premier Andrew Furey, before Carney was even prime minister.

The 2026 deal isn’t just that 2024 MOU with federal money bolted on, though — it’s a materially better renegotiation of it. Newfoundland’s own government puts its estimated benefit at $49 billion under the new arrangement, up from $36 billion under the 2024 MOU, with more retained power and better export access. Some of that improvement came from Ottawa: Newfoundland credits the federal contribution — wind equity participation, transmission support, the Gull Island loan guarantee — with adding roughly $3.5 billion in present value to the deal. So the fair account is this: Quebec and Newfoundland reached the fundamental rapprochement in 2024, on their own, before Carney was in office. Ottawa then helped turn that rapprochement into a bigger, better-financed package.

Carney’s government put real money on top of an already-negotiated deal: the $10 billion in federal financing and loan guarantees, clean-economy tax credits, and an expedited review through the Major Projects Office. That’s not nothing. It’s also not the thing that broke a fifty-year impasse. The thing that broke the impasse was Quebec doing its own actuarial math and Newfoundland finally having enough leverage to get a deal it could live with.

This isn’t a Churchill Falls problem. It’s a Canada problem.

The usual story Canadians tell themselves about their economic problems runs through Washington — tariffs, trade threats, a mercurial neighbor to the south. That story isn’t fictional. It’s also not the whole story, and arguably not the main one. Interprovincial trade barriers, jurisdictional standoffs, resource nationalism between provinces that are supposed to be on the same team — these have been quietly compounding for decades while getting a fraction of the political attention tariffs do. Churchill Falls took fifty-seven years to unstick not because Washington stood in the way, but because Quebec City and St. John’s did.

To be fair

A few things this piece is not claiming. Carney’s federal financing was real, not cosmetic — $10 billion in guarantees and tax credits is a meaningful de-risking of a $70 billion buildout, and it’s fair to ask whether Hydro-Québec and NL Hydro could have financed this entirely on their own credit without it. Some of the friction in this story wasn’t dysfunction — the unresolved Aboriginal-rights issues that helped sink the 1998 deal were a legitimate consent process working as intended, not an obstacle to be waved away. It’s less clear the process has actually improved this time: Ottawa and St. John’s describe the Innu Nation as a partner in the new deal, but Innu Nation Grand Chief Jodie Ashini has said the new framework cuts the royalties and benefits Labrador Innu were due under the 2024 MOU and that her nation cannot support it as written — a live consent fight, not a settled improvement. And it’s possible, even likely, that Carney’s timing here does real work for him in ongoing trade negotiations with the US, whatever its causal role in the deal itself. One more thing worth saying plainly: the August agreement is signed — the Definitive Cooperation and Implementation Agreement is a real, executed document — but the deal isn’t finished. What’s still outstanding is a set of long-form definitive agreements meant to give the framework full legal effect. The parties hope to have those done by the end of the year; the framework itself holds until March 2027 if it takes longer than that. Quebec holds a provincial election on October 5, less than two weeks after this piece went up, and the Parti Québécois — which has questioned whether an outgoing government had the standing to lock Quebec into a roughly 50-year deal days before an election, though its leader has said he wouldn’t tear it up for its own sake — has led most recent public polls. Quebec’s own premier, Christine Fréchette, has said plainly that a change of government could put the whole thing at risk. The “getting out of its own way” story could still stall on its way to the finish line.

None of that changes the headline. The obstacle standing between Canada and this outcome for half a century wasn’t a foreign government. It was two Canadian provinces that couldn’t agree on terms, with a federal government mostly on the sidelines.

Coming next

Churchill Falls isn’t the only place this pattern shows up. Canada’s dairy quota system runs on the same logic on a smaller scale — a domestic cartel that predates any American complaint, protecting incumbents from competition, foreign or Canadian, and then blaming Washington when the arrangement gets challenged. Later in this series: the internal trade barriers between provinces — which the IMF estimates run to the equivalent of roughly a 9 percent tariff nationally, more than the current US tariff dispute by most estimates — the pipeline fights that have killed projects with willing buyers waiting at the other end, the equalization arguments that turn provinces against each other every time the topic of who’s paying for what comes up, and a look at just how much this pattern has cost Canada in the growth it never had.

Canada’s enemy, on file after file, keeps turning out to be Canada.

None of that happened. Obviously.

What was actually signed

On September 22, the United States, Denmark and Greenland signed an agreement amending the 1951 Greenland Defense Agreement. The text is public, twelve articles, and worth reading, because it is not what either side’s press releases say it is.

The US gets the right to expand Pituffik Space Base, new defense areas at Narsarsuaq and Mestersvig, and a fast-track process for proposing more. US military aircraft can fly over and land anywhere in Greenland. US vessels get undersea access to Greenland’s territorial waters. No non-NATO state may put a base or a persistent military presence on the island. Investors from countries outside NATO, its partner nations and the EU are barred from control of critical infrastructure and resource extraction. The treaty doesn’t name China. It doesn’t have to.

Then there’s Article XI: the agreement has no end date. If Greenland ever votes for independence, Denmark and Greenland are obligated to ensure the new state stays in NATO and inherits every one of these obligations.

It isn’t annexation, and it isn’t “permanent control over security and all other needs on that territory,” as the President put it at the UN. Sovereignty is explicitly reaffirmed. But it is the first real change to America’s position in Greenland in decades, and it locks that position in beyond any future independence vote.

What they actually said

In January 2025, days into Trump’s second term, Hakeem Jeffries said “the problem is not Greenland.” A year later, when the pressure campaign was at its height, Chuck Schumer called it a “quixotic quest to takeover Greenland.”

The more revealing moment came in April 2025, when Jeffries led a congressional delegation to Copenhagen. His argument was that none of this was necessary: the 1951 agreement “is still valid, it’s alive,” and, he said, Danish leaders had made clear it was available. By his own account, the delegation did not discuss increasing the US military presence in Greenland with Danish officials.

Read that again. The Democratic position was that the tool was sitting right there, and they never picked it up.

The day before the signing, Foreign Policy called it a “pyrrhic victory” and argued the US could have expanded its footprint without any new deal.

Meanwhile, the ice melted

While Washington treated Greenland as a real-estate joke, the Arctic opened up.

Russia reactivated its Cold War-era Arctic bases and moved in advanced air and missile defense systems. It runs more than 40 icebreakers, eight of them nuclear-powered. The United States has three, and one is 50 years old.

China declared itself a “near-Arctic state” in 2018, which is a bit like Nebraska declaring itself a near-Pacific state. It has since built its own icebreakers and sent research submersibles under the Arctic ice. Chinese container voyages on Russia’s Northern Sea Route went from 7 in 2023 to 14 in 2025, and this year a Chinese carrier is running a scheduled, roughly weekly Arctic service to Europe: about 20 days instead of 40 via Suez.

And they’re doing it together. In 2024, Russian and Chinese bombers flew a joint patrol near Alaska, and their coast guards ran their first joint Arctic patrol. Both happened on Biden’s watch.

To be fair, analysts point out that neither country has a real foothold in Greenland itself. That’s the point. Articles IX and X of the new agreement make sure they never get one. It’s cheaper to lock the door before someone moves in than after.

The obvious question

If the 1951 agreement was always available, why did nobody use it?

Obama had eight years. His 2013 Arctic strategy listed national security as a priority. His administration finished a radar upgrade at Thule that George W. Bush had started, and added no new bases in Greenland. When a Chinese state construction firm bid to build Greenland’s airports, it was the Trump Pentagon in 2018 that pressed Denmark to finance them instead.

Biden had four years. His 2022 Arctic strategy named Russia and China as competitors and called for stronger Arctic defense. He kept the Nuuk consulate that Trump reopened in 2020, and renamed Thule to Pituffik. That was about it: no new bases and no investment screen. Both administrations saw the problem on paper; neither put money where the paper was. The US left its air base at Narsarsuaq in 1958; it became a small civilian airport, and neither administration asked to go back.

“We could have done it anytime” is not a defense. It’s an admission. It’s the kid telling his parents they didn’t need to bug him about his homework — right after they spent a week bugging him, and the homework finally got done. A treaty sitting in a drawer since 1951 is not the same as a solved problem. Bases get built or they don’t.

