Democrats still think politics runs on money. It runs on attention.

Lex Fridman sat down with Ezra Klein and Derek Thompson for more than three hours after Democrats lost the White House, the Senate, and the House, with no Supreme Court justice of their own appointment and no obvious next nominee. Klein is a *New York Times* columnist and host of *The Ezra Klein Show*; Thompson writes for *The Atlantic* and hosts *Plain English*. Together they have written *Abundance*, a book they present as a manifesto for what the Democratic Party could stand for — and, just as much, a reckoning with how liberalism learned to stop things and forgot how to finish them.

Asked to define the American left, Klein started with unfairness. People are born with different talents, into different countries and families; luck is omnipresent. The left’s job is not perfect equality but a kind of universal dignity, plus skepticism of unchecked capitalism and an instinct to rectify power imbalances — a minimum wage, for workers who have no market power, being one example. Thompson offered a higher-altitude sketch: liberals fear injustice, value change, and tolerate a bit of overreach; conservatives fear cultural radicalism, value tradition, and often tolerate inequality as natural. Klein added a correction that “big government versus small government” oversimplifies. The left is comfortable with the government of the check. The right is comfortable with a very powerful police, surveillance, and national-security state — including the extraordinary faith it took to think the United States could invade Afghanistan and Iraq and rebuild them as liberal democracies.

Then they came down from the Platonic forms to who actually runs things. On the right, Klein said, it is straightforward: Donald Trump, and the people he has nodded toward — Elon Musk, in some degrees JD Vance and Russ Vought. Senate Republicans did not want Pete Hegseth, Kash Patel, Tulsi Gabbard, or RFK Jr., and they confirmed them anyway. Democrats, by contrast, are in fundamental disarray. Hakeem Jeffries and Chuck Schumer could not even agree on whether voting for a continuing resolution was leverage. There is no presumptive next nominee the way Hillary Clinton followed Obama, or Biden and Sanders followed Clinton. “In my lifetime, the Democratic Party has never been as internally fragmented and weak, leaderless, rudderless, as it is right now.” They do not hold the presidency, the House, the Senate, or the Court, and they do not have a next in line.

Fridman asked why the party is so bad at generating an inspiring figure — he named Alexandria Ocasio-Cortez as someone who might be that person — and why Klein had said early that Biden should not run again. Klein’s 2023 argument was that even with Biden locked into the primary, Democrats still had time for something parties had done before: an open convention, a mini-primary whose audience was the delegates. Harris, he said, had been picked for a different moment, a coalitional signal from a white male nominee trying to reconstruct the Obama coalition. By June 2024 there was no time, the administration was unpopular over inflation, and she had no lane on cost of living. Primaries educate candidates. She got three months.

Thompson tried to click two of Fridman’s questions together: why Trump won, and why Democrats seem less willing to sit for an unstructured three-hour conversation. The 2024 result, he said, was overdetermined. Incumbents lost across the developed world at rates not seen in fifty years, largely because post-COVID inflation punished whoever happened to be in office. Democrats were in office. On top of that, twenty-first-century American elections have become bizarrely close: each side gets to about 48 percent, and every contest is a fight over the next 1.5 percent, so small thermostatic swings flip the outcome.

Then there is the communications technology of the age. Radio made FDR powerful in a way he could not have been in the 1890s; the 1960 debate rewarded Kennedy’s face on a screen. The current screen, Thompson argued, is not television on steroids but a different species, and it pays for something like live-wire authenticity — the paradoxical skill of performing uncalculated presence. Trump is a marvel at it even when the audience knows he may be bullshitting. There does not seem to be a background algorithm in his head crafting the line for each coalition. That compares, Thompson said, with a Democratic character of bureaucratic caution: always balancing factions, a little afraid of unstructured talk.

In 2015, Thompson joked, America had four parties: MAGA, the center right, the Bernie wing, and the Biden-Clinton-Obama wing. Trump killed the center right and is now wearing it as a hat. The Democratic Party is still a jungle, which is why its leaders perform coalitional calculation before they walk into a room that might not pay off in a pre-planned way. Klein’s most parsimonious model of the last two elections is even colder. In a roughly 48–48 country, the pandemic inflicted about a 2.5-point incumbent penalty, and Biden won the popular vote by 4.5. In 2024, inflation inflicted about a 2.5-point penalty on Harris, and Trump won by 1.5. You do not need much more than that to explain the result. You do need more to explain the rest.