Could quiet diplomacy have gotten this deal? Maybe. Greenland did propose its own investment-screening law — in late 2025, ten months into the pressure campaign — and then withdrew it this spring. But nobody tried the quiet way. And the quiet way never answered the biggest question of all: what happens to American bases if Greenland votes for independence? The 1951 agreement was signed with Denmark. Nothing in writing said an independent Greenland would be bound by it. Now Article XI does. For twenty years the quiet approach produced no new bases, no investment screen and no successor-state clause. The loud approach got all three into a signed treaty in twenty months. Critics who say pressure wasn’t necessary are welcome to point to the quiet version of this agreement. There isn’t one.

The honest caveats

The Greenland agreement is not in force yet. Article XII says it takes effect only after both the Danish and Greenlandic parliaments complete their procedures. The investment screen depends on a Greenlandic screening law that doesn’t exist yet; the last attempt was withdrawn this spring. And there is no money anywhere in the text: no appropriation, no timeline, no troop numbers. And Greenlanders themselves aren’t cheering. Thousands marched in Nuuk in January, and Greenland’s parliament still has to approve this.

So here is the falsifiable test. If Narsarsuaq and Mestersvig appear in a defense budget, with construction contracts going to Greenlandic firms as Article IV calls for, this was a real strategic gain. If they don’t, it was a signing ceremony.

Either way, one side spent twenty years saying the door was open, and the other side walked through it.

But of course, Trump didn’t really do anything here. The Democrats had already solved Greenland, just like they solved Iran with the Obama deal. Ditto for the border.

Don’t wait for the congratulations.

Watch any interview on cable news and you’ll see the same ritual. Every guest genuflects before the host. “Thanks for having me.” “Great to be with you.” “Always a pleasure.”

Even the biggest names do it. Senators, cabinet officials, authors at the top of the bestseller list. People who could fill any studio in the country still open with the thank-you. So do the host’s favorites, the regulars who are on every week and have nothing left to prove. They thank the host the loudest.

It doesn’t stop at hello. Once the segment starts, the guest agrees. “That’s exactly right.” “Great point.” The host offers a take, the guest hands it back in slightly different words, and both call it analysis.

Some of it is manners. You see it on the network news and on some guy’s podcast out of his spare bedroom. But the incentive is the same everywhere: the host decides whether there’s a next time. The bigger the audience, the more the next time is worth, and the heavier the deference gets.

And it isn’t one host’s door. Bookers talk. A guest who argues with a host or shows one up on air gets a reputation as difficult, and the reputation travels. Embarrass one host and you’ve made yourself hard to book on everyone else’s show too.

Now imagine the White House required the same of reporters. No hard pass unless every question opens with “Mr. President, we love you.” The press would go ballistic. There would be lawsuits by lunch.

The networks already run that system. They just call it booking.

The Seat Is the Paycheck

Start with what the news is: a business. It sells ads, subscriptions and cable fees, and it books whatever keeps the audience watching. “Journalism” is how the business describes itself. It’s a claim, not a credential, and the way to test it is to watch what the business does. Even the Times now grows on Wordle, recipes, sports and product reviews, which is why I asked whether the New York Times is still a newspaper.

News divisions don’t pay ordinary guests. The airtime is the payment. A guest gets exposure for a book, a campaign, a consulting practice, a career. A booker controls who gets it.

I argued in Freedom of the Press Is Not a Reserved Seat that the White House owes no reporter a seat. The same holds in the other direction. A network owes no guest a booking. Whether the White House can pull hard passes, the courts are still sorting out.

The question is whether the press lives by the rule it demands of everyone else.

Who Gets Invited

It doesn’t. Researchers have checked. A 2022 study in the Proceedings of the National Academy of Sciences tracked the political leanings of the people cable news put on screen from 2010 to 2021. After 2016, Fox’s lineup ran steadily more conservative, and CNN’s and MSNBC’s steadily more liberal. Each network invites its own side, and its own side thanks it for coming.

Who Gets Thrown Out

In March 2024, NBC News hired Ronna McDaniel, the former Republican National Committee chair, as a paid contributor. Within days, the network’s own anchors took to the air to denounce the hire. Chuck Todd, Rachel Maddow, Nicolle Wallace, Joe Scarborough. Four days later she was gone.

The head of NBC’s news division explained why in a memo to staff: “No organization, particularly a newsroom, can succeed unless it is cohesive and aligned.”

Hold that next to the White House. When the President banned CNN, MS NOW and Politico on September 18, the three outlets sued three days later. “This ban could not be a more direct assault on the First Amendment,” the lawsuit says. The White House answered with a press release titled “White House access is a privilege—not a right.”

Nobody is suppressing the press, as I pointed out in Freedom of the Press Is Not a Reserved Seat. They can still write whatever they want.

NBC runs on the same principle. A contributor’s chair is a privilege, not a right, and McDaniel lost hers in four days. NBC’s defenders say her conduct disqualified her. The White House says the same about its bans: the outlets, Trump wrote, “constantly write or report fiction and lies.” Either conduct can cost you a seat or it can’t. The press claims the power for itself and denies it to everyone else.

The Shield

There’s one more asymmetry. The press answers to almost no public figure for what it says. Since New York Times v. Sullivan in 1964 and the cases that followed it, a public figure who sues for defamation has to prove the outlet knew the story was false or recklessly didn’t care. Honest mistakes, even careless ones, aren’t enough for a public figure to win. The occasional nine-figure settlement makes news because it’s rare.

So the press holds the gate, sets the terms of entry, collects the thank-you, and is nearly immune to losing a lawsuit to the people it covers most. That’s a lot of power for an institution whose whole brand is holding power accountable.

Honest Caveats

Not every host books only friends, and not every guest is a courtier. The hard-news hours are far less skewed than primetime. Audiences reward agreement, so part of this is the networks giving viewers what they pay for. And the thank-you, taken alone, is often just courtesy. The argument isn’t that manners are sinister. It’s that the press already practices, quietly, the access politics it denounces loudly when someone else does it.

If I Were King

The briefing room works like a cartel. The big outlets hold the front rows, and for decades the incumbents’ own association decided who sat where and who rotated into the pool. When one member gets squeezed, the rest close ranks. After the CNN ban, ABC, CBS, NBC and Fox News all refused to replace CNN’s pool crew. Fox included.

The pool is supposed to be the public's eyes on the President. The networks shut it off to protect CNN's seat. The public loses the pictures. The club keeps its rules.

If I were king, I’d take away every permanent network seat in the briefing room and put them on rotation among the thousands of journalists and newspapers in this country. The small-town daily, the trade reporter, the freelancer with a newsletter. Everybody gets a turn. Nobody owns a chair.

The networks would hate it. Which tells you how much they believe in open access when the access is theirs.

Do Unto Others

Swap the chairs. Give Trump his own show and book the White House press corps as his guests. To get invited back, they’d have to thank him for having them, nod along, and tell him what a great point he just made. That’s the deal every cable guest takes tonight.

It came up again yesterday. The President announced on Truth Social that he was banning CNN, MS NOW and Politico from the White House, effective immediately, for reporting “FICTION and LIES.” This morning their reporters’ hard passes didn’t work at the gate. “Other Fake News Media Outlets to follow,” he wrote.

All three will publish whatever they want tomorrow. Nobody has stopped them from printing a word. What they lost is a badge.

It comes up every week in smaller ways. A reporter asks the President a question that is half accusation and half performance. The President snaps back, calls the question stupid, and moves on to someone else. By evening the headline is that the press is under assault.

Nobody was stopped from printing anything. The reporter went back and wrote what he wanted. The outlet ran it. The only thing that happened is that someone with a microphone got talked back to.

This piece isn’t claiming reporters should be punished for what they write, or that the government should decide what counts as fair coverage. It’s claiming something narrower: nobody owes a rude or biased reporter an interview, a seat, or a ride.

What the First Amendment Actually Says

“Congress shall make no law … abridging the freedom of speech, or of the press.” That’s the whole thing. It restricts what the government can do to the press. It doesn’t give the press a claim on anyone.

The Supreme Court has said so directly. In Pell v. Procunier (1974) and Houchins v. KQED (1978), it held that the press has no constitutional right of access to information beyond what the general public has. Freedom of the press is freedom to publish. It isn’t freedom to be invited.

There’s no clause requiring an interview. There’s no clause requiring a seat on Air Force One. There’s no clause requiring anyone to stand still and absorb a loaded question.

Access Was Always Rationed

Every argument about “access” skips over one plain fact: almost no reporter has it.

There are tens of thousands of working journalists in the United States. A small fraction hold a White House hard pass. Getting one requires congressional press gallery credentials and a Secret Service background check. The briefing room has 49 assigned seats. The press pool traveling with the President is a handful of people covering for everyone else.

For decades, the White House Correspondents’ Association decided who sat where and who rotated into the pool. The WHCA is a private association of the reporters already in the room. The incumbents decided who joined the incumbents.

In February 2025 the White House took over pool selection, and a month later moved to take over the seating chart. The reaction treated this as an attack on press freedom. It wasn’t. Control of a scarce privilege moved from one gatekeeper to another. Neither arrangement was ever a constitutional right.

The rationing isn’t new or partisan either. In 2023 the Biden White House tightened hard-pass eligibility, and 442 passes lapsed. It drew nothing like the outrage of 2025. In 2009 the Obama White House tried to cut Fox News out of an interview with the Treasury’s pay czar. The other networks refused to go along, and the White House backed down.

Where the Courts Draw the Line

The Constitution doesn’t mention press passes, but the courts have built some doctrine around them. Sherrill v. Knight (D.C. Cir., 1977) held that once the White House opens press facilities to journalists generally, it can’t deny passes arbitrarily. It needs published standards and some due process.

That’s how CNN’s Jim Acosta got his pass back in 2018, and Playboy’s Brian Karem in 2020. In March a federal judge ordered the Pentagon to restore credentials to seven New York Times reporters, finding its policy was built to exclude disfavored journalists. This week’s ban is a hard-pass ban.

The Associated Press was shut out of the Oval Office and Air Force One for refusing to write “Gulf of America.” It sued and won in April 2025. In June the appeals court put most of that win on hold. Two of the three judges said the Oval Office and the President’s plane are his working spaces, not open press venues, and he can choose who comes in. The full appeal was argued last November. No ruling yet.

So far the courts agree with the common-sense answer. The President’s office and the President’s plane are his to fill. The pool is a custom, not a right.

If the case reaches the Supreme Court and the Justices side with the White House, the headline is already written: the conservative Court strikes again. The preview has already run. When the D.C. Circuit stayed the AP’s win, Courthouse News led with “Trump-appointed judges restore White House ban on AP.” The district judge who had ruled for the AP, Trevor McFadden, is a Trump appointee too. I’ve covered this reflex before in Is the Supreme Court Really Partisan and Unethical?

There’s no statute behind any of this. Congress has never passed a law on White House press passes. The rules are custom, White House policy, and a 1977 appeals court ruling the Supreme Court has never reviewed. If the Court someday gives the President more room, that answer could last. Three of the Justices are in their 50s or early 60s and could sit for another twenty or thirty years. Two more, Thomas at 78 and Alito at 76, may retire while Trump is still in office, and their replacements would likely be younger still.

A word of caution for CNN, MS NOW and Politico: be careful what you litigate. Push this case all the way up and lose, and a narrow precedent becomes the law of the land for a generation. Don't push your luck.

Why the Questions Are So Bad

Access is scarce and it’s granted. That makes behavior matter. The question is how the people holding the seats have chosen to behave.

Start with who holds them. The 2022 American Journalist survey from Syracuse and Indiana found that 3.4% of American journalists identify as Republicans. Gallup’s 2025 reading put Americans’ trust in the mass media at 28%, the lowest it has ever recorded.

Then look at the business model. News used to sell soap. Now it mostly sells subscriptions and cable fees, which is worse. When the audience is the customer, the product is whatever the audience wants to hear. Clicks still matter too. They're how an outlet proves people are reading. An outlet whose audience leans one way has no commercial reason to cross it. Its reporters have every reason to ask the question that will clip well that evening.

The result is a question built for the clip, not the answer. It’s framed so any response confirms the premise.

Framing doesn’t require lying. Selective quotation, missing context, a loaded headline over a technically accurate story, and a decision not to cover something at all can each leave the reader believing something false without a single false sentence. The effect is the same as lying. The deniability is better.

Anonymous sources are the wildcard to report anything without penalty. If it turns out to be wrong, the source was wrong. The source can’t be checked, can’t be questioned, and never has to answer for it.

I’ve cataloged these techniques, including the loaded question at the press briefing, in The Modern Journalist’s Field Manual. I’ve also built an app, Bias Unlocker, that checks news reporting for them.

Pushback Is Not Suppression

When a President calls a question stupid, refuses an interview, or chooses who rides on his plane, he’s exercising the same freedom the reporter is. That’s speech. It isn’t censorship.

Declining to reward what he considers bad behavior with a scarce seat isn’t censorship.

The Entitlement

In Is Democracy Really Under Siege, and If So, From Whom? I asked whether the current alarm about losing our democracy is really just the side that’s used to having carte blanche discovering it doesn’t have it anymore.

The legacy press has the same reflex. The same outlets held the front row, the pool rotation, and the gatekeeping for decades. Held long enough, a privilege starts to feel like property. Property doesn’t get revoked for bad behavior, so the behavior stopped mattering. The loaded question costs nothing. The distorted story costs nothing. The seat is still there the next morning.

In almost any other line of work, abusing a scarce privilege costs you the privilege.

Crying Wolf

There’s a cost to the press here too, and it’s self-inflicted. When every insult is an “assault on the press,” the phrase stops meaning anything. If a real attack ever comes, a genuine attempt to use state power to shut someone up, the alarm will already have been rung a hundred times over a rude answer at a press gaggle.

Honest Caveats

Not every tough question is a rude one, and plenty of reporters in that room do the job straight. Presidents of both parties have confused hard questions with hostile ones. And the White House picking the pool puts the gatekeeping in the hands of the person being covered, which is a real tradeoff, not a free lunch. The argument here isn’t that the new gatekeeper is better. It’s that there was always a gatekeeper, and the seat was never a right.

Freedom to Publish

Freedom of the press means the government can’t stop you from printing what you want. It has never meant anyone has to listen, answer, or give you a ride.

The press is free to say whatever it wants, however it wants. Everyone else, including the person it’s shouting questions at, is free to stop listening.

Witkoff and Kushner hadn’t set foot in Kyiv the entire war, until this trip. Days after Putin told a Vladivostok forum there was “a chance” of a deal, they went — Moscow on September 5, Kyiv on September 6. Coverage called it a revival of stalled talks. Nobody asked why now.

The Price Problem

Here’s why now. Oil had dropped nearly 20% in May on hopes a U.S.-Iran ceasefire MOU would hold. It didn't — renewed strikes in July sent prices back up, and they've kept climbing since. Brent was already near $94 by September 1, past $100 before the envoys even left for Moscow. The midterm framing predates the trip too: the White House convened refiners on August 31, explicitly to find ways to cut prices 'ahead of November's congressional midterm elections,' per Reuters. On September 9, WTI closed at $96.30, Brent over $101, regular gas at a national $4.28 and diesel at a record $5.85. The administration ordered a 172-million-barrel emergency release from the Strategic Petroleum Reserve in March, after the Strait of Hormuz closure, trying to hold prices down. Trump told reporters that day gas won’t fall below $2 until “right after” the midterms. White House spokeswoman Taylor Rogers, from the RNC convention floor, promised prices would fall “back to pre-conflict levels.”

The Ukraine Side

Ukraine’s refinery campaign is working — against Russia. Eighteen refineries hit in July, twenty-one in August, Russian crude processing down to a 2002 low. That’s also, by the administration’s own account, pushing prices up at home. Separately, Vance called Zelensky on July 31 about a different target: a drone strike on the Caspian Pipeline Consortium terminal near Novorossiysk, which moves Kazakh crude — not Russian refining capacity — through Russian territory, and in which Chevron and ExxonMobil hold stakes. He complained it was “destabilizing oil markets.” Bessent said on air, September 2, that the world is experiencing an “energy shock” from the Iran war and from Ukraine, which “decided that they want to blow up Russian energy assets and refined properties.”

The Tell

So: the administration blames a tactic that’s hurting Russia for a price problem hurting Republicans in November. And its peace envoys land in Kyiv for the first time in the war the same week gas prices become the GOP’s own midterm message.

Security Guarantees, Already Proven Worthless

Ukraine gave up its inherited nuclear arsenal in 1994 for written security assurances from the U.S., U.K., and Russia — the Budapest Memorandum. Russia violated them twice; Washington and London did nothing binding either time.

The current guarantee isn’t holding up better. Europe pays for Ukraine’s Patriots under PURL, but Zelensky said in April the U.S. was diverting that stock to Iran instead — “we are not the priority for today.”

Weeks before this round of talks, Trump also reversed the one guarantee he’d made in writing — a Patriot production license, offered at the NATO summit, withdrawn at Camp David. It’s the same license Kyiv was reportedly counting on when it agreed to Vance’s request and stood down on the CPC terminal. The U.S. got the concession; Ukraine didn’t get the license. That’s not an unfinished draft. It’s a pattern.