Klein wanted to stay on this even though they were supposed to be talking about the book. “Democrats still think the currency of politics is money, and the currency of politics is attention.” Harris came in and raised a billion dollars in record time — more money than Trump — and used it to buy attention. On the left, being good at social media means having a team at headquarters that posts memes. On the right, it means Vivek Ramaswamy, JD Vance, Donald Trump, or Elon Musk: you personally are the product. Harris, Klein granted, was a hell of a performer on the stump and had trashed Trump in the debate. She does not do social media herself, because she is not going to take risks. Most Democrats still optimize attention by avoiding negative attention.

The Trumpist wing understands that a volume of attention is itself good, and that you only get a critical mass of it if you will accept the negative kind. Agenda control does not come from positive coverage. It comes from conflict: you do things the other side disagrees with, so they enter into a functioning agreement to keep your issue at the front. That does not make you popular. Nate Silver’s count, Klein said, was that it took 221 days for Biden’s net favorability to go negative, and something like 55 days for Trump in the second term. Trump is not optimizing for favorability or for big wins. Democrats felt thrashed in 2024; his popular-vote victory was the smallest since 2000.

The deeper puzzle is the monopoly. From the golden escalator in 2015 through 2020, American politics was about Trump. From 2020 through 2024, with Biden in the White House, it was still about Trump. Biden was an intentional void. When he went back on the trail, he went to Valley Forge and talked about January 6. Klein’s tell that the 2024 bid would fail was the missing Super Bowl interview in 2023: a president behind in the polls refusing the biggest free television of the year meant the campaign did not trust him to do an extemporaneous interview. “If you can’t turn your candidate into the product, if you don’t trust your candidate to be the product, in an election, you’re fucked.” Attention is the coin of the realm. What makes Trump good at politics is also what makes him bad at it. For Democrats, caution is optimized for a different attentional era.

You cannot change American politics, Klein said, and you cannot change the Democratic Party, if you are not willing to upset people. Trump reformed the Republicans by fighting Republicans: he ran against George W. Bush, Jeb Bush, Mitt Romney, and the trade deals treated as sacred cows, attacking Romney and McCain right after their races. The Obama coalition is exhausted. The party will not take a different form unless standard-bearers are willing to say they were wrong about some things and have to speak and act differently. Thompson did not begrudge the left its intra-party fight. In an anti-institution age, Democrats are at risk of becoming the party that always defends the establishment — “an absolute death nail.” *Abundance* is an attempt to channel anger at the status quo in Democratic cities and states toward housing, energy, and competent government, oppositional in a way that is constructive rather than merely destructive.

Fridman wanted a pretty bow: what does he need to do to get AOC for three hours? Klein would not pretend to know her office. He did think you can see Democrats taking different media risks — Gavin Newsom podcasting with figures including Steve Bannon; Pete Buttigieg likely holding his powder until 2027 if he runs. The three-to-four-hour ask, Klein said from his own booking experience, is hard even for 90 minutes. Fridman suspected fear of saying the wrong thing. Klein suspected culture. Politicians are scheduled by schedulers. Reporters do not get three hours, donors do not, policy staffers do not. When an interview requests it, the default is no — unless AOC herself, or Buttigieg himself, tells the staff to make time. They have the hours if a child is graduating. The hours have to come from them.

Thompson offered a riot theory. The person who throws the first stone has to be very courageous; the 101st does not. One good interview — say Wes Moore articulating a vision of 2020s liberalism that clips well — and the risk valence switches: leaving the appearance on the table starts to look like leaving money on the floor. Klein pushed back on waiting for poll leaders. He had just booked Jake Auchincloss, a not-that-well-known House member from Massachusetts, because the man thinks out loud. It is not 2028. Pre-running the primary is stupid; the world might be in World War III, or there might be AGI. Bernie Sanders, in 2015, was the guy in the wilderness whom the left treated as a consolation prize after Elizabeth Warren did not run. He still has the instinct that the most important thing is to find a place that will let you talk, and let you talk for a while. Thompson’s version: they came on because they have something to say. They wrote a 300-page book. As a writer, he wants his software running on as many pieces of hardware as possible.

The last chapter of *Abundance* borrows from historian Gary Gerstle the idea of political orders: periods with a structure of consensus and a zone of conflict, tacitly accepted by both sides. The New Deal order was founded by FDR and entrenched when Eisenhower accepted it, held in place by the Soviet Union as external antagonist. Stagflation and Vietnam opened the neoliberal order, founded by Reagan and entrenched by Bill Clinton’s “era of big government is over,” with the fall of the USSR as proof point. That order, Klein and Thompson argue, collapsed under the financial crisis, climate change, and China. Integrating China into the WTO was supposed to liberalize it. China became more authoritarian, built an industrial base, and became a country you do not want to have outsourced your key manufacturing to.

Once an order falls, things that were ridiculous become possible: a self-described socialist nearly winning a Democratic nomination; Trump running against trade, against Bush-McCain immigration liberalization, against Iraq. Nothing has found an equilibrium. Both parties now take a harder line on China — Biden kept and strengthened Trump’s policies — but they have not agreed on the rest. Klein’s warning was not to over-code people, platforms, and podcasts. Elon Musk was an Obama-era liberal in 2012. A lot of MAGA-curious people in 2025 will feel differently in 2028 if the experiment fails, and they will not want the last Democratic Party back either. Reaching independents through Liz Cheney-style never-Trumpers misunderstood the independent: they are independent because they also do not like *your* party. You have to explain why they should like you, which means accepting some fault.

Underneath the order talk is a shift from demand to supply. After the financial crisis the problem was too little demand. Behind that, Klein said, a cost-of-living crisis had been building: housing, health care, elder care, child care, higher education. His wife, Annie Lowrey, wrote a viral 2020 piece, “The Affordability Crisis,” asking why everyone was furious when consumer confidence looked terrific. Then the pandemic, then inflation, and the public’s attention moved from getting people back to work to not having enough of anything. Democrats in blue states failed on cost of living, which is why California, Illinois, and New York have been losing hundreds of thousands of people to Florida, Texas, Arizona, and Colorado. They did not make enough of what people needed, and in many cases they made it too hard to make it. Trump has said he won because of groceries. Democrats believe cost of living is his weakness. Klein’s retort: they are probably right, but they do not have a strength on it.

Fridman called *Abundance* a manifesto for what the left could represent. Thompson’s 2022 essay on the subject started literally outside, in 21-degree weather, waiting for a rationed COVID test two years into the pandemic. Tests were scarce, vaccines had been scarce, PPE had been scarce. Zooming out, he thought the 21st-century economy had been defined by scarcity — especially housing unaffordability in productive, often Democratic cities, plus clean energy that environmentalists on his own side had made hard to site, plus a pandemic that constricted supply worldwide and sent inflation global. Their definition, on page one of the book: America needs to build and invent more of the things it needs. Housing, energy, science and technology — and, at the heart of it, government that can actually finish the job.

In the New Deal decades, liberalism transformed the physical world. In the last half-century, Thompson argued, it became very good at the politics of blocking rather than building, and at defining success as how much money you can spend rather than how many things you can put on the ground. California authorized more than $30 billion for a high-speed rail system that basically does not exist. The mayor of Chicago had just bragged about spending $11 billion on 10,000 affordable units — $1.1 million apiece. San Francisco built a public toilet for $1.7 million. The book is trying to nudge Democrats back toward thinking about what Americans need and making it easier for government to provide it, beginning with housing and energy.

Klein wanted to uncouple “deregulation” from its right-wing coding. Sometimes the market should be deregulated, sometimes regulated more — he does not want coal-fired plants to be legal. What does not get talked about enough is how much the government regulates the government. Friends on the left tell him the answer is social housing; Singapore does a great job of it. Look at Los Angeles: voters passed a bond of about a billion dollars for affordable housing, and six years later a couple thousand units had come in at $600,000 to $700,000 each, more than a market-rate home in Denver. Public money layers on rules market developers skip: union or prevailing wage, greener codes, design concessions, grants that require you to house homeless veterans before you can open, a second disability review that wants the doors three inches wider. Klein calls it everything-bagel liberalism: enough toppings make a bagel; everything everywhere all at once makes a black hole. He has a hundred disagreements with Musk. He agrees it is functionally illegal to build high-speed rail in California. Europe, Japan, and China build it. The United States has regulated a public project until you cannot finish it.

Fridman asked about the tension between progressive redistribution and “getting out of the way.” Klein said the sharp tension is between equality and constricting the supply of housing. When you choke supply, people who bought in 1977 for $220,000 sit on a $2.7 million asset and working-class families cannot live there anymore. That is not even a redistribution debate; opportunity requires enough of the thing. He is pro more redistribution than the country currently does. Operation Warp Speed is his example of the two tasting good together: public money and DARPA underwrote mRNA