Washington isn’t funding any of this, either. No new U.S. money for Ukraine since January 2025 — intelligence sharing continues, but Europe is covering the weapons, and what’s still arriving is leftover Biden-era contracts working through the pipeline. Ending the war is a free win for an administration that isn’t covering any of its cost.

Nobody’s even decided who delivers the next security guarantee or what it would look like. The current proposal has Europe deploying troops and the U.S. “backstopping” with intelligence and logistics — not a treaty, not Senate-ratified, the same legal weight as 1994. And it only activates after a ceasefire, which hands Russia leverage to shape or block it before it exists.

Ukraine needs a real guarantee to sign anything. Russia doesn’t want Ukraine to have one — that’s the whole point of insisting on no NATO membership and no binding Western commitment. Nobody has offered anything concrete yet.

Has Anything Changed?

Putin’s territorial demands haven’t moved. His adviser, after the Moscow meeting, said Russia still expects “all the root causes” addressed.

The Sequence

This administration let Ukraine diplomacy sit frozen for six months. It picked it back up the same week gas prices became its own party’s midterm liability. That’s the sequence. Washington’s coverage hasn’t caught up to it.

The Mechanics Are Already Settled. This Is About the Politics.

I already walked through the institutional mechanics in “Does Winning Congress Give Democrats an Iran-War Off Switch?” — the funding lever doesn’t work because the Pentagon can cover most of this war out of its existing budget, a shutdown fight favors a president who’s comfortable holding one indefinitely, and impeachment dies in the Senate before it gets near removal. None of that has changed. But set the mechanics aside for a moment and ask a different question: even if Democrats somehow had the votes to force an end to the war, would they actually use them?

The answer is no, and the reason has nothing to do with procedure. It’s about what happens to a party that actually ends a war it campaigned against.

That’s also the more coherent way to read Iran’s own strategy. Reporting throughout this year has described Iranian officials, including an IRGC adviser, framing their bet as outlasting Trump rather than counting on Congress — waiting until he’s out of office in January 2029, not until a midterm reshuffles committee chairs. The theory floating around Washington that a Democratic Congress represents near-term relief is a mismatch with what Tehran itself appears to be counting on.

But as we’ll see in this piece, November 2026 or January 2029 makes little difference — neither will likely bring the exit Tehran thinks it will, because some of what the war has already done by either date won’t be something any president, of either party, can walk back.

Obama Talked About Ending Wars. He Didn’t.

Obama ran in 2008 on ending the wars he inherited. In office, the record looks different. Iraq withdrawal happened on the timeline Bush had already negotiated in the 2008 status-of-forces agreement, not an accelerated one. Afghanistan went the other way entirely — Obama surged troop levels from about 30,000 to a peak of roughly 100,000 by 2010–2011, more than tripling the force he inherited, and by the time he left office in January 2017 there were still roughly 8,400 American troops there. Eight years of a president who ran against the war, and the war he actually escalated outlasted his presidency.

That’s not a gotcha about Obama personally. It’s a data point about what happens when the party that campaigns against a war has to actually own the consequences of leaving. The incentives point toward managing the war quietly rather than ending it loudly.

Biden Actually Did It. It Was a Disaster. The Kabul Paradox

Biden did end the war abruptly — and the polling shows exactly what it cost him.

I laid out the polling on this in “What the Iran War Polls Really Say — and What They Don’t”: in September 2021, 77% of Americans supported withdrawing from Afghanistan, one of the most lopsided consensus numbers you’ll find on any policy question. In that same poll, 60% disapproved of how Biden actually handled it, and his approval had dropped six points in two months and never recovered — he left office three years later at 40%, having spent his term averaging the second-lowest approval rating of any post-WWII president. Wanting a war over and accepting the price of ending it turned out to be two entirely different votes, measured in the same survey, at the same time. Call it the Kabul Paradox: the public will back ending a war right up until someone shows them what ending it actually requires. Kabul gave every Democratic strategist a vivid demonstration of that paradox — that ending a war on camera can be more dangerous than continuing one quietly.

The Public Wants It Over — But Not Irrespective of What the Ending Entails

The same paradox doesn’t just catch presidents after the fact. It’s visible in the Iran polling right now, before there’s even an ending to grade. Economist/YouGov found 68% of Americans want a deal to end the war as fast as possible — until they were shown a specific deal that left Iran’s enriched uranium intact. Support collapsed to 34%.

That’s the political exposure Democrats would inherit if they actually forced the war’s end.

The Blockade Is a Much Cheaper War to Own

Compare that exposure to what the blockade currently costs, politically. Higher gas prices. Some inflation pressure. Media narratives about American ships and Iranian tankers that most voters don’t follow closely enough to hold anyone accountable for day to day.

That’s most likely a manageable, diffuse cost. As I wrote in “Is Trump Cornered, or Is He Managing a War He’s Already Winning?”, it’s a war being managed for cost on both sides.

This is why the War Powers votes keep happening and keep failing: a successful vote means owning the aftermath, a failed vote lets everyone perform opposition risk-free — the same dynamic “Does Winning Congress Give Democrats an Iran-War Off Switch?” already showed has no real exit ramp attached to it.

Promising something, like the end of the war, that cannot likely be delivered carries a cost too.

What a Midterm Win Actually Changes

None of this means a Democratic majority buys nothing. It buys oversight, hearings, and two more years of “no” votes that go into the 2028 record. What it doesn’t buy is an actual end to the war, because ending it is the one move with a real downside precedent — Biden’s — and continuing it is the move with no downside precedent at all, just an invoice at the gas pump that voters have already been paying for months.

If there’s one thing a Democratic Congress could plausibly claim credit for foreclosing, it’s a full-scale ground invasion of Iran. But that’s not a new constraint a midterm win would create — it’s already the reality under the current Republican House and Senate. Nobody in this Congress, in either party, is voting to authorize an occupation of Iran. A Democratic majority wouldn’t be closing off an option that was open.

Iran’s calculation that a change in Congress will change American policy assumes Democrats want the thing they say they want. They don’t: Democrats want credit for opposing the war, not responsibility for ending it. Those are different products, and only one of them is actually on offer.

And even if that weren’t true, the three structural clocks I wrote about back in April in “The Three Clocks Running Against Iran” don’t stop for an election. Congress can vote on the war. It can’t reverse what’s already happening to Iran’s oil fields, its export revenue, or the routes the rest of the world has built around the Strait.

The Media Is Watching the Wrong Clock

News coverage keeps framing Iranian strategy around the midterms, as if a shift in Congress is what Tehran is actually waiting on. Nobody covering this war has a source inside Tehran confirming that — the midterms just make a better daily story, with a countdown and a weekly scoreboard update, exactly the mechanism I described in “Is Trump Cornered, or Is He Managing a War He’s Already Winning?”, where the horse-race frame wins column inches by being compelling, not accurate. A story that only resolves in 2029 generates no clicks in the meantime, which is reason enough to be suspicious of the framing independent of whether Iran actually holds it.

We don’t actually know which clock Tehran is watching. A Democratic Congress raising the war’s political cost could still be something Tehran hopes pressures Trump into a faster deal, and there’s no way to verify that from outside. It wouldn’t be the first time: I wrote in “The Hall of Mirrors: What the Coverage Cannot Stop” about how Iran spent the run-up to this war reading a press narrative of American exhaustion that didn’t match reality, and paid for that miscalculation. If Tehran is doing the same thing again with the midterms, that’s a documented pattern, not a new risk.

Either way, it misses the bigger point: it doesn’t matter whether Iran is playing a two-month game or a three-year one, because neither timeline undoes what the three clocks already do. By 2029 the economic bleed is further along, more of Iran’s aging fields have crossed into permanent reservoir damage, and Saudi Arabia, the UAE, and Iraq will be running that much more pipeline capacity around the Strait instead of through it. Some of what the blockade wins doesn’t require the blockade to still be in place to keep winning it — once the world has built its way around the Strait.

Where This Could Break

The blockade’s low cost depends on staying diffuse. A single successful strike, a naval miscalculation, or an oil spike past $120 a barrel could turn a background grumble into a front-page crisis overnight. History cuts both ways here, too: after the 1983 Beirut barracks bombing killed 241 American troops — an attack linked to Hezbollah with Iranian backing, though how certain the intelligence was on direct Iranian sponsorship remains disputed — Reagan pulled the Marines out of Lebanon rather than escalate against Tehran. A mass-casualty event doesn’t reliably produce more resolve; a president facing that choice today could go either way.

Even the Last Withdrawal Took Twelve Years

Winding down Afghanistan wasn’t a single president’s decision. Obama promised to end it, then more than tripled troop levels to a peak of about 100,000, and only drew down to roughly 8,400 by the time he left office — well short of an actual exit. Trump negotiated the withdrawal deal with the Taliban in 2020, and Biden was the one who finally executed it in 2021. From the presidency claiming to end the war to the last plane out of Kabul, the whole arc spanned three administrations and twelve years. If ending America’s longest war took that long once a president had already promised to leave, there’s no reason Iran policy resolves cleanly just because a different name is on the door in January 2029. Outlasting Trump assumes the exit happens on schedule the moment he’s gone. The Afghanistan precedent says the opposite.

The stakes are also larger here than they were in Afghanistan. Leaving the Taliban in control of Kabul was a strategic embarrassment; leaving Iran’s regime in place to sponsor terrorism across the region, destabilize its neighbors, and potentially field a nuclear weapon is a different order of consequence.

It’s not clear that waiting until 2029 changes much.

The film’s throughline is Moore trying to get GM chairman Roger Smith to walk through Flint and see what the company’s decisions did to it. Moore shows up at GM headquarters, at shareholder meetings, at a country club, at a company Christmas party, camera rolling, trying to corner Smith into an interview. Smith never grants one. Security and PR staff intercept Moore repeatedly. The film cuts these attempts against footage of the town’s decline: a sheriff’s deputy repossessing furniture, an auto show sponsored by GM in the same period Flint was hemorrhaging jobs.

Roger & Me became a box-office hit for a documentary, launched Moore’s career, and set the template for the confrontational, camera-forward, no-cooperation-required style he’d use again in Bowling for Columbine, Sicko, and Fahrenheit 9/11. It’s taught in film and journalism courses as a landmark of the genre.

Nothing in the film alleges that Roger Smith broke the law. GM’s plant closures and its move overseas were legal decisions made in response to a wave of global competition and labor costs reshaping manufacturing everywhere, not just at GM. Smith’s legal obligation ran to GM’s shareholders, not to Flint — corporate law gives management wide discretion to run a company in its own interest, with no duty to keep an uneconomic plant open for a hometown’s sake. The case against Smith was that he made a decision he was legally free to make, without the human sympathy Moore thought the moment demanded, and that he wouldn’t sit for an interview about it. That’s a moral complaint. It was never a legal one.

Roger & Me wasn’t an outlier. It’s one entry in a much longer tradition of journalism that operates without its subject’s consent and rarely draws fire from the left when the subject is a cause the left already opposed. 60 Minutes built a fifty-year run on hidden cameras and ambush interviews aimed at corporations, scam artists, and government agencies — nobody proposed shielding those targets by statute either. The method in all these cases is the same: show up where you’re not wanted, publish without permission, let the subject explain itself after the fact if it wants to. Nobody organized a bill.

Vilified First, Vindicated Later

In December 2025, independent journalist and YouTuber Nick Shirley posted a video alleging widespread fraud at Somali-run daycare centers in Minnesota — operators, he claimed, collecting taxpayer-funded child care payments for children who weren't being served. The video went viral within days, amplified by Vice President JD Vance and others, and triggered an immediate federal response.

Before any of Shirley’s Minnesota claims were adjudicated, the press response was immediate and dismissive. A University of Minnesota media law professor told NPR that influencers like Shirley “prioritize fearmongering over fact-checking.” PBS labeled him a “right-wing influencer” pushing fraud claims “without proof.” The New York Times downplayed the story as a single viral video. Some media-law commentators went further, framing the reporting itself as pushing an “anti-immigrant” and “anti-Muslim” narrative. Notice the label doing the work before the facts get a hearing: Moore has been called a documentary filmmaker for thirty-seven years, and an Oscar winner for the last twenty-three. Nobody called Roger & Me an influencer’s video. Shirley, doing the same thing outside the same traditional outlets, gets “influencer” — a word that pre-loads the dismissal before a single claim is checked.

At least one of the centers Shirley named has already resulted in federal action. Fahima Egeh Mahamud, who ran Future Leaders Early Learning Center — one of the businesses in Shirley’s video — was charged and pleaded guilty in July 2026 to wire fraud and conspiracy to defraud the United States. She admitted to falsely certifying that she was collecting required parental co-payments when she wasn’t — which made her center ineligible for the $4.6 million in Child Care Assistance Program reimbursements it received — on top of $850,000 tied to a separate federal food-program fraud scheme. That’s not proof of Shirley’s specific claim that centers were billing for children or hours never actually served. It is proof that a center state inspectors described as “operating as expected” in January was criminally charged the following month, and its owner later admitted to multimillion-dollar fraud in the very program funding it. Shirley was dismissed as fearmongering months before any of that came out.

Five days after the video went viral, Minnesota’s Department of Children, Youth, and Families announced that inspectors had visited the nine centers named in Shirley’s video — Future Leaders among them — and found them “operating as expected.” That finding got wide pickup as evidence Shirley had overreached. But look at what the check actually measured: whether children were physically present and licenses were active on the day inspectors showed up. That’s a same-day compliance snapshot. It says nothing about whether a center billed the state for hours or children it didn’t serve, which was Shirley’s actual claim. It’s also worth noting DCYF is the same agency whose oversight of Minnesota’s child-care assistance programs sits at the center of a fraud scandal that predates Shirley’s video by years, with more than 90 people already charged in related cases since 2022. An agency clearing its own licensees five days after a viral video put its oversight failures in the national spotlight is not a disinterested source.

The federal government didn’t treat the matter as closed the way the “operating as expected” headline suggested. HHS froze child care payments to the state pending investigation within days of the video going viral, and Homeland Security Investigations and ICE were on the ground in Minneapolis before the week was out. Four months later, in April, FBI and HSI agents were still actively raiding businesses tied to Shirley’s video — the former Quality Learing Center location among them — collecting records and computers. The investigation the state’s five-day clearance was supposedly answering was still open when federal agents were carrying out evidence boxes months later.

Two different outcomes for two different centers named in the same video. Future Leaders produced a federal guilty plea. Quality Learing Center — the business whose misspelled sign became the meme of the whole story — voluntarily surrendered its license in January with no reason given by the state, after officials had earlier said they’d found no fraud evidence there specifically. Both were waved off as debunked in the same news cycle.

Same Method, Different Verdict

Thirty-seven years later, California passed a law about a journalist doing something close to what Moore did — except this time, the journalist is alleging something Moore never alleged against Smith: fraud.

Nick Shirley is an independent journalist and YouTuber who built a following exposing what he says is fraud in taxpayer-funded programs — first at daycare centers in Minnesota, then at immigration-support nonprofits and healthcare facilities in California. His method is the Moore method: show up uninvited, film, ask questions the subject doesn’t want to answer, publish.

Here’s the legislative sequence, in order. Shirley’s Minnesota daycare video goes viral in late 2025. He travels to San Diego to shoot similar content at Somali-run daycares in California; two days after Shirley posted that he’d arrived in the state, Governor Newsom says the local Somali community feels “under siege” by “right-wing provocateurs.” Assemblywoman Mia Bonta — married to California Attorney General Rob Bonta — introduces AB 2624 that same month, titled “Privacy for Immigration Support Services Providers,” extending the state’s Safe at Home address-confidentiality program to immigration-service workers; her office says the bill responds to a broader pattern of harassment predating Shirley. Republican Assemblyman Carl DeMaio attaches Shirley’s name to it anyway, calling it the “Stop Nick Shirley Act.” The bill passes the Assembly in May, the Senate later, and Governor Newsom signs it on August 22, effective October 1, 2027. Shirley sues on September 4, represented by America First Legal, arguing the law is an unconstitutional content- and viewpoint-based restriction that chills investigative journalism into fraud at taxpayer-funded nonprofits.

Bonta’s office says the bill was never about Shirley or journalism — it protects workers facing documented threats, full stop. On what the final text actually prohibits, her office has a real textual argument: it bars posting a protected person’s information or image with specific intent to incite imminent violence or create reasonable fear for their safety, not filming, publishing, or accusing someone of fraud. On a narrow reading of that text, a reporter can still show up at an immigration-services office, film, ask hostile questions, name the organization, allege fraud, and publish. Shirley doesn’t buy that reading, and neither does his legal team — which is exactly what his lawsuit is about. Free-speech lawyers, including the First Amendment Coalition, raised real concerns about an earlier, broader version of the bill; lawmakers narrowed it before passage, but narrower isn’t the same as settled, and no court has yet ruled on which reading is right.

IMO, that legal uncertainty is part of the problem, not an answer to it. A law doesn’t have to ban journalism outright to intimidate the people doing it. It just has to make them uncertain whether today’s story is the one that gets them sued, demanded from, or referred for prosecution — and let them find out only after they’ve already spent the money defending themselves. I think that’s what this bill does to reporters covering this beat, whatever its authors intended and whatever a court eventually rules.

So the accurate claim isn’t that California outlawed the Roger & Me method. It’s this: California already has general anti-doxxing law — a criminal statute for publishing someone’s information with intent to cause fear, a civil law letting anyone doxxed sue for damages. Those already covered Shirley, the immigration workers, everyone. AB 2624 didn’t fill a gap. It handed one occupational class a second, class-specific legal weapon that can be invoked against anyone publishing information about them — a civil and criminal claim, plus a formal demand process a reporter has to respond to — on top of the general law that already covered them. A reporter doesn’t have to lose that claim to be hurt by it. Facing a lawsuit or a demand letter over a story you have every right to publish costs money and time whether you win or not — a way to punish a reporter through the process itself, without ever needing a verdict. And that’s exactly what an intent-based standard like this one makes possible: nobody has to prove Shirley intended to incite violence before they can accuse him of it, send the demand letter, or file the claim. Intent lives in someone’s head. Alleging it is free; disproving it in court isn’t. That gap between what it costs to accuse and what it costs to defend is the actual mechanism, regardless of who eventually wins. Roger Smith had no comparable tool available against Moore. The workers named in AB 2624 have it now, against reporters doing something just as legal as what Smith did.

Whatever the bill’s authors intended, here is the actual asymmetry: the confrontational, no-consent style of journalism that made Michael Moore’s career when the target was a legal corporate decision produced no legislative response at all. The same style, aimed at a falsifiable fraud claim, produced a special legal weapon for its targets within months.

The Silence

Michael Moore has not gone quiet on immigration enforcement. He has written at length on his Substack and posted regularly on X about ICE, deportations, and the people he says are being swept up by them — including a widely covered piece arguing that deporting immigrants risks removing the next cure for cancer. He has weighed in on detention conditions, on ICE tactics, on nearly every immigration-adjacent controversy of the last two years.

I can find no public statement from Moore on AB 2624 or Shirley — no Substack post, indexed X post, interview, or podcast discussion. A filmmaker built entirely on the premise that cameras belong wherever the powerful don’t want them has had nothing to say, as far as I can find, about a law aimed at burdening exactly that premise — as long as the camera is pointed somewhere he’s chosen not to look.

Not a New Pattern

This isn’t the first time the same reversal has shown up. In Is Democracy Really Under Siege, and If So, From Whom?, I argued that “democracy under siege” language is less a description of an actual rule change than the sound a party makes when it discovers, after decades of assuming otherwise, that the rules don’t favor it anymore. AB 2624 is one more data point for that argument, but it’s not the only one on record.

I made this case at length in The Democrats’ Love/Hate Relationship With the Supreme Court: nobody called the Court a threat to democracy while it was ruling their way. FDR tried to expand the Court in 1937, when it was declaring his agenda unconstitutional; for various reasons too complex to get into here, that effort was dropped, and the idea sat unused in mainstream Democratic politics for decades. It came back in 2021, when Senate and House Democrats introduced a bill to expand the Court from nine to thirteen seats, a year before Dobbs, once the Court stopped ruling their way. The “assault on democracy” framing arrived with the losing streak, not before it.

The same selective standard shows up in ethics scrutiny, which I covered in Is the Supreme Court Really Partisan and Unethical?: undisclosed gifts and fishing trips became a national media obsession the moment the Court’s majority stopped ruling the way critics wanted, while the revolving door, congressional stock trading, and the post-office book deals and speaking fees that make ex-officials rich — structural corruption that dwarfs any fishing trip — drew no comparable outrage from the same critics. Same category of conduct, opposite level of scrutiny, depending on which branch was doing it.

The same shape shows up in redistricting, which I wrote about in What Is the Uproar About the Latest Supreme Court Ruling on Gerrymandering? Louisiana v. Callais drew a wave of civil-rights outrage this year for a ruling that made it harder to compel a majority-Black district in a state that’s one-third Black. Illinois Democrats, in the same period, drew a map producing fourteen of seventeen congressional seats from a statewide vote split roughly 56-44 — and it generated nothing resembling the same civil-rights framing in the same press corps. Same tool, opposite verdict, depending on who the map benefits.

AB 2624 is that shape again, at statute scale. A confrontational, no-consent style of exposure was a career-making, celebrated genre for four decades when the target was a legal, undisputed corporate decision. The moment the same style produced a specific, falsifiable fraud claim against an immigration-services nonprofit, its targets got an extra layer of legal shielding Roger Smith never had against Moore, plus a governor’s signature and a critic’s nickname stuck to the bill. Nobody has to guess why the reaction changed. The pattern already has data points on the record before this one.

The Doxxing Problem Is Real

None of this means doxxing isn’t a real problem. It is, and the internet and social media have made it worse than anything Moore’s generation of documentarians had to reckon with — a viral clip is permanent and searchable in a way a theatrical release never was. Shirley himself says he can no longer investigate in major cities without a security detail, because activists photograph him and circulate his location to disrupt his reporting. California already has general doxxing laws covering exactly that. AB 2624 added a special layer on top for one occupational class anyway.

The Question Newsom’s Signature Didn’t Answer

Michael Moore never got a bill with his name on it. He spent decades perfecting the exact method that Shirley used — no consent, no cooperation, camera rolling until the subject cracks or the story’s told anyway — and it made him rich, celebrated, and Oscar-decorated instead of regulated. Nick Shirley got a bill with his name attached to it within months of using that same method to allege something Moore never had to prove: an actual crime. On its narrowest reading, the bill doesn’t ban what Shirley does — but Shirley is suing precisely because he doesn’t accept that reading, and no court has settled it yet. What isn’t in dispute is this: California already had a generally applicable doxxing law covering him and everyone else before AB 2624 existed. What Shirley’s targets got, and Moore’s target never did, is a legal weapon Roger Smith never had against Moore, no matter how legal Smith’s conduct was.

IMO, the timeline tells you why: this bill is part of the broader resistance to Trump and Republican immigration enforcement, and its practical effect is to intimidate reporters who go after liberal causes. Nobody has to admit that for it to be true.

IMO, and this is opinion, not something the record above proves on its own: this isn’t really about doxxing, or about Shirley, or even about immigration policy specifically. It’s what happens any time a set of tools — a free press, a Supreme Court, a redistricting map — worked in your favor for long enough that you never had to think about the tool itself, only the outcome. The tool was fine when it expanded rights you wanted expanded, or exposed a corporation that had done nothing illegal, or drew districts that happened to favor you. The moment the same tool turned and started producing outcomes you didn’t want, the complaint stopped being about the outcome and became about the tool — the Court is illegitimate, the map is a civil-rights violation, the reporter is a safety threat. I’ve made this case about the Court more than once. AB 2624 is the same shape, just smaller and easier to see all at once.

Same camera. Same ambush. Same refusal to take no for an answer. The variable that best explains the radically different reaction is who was standing on the other side of the lens.

Sources

Nick Shirley / Minnesota daycare fraud

• NPR: What to know about Nick Shirley

• Fox News: PBS/media reaction to Shirley’s reporting

• Fox 9: Future Leaders Early Learning owner pleads guilty

• CBS/Fox Baltimore: DCYF “operating as expected” finding

• Fox News: Quality Learing Center voluntarily surrenders license

• Fox News: HHS freezes payments, federal “massive investigation” launched

• CNN: FBI/HSI raid Minnesota businesses in April 2026

AB 2624 / California

• SF Chronicle: Newsom’s “under siege” comments, Shirley’s San Diego trip

• CalMatters Digital Democracy: AB 2624 bill text and history

• SF Chronicle: Newsom signs AB 2624

• America First Legal: Shirley’s lawsuit against California

Historical pattern / prior Cranky Old Guy pieces

• Is Democracy Really Under Siege, and If So, From Whom?

• The Democrats’ Love/Hate Relationship With the Supreme Court

• Is the Supreme Court Really Partisan and Unethical?

• What Is the Uproar About the Latest Supreme Court Ruling on Gerrymandering?

• NBC News: 2021 Democratic bill to expand the Supreme Court

• Wikipedia: 2022 Illinois congressional election results

How? The raw trees are owned by the government in Canada. In the US, the land is mostly private. Even if the land in Canada were private, the two countries might not have equal costs. Either way, the way the Canadian government charges for logging is called a stumpage fee — not an open-market price, but a rate set administratively. The claim has been that those fees are too low, and that this makes it impossible for US firms to compete.

The fight was never winnable on its own terms

Ninety percent of Canadian forest land supplying the lumber industry is Crown land. Provinces set stumpage fees by schedule. Most US forest land supplying the industry is private — sometimes owned outright by the mills that cut it, sometimes bought from independent landowners through a negotiated sale or a competitive auction. Either way, when a price does get set, it comes from a buyer and seller striking a deal or bidding against other buyers, not from a government schedule. The US Commerce Department revisits the rate every year in what’s called an administrative review, or AR — a fresh recalculation, with a new number attached, on a clock that never stops. Four of the most recent reviews alone have tried to benchmark the two systems against each other, as if an administered fee and a market price were the same instrument measuring the same thing.

They aren’t, and USMCA and NAFTA Chapter 19 panels have said so, repeatedly — ruling the US benchmark methodology unsound. The US kept using it anyway. AR5 became AR6 became AR7 became AR8, each one re-litigating the same comparison, because the two countries’ forestry systems were never going to converge into something a single price benchmark could fairly measure.

British Columbia already ran that experiment, partially. Since the 2006 Softwood Lumber Agreement, BC has sold roughly 20 percent of its Crown timber at open, competitive auction through BC Timber Sales, then used those auction results to extrapolate a stumpage rate for the rest of the province’s harvest. It’s not a full market conversion — most of BC’s timber still moves at the extrapolated rate, not an actual auction price — but it’s the closest thing to the private-market pricing the US says it wants, built specifically to satisfy Washington. The duties came anyway. Commerce has leaned on other provinces — Nova Scotia’s private stumpage rates, at points — as the benchmark instead, over Canada’s objection that the comparison doesn’t hold. Even a partial, one-province market experiment, built to the US’s own specification, didn’t end the dispute.

The way out Canada already uses — for dairy

There’s a way to sidestep the stumpage comparison entirely: stop trying to align two structurally different pricing systems, and just fix a volume. Canada already does this for its own protected industry. The dairy tariff-rate quota doesn’t ask whether Canadian milk pricing is comparable to US milk pricing. It sets a number and stops arguing.

Softwood never went that route. The US lumber industry chose litigation instead, filing the first countervailing duty petition in 1982, claiming Canadian stumpage fees were an unfair subsidy. That’s a legal remedy under US trade law, triggered by a private petitioner — not a negotiated instrument between states, the way dairy’s quota was. Softwood’s fight was picked by an industry, in court, and has stayed there ever since.

Earlier this year, that looked like softwood was headed to a quota system also. BC Premier David Eby raised quotas as a path to resolving the dispute. Prime Minister Mark Carney said a future deal could include “some element of managed trade” on softwood. For a few months, quotas were being discussed as a real option — the first genuinely new instrument on the table in this dispute in years.

What was actually offered

That’s not what showed up in the room at the end. When the US and Canada sat down for the final push in August 2026, the offer on softwood wasn’t a quota. It was narrower: drop the 10 percent Section 232 tariff — the national-security tariff added in October 2025 — by September 1. The underlying “anti-dumping”/”countervailing duty” (AD/CVD) tariffs, the ones running through AR5 to AR8 since the last agreement lapsed in 2015, would stay exactly where they are.

So the quota talk that ran through the first half of 2026 didn’t survive contact with the actual negotiation. What survived was an offer to peel back the newest layer of the dispute while leaving the oldest layer — the one built on the stumpage argument — untouched.

Why it fell apart anyway

Talks collapsed on August 21–22, and softwood wasn’t the reason. Carney said the deal broke down over US terms he said would “destroy” Canada’s auto, steel, and aluminum sectors — sectors with nothing to do with stumpage fees or Crown land. He also named bilingual labeling as a red line, USTR Jamieson Greer called that a “funny, fake story,” and Carney fired back that it was real. I’ve covered that dispute on its own terms in “Washington’s Complaint About Canadian Bilingual Laws” — the short version is that Carney folded a fifty-six-year-old, uncontested federal labeling law together with a one-year-old, already-softened Quebec trademark rule into a single undifferentiated grievance, because the merged version plays better as a soundbite than either piece does on its own. That’s not a one-off. The same flattening shows up in how he’s handled dairy, covered in “Canadian Dairy Quotas: The Taxi Medallions of Their Dairy Industry.” Trump makes these fights harder than they need to be with the “51st state” routine — see “Does Taunting Canada Have Any Rational Purpose?” — but Carney isn’t a plain-spoken counterweight to that. He’s playing the identical game to his own audience: real grievances and manufactured ones collapsed into one outrage, engineered for the domestic news cycle rather than for accuracy. Softwood was part of the package that fell — Trump had also reportedly offered to cut the auto tariff — but it wasn’t what killed the talks, on either side’s account.

Which means softwood lumber didn’t get resolved and didn’t get rejected. It got shelved, dragged down by sectors that were never part of the stumpage argument in the first place.

Where that leaves it

Forty years in, the pattern holds. A new instrument gets floated — this time a quota, dairy-style, treaty-durable in a way the old side deals never were — and it evaporates before it’s tested, not because either side rejected it on the merits, but because it was never negotiated on its own this time. Softwood has been settled on its own terms before — a 1986 MOU, the 1996 Softwood Lumber Agreement, the 2006 SLA — and none of the three lasted. Canada unilaterally withdrew from the 1986 deal in 1991; the 1996 and 2006 agreements ran their fixed terms and lapsed. Either way, the underlying AD/CVD machinery was always still there, waiting. That’s arguably worse than never settling at all: even a negotiated fix doesn’t retire the statutory process that produced the dispute in the first place. It just pauses it.

The underlying AD/CVD duties are still there, unaffected by any of this. AR8 is still running its own review cycle, on its own schedule, indifferent to whether the two governments are speaking to each other. Whatever the next opening looks like, it’ll have to survive being attached to something else again — because in this dispute, it never gets to stand on its own.

That’s the entire confirmed record. Everything else arrived over the next 72 hours as a chain of unnamed-official claims, each laundered by the next outlet into something that reads like established fact by the third retelling.

The premise

Here’s the thesis this piece is going to argue, stated plainly and up front, because it’s unproven and should be read that way: this was a personal message from Trump, requiring total secrecy and zero chance of leaking, with Ratcliffe as the courier. Everything CNN, the Journal, and Axios reported — NATO warnings, Iran pressure, an “escalate to de-escalate” intelligence assessment — is noise. Real reporting, sourced to real people, but missing the actual purpose of the trip by a country mile, because the actual purpose was never going to be leakable in the first place.

What follows is the case for that reading, built only on what’s confirmed.

What’s actually confirmed

Strip out every “according to sources familiar with the matter” and six facts remain, all attributable to a named person or a verifiable data source:

1. Ratcliffe flew to Moscow aboard a U.S. Air Force C-17, callsign REACH 4555 — a class of Air Mobility Command aircraft carrying almost no identifying markings beyond a flag and a low-visibility roundel. Confirmed by flight-tracking data and aviation trackers, not by either government.

2. Kremlin spokesman Dmitry Peskov initially denied any knowledge of the aircraft the morning it landed — “no such information,” “no such plans.” He confirmed the visit later that day, and by the next day said talks had taken place with Russian intelligence officials, not Putin, who was briefed afterward, according to Peskov.

3. Peskov called U.S.-Russian relations a “deep crisis” and said it was too soon to know whether the talks would change that.

4. Trump confirmed the trip and called it “semi-routine,” explicitly denying it was meant to warn Russia off attacking NATO.

5. The CIA has issued no statement — no confirmation, no denial, nothing.

6. This is Ratcliffe’s first known trip to Russia as CIA director.

One of the six deserves an asterisk: the no-Putin-meeting claim is just what Peskov said, not a fact.

Why the reported explanations are noise

NATO warnings, Iran pressure, escalation assessments — all plausible things the U.S. might discuss with Russia. None of them require the head of the CIA to personally, and secretly, fly to Moscow. The explanations are a grab bag of whatever’s already in the news, buttressed by the usual invisible people corroborating each other.

Why the Burns comparison doesn’t apply here

Several outlets reached for the same frame: the last CIA director to visit Moscow was William Burns in 2021, warning Putin directly against invading Ukraine — invasion followed three months later. Two unconnected events from different administrations years apart does not allow the first to explain the second.

Why it had to be Ratcliffe, and why in person

Modern U.S.-Russia intelligence contact doesn’t require the director to board a plane. Liaison work, deconfliction, prisoner-exchange coordination all run through station chiefs and secure channels without a principal leaving Washington — a small prisoner exchange happened the same week through exactly those ordinary channels, no CIA involvement reported. So the question isn’t why contact happened. It’s why this contact needed Ratcliffe himself, in person, unannounced, off the books.

A spoken, undocumented conversation doesn’t eliminate every disclosure risk — participants can still describe it afterward, and people can be questioned even without a written record. But every additional document, agency, and government a message passes through creates another point where it can leak. An in-person conversation between two intelligence principals, with nothing put in writing, collapses nearly all of those points into one room — not because Moscow can be trusted, but because Washington can’t be.

The plane sat on the ground in Moscow for roughly eight hours. That’s not a handshake and a sealed envelope — that’s time for a real exchange, possibly documents changing hands or supporting material Ratcliffe needed to reference or show, not just a verbal message delivered and received. Whatever it was, it wasn’t something Witkoff was able to know or handle. The length and apparent detail may also explain why this wasn’t a call or meeting Trump conducted himself — a message this involved needs someone who can sit with it for hours and handle the specifics, not just carry a short, personal line. Ratcliffe is also one of Trump’s most trusted people — not a career officer inherited from a prior administration, but a loyalist appointee, which matters for a message this sensitive as much as clearance does.

Who it had to be kept from

Washington routinely shares assessments with NATO allies and Kyiv — that’s how Ukraine got told to pause strikes near Moscow in the first place, through entirely normal channels. Anything written down about the actual conversation carries that same risk of circulating to allies, to Kyiv, potentially to Congress. A message that skips interagency clearance and allied notification isn’t an intelligence community position — those get briefed and filed by definition. It’s something with no institutional owner at all: a message from one man, carried by one man, meant to exist nowhere else.

Where that leaves us

We may never know what was said, and reporting three separate cover explanations in one week isn’t evidence anyone got closer — it’s evidence of how well the concealment worked. Two governments that agree on almost nothing right now agreed, implicitly, that this conversation would not be written down, characterized, or confirmed in any retrievable form. Everything published about why is somebody’s honest reporting on the wrong story. The secrecy itself, and the fact that it succeeded, is the only finding sturdy enough to stand on.

Here is the basic idea, before any of the complication.

Canada agreed to let a certain maximum amount of American dairy into the country each year, duty-free. Trade lawyers call this a tariff-rate quota, or TRQ — a quota that only covers a set volume, with steep tariffs kicking in on anything above it. That’s the number everyone reports: the quota. What almost nobody reports is that the number isn’t what actually crosses the border. It’s a ceiling, not a guarantee. How much of it actually gets used in any given year is not fixed by the agreement at all — it’s decided, year by year, by Canadian companies operating a set of allocation rules complex enough that describing them accurately takes several thousand words. In practice, usage has averaged around 40 percent of the maximum. In a given year, it could be less. It could be zero, and nothing in the agreement would stop it.

Is that fair? I don’t think anyone would look at that arrangement and call it fair, once they understand what it actually is: a promise Canada never has to keep.

There’s also a more specific grievance sitting on top of the general one: on cheese, Canada lets European retailers buy and sell directly under its trade deal with the EU, and doesn’t extend the same eligibility to American retailers under the US deal. On the identical product, the US gets the worse deal. It’s the one piece of this Washington has actually cited as formal legal grounds for the tariffs now in effect — two rulebooks, same country, same cheese, and the tighter one is the one written for the United States. But it’s an add-on to the real complaint, not the substance of it. Washington doesn’t like the whole arrangement.

There is nothing deeper to this part of the trade dispute apart from bogus excuses from Canada.

None of this gets a fair hearing right now, and it’s worth saying why. Trump’s own conduct — the insults, the “51st state” routine, renaming Lake Ontario to “Lake America” by executive order, the tariffs imposed by fiat — makes it easy for anyone watching to write off every American complaint as more of the same bluster, without checking whether this particular one holds up. And Carney has folded this dispute into the same undifferentiated pile as “Washington’s Complaint About Canadian Bilingual Laws”, which is also very overblown and misrepresented by him.

If you want to know what’s actually behind the convoluted dairy system in Canada — which is not a free market system at all — then read on.

Before Any Import Ever Enters the Picture

Start with a version of this system that has nothing to do with the United States at all.

In most of Canada, you cannot legally sell milk in commercial volume without first owning “quota” — a government-created permit specifying exactly how much you’re allowed to sell. This has nothing to do with tariffs, trade deals, or imports. It’s a purely domestic rule, and it applies with equal force to a Canadian farmer down the road as it does to anyone outside the country.

You can still buy land in Canada. You can still build a barn, buy cows, and produce all the milk you want. What you can’t do is sell more than your quota allows — sell beyond it, and depending on the province, you either don’t get paid for the excess or face a financial penalty. Production itself was never restricted. Selling it is, and selling requires a permit that isn’t earned, it’s purchased, on a secondary market, from someone who already holds one.

The closest American equivalent isn’t another farm program. It’s the taxi medallion.

In New York and other cities that ran medallion systems, you couldn’t legally operate a cab without owning one — a government-issued permit, fixed in supply, tradeable on the open market, worth whatever the market would bear. It had nothing to do with whether you could drive, or whether you owned a car. It was a separate, purchasable right to participate in the business at all, and it became eye-wateringly expensive precisely because the number of medallions never grew while demand for cabs did.

Canadian dairy quota works the same way, medallion for cow. A single cow’s worth of quota ran $25,000 in Ontario and Quebec, up to $42,500 in British Columbia, as of 2015 — meaning a typical 70-cow farm required more than $3 million in quota alone, before a single acre of land or a single barn. Like the medallion, it was originally handed out for free to whoever was farming when the system began, in the early 1970s. Like the medallion, decades of a fixed supply against rising demand turned a free permit into a multi-billion-dollar asset. Nationally, Canadian dairy quota is now valued in the tens of billions of dollars — an appreciation that reflects nothing about how much better Canadian milk has gotten. It reflects only how much more valuable it’s become to be one of the people already holding the right to sell it.

This is the part of the system that predates, and doesn’t depend on, any question about American imports. Even if the United States didn’t exist, a Canadian who wanted to become a dairy farmer next door to an existing one would face the identical wall: not “can you produce good milk,” but “can you afford to buy someone else’s permission to sell it.” That’s worth understanding on its own before getting to the part of the story that does involve the US — because it shows the system isn’t really about protecting Canadian dairy from foreign competition. It’s about protecting the people who already hold quota from anyone at all, foreign or Canadian, who might want to compete with them.

And the quota isn’t the only lever. Once you hold one, the price you’re paid isn’t set by the market either — it’s negotiated by provincial marketing boards on behalf of every quota holder at once, with a national reference price set annually by the Canadian Dairy Commission using a formula, not a market. That’s a second layer of control stacked on top of the first: not just who’s allowed to sell, but what they’re guaranteed to be paid for it. The parallel with taxis in medallion cities where the city sets the rate holds.

The complaint, stated plainly

Canada routes the entire US import quota to dairy processors and distributors — the companies that turn milk into cheese and butter, or buy from those companies — rather than to retailers or anyone else who’d actually have a commercial reason to import regardless of their own supply position. A processor only imports when it’s short of something it can’t source domestically. When its own production already covers what it needs, there’s no reason to bring in a competing product, and nothing in the system requires it to. That’s who’s deciding the real number every year, and why it isn’t the number Canada negotiated.

The rest is Canada’s problem, not Washington’s

Ottawa defends the system as price stability, food security, and a fair return to farmers. None of it holds up: “stability” means consumers pay a premium so quota holders don’t have to compete, “food security” means capping supply rather than building resilience, and “fair return to farmers” has presided over a 92 percent collapse in farm count while the asset protecting them climbed into the tens of billions.

None of that is Washington’s fight to referee. How Canada chooses to run its own dairy cartel — who it protects, who it compensates — is entirely Canada’s business. What isn’t Canada’s business to define unilaterally is whether the number it agreed to in a treaty means anything.

And that part isn’t complicated, whatever else gets mixed into this fight. American dairy farmers were never entitled to sell every gallon the quota covers. What they were entitled to was a genuine opportunity to compete for it — a market-access system where the incumbents they’re competing against don’t also get to decide how much of that competition they’ll allow themselves to face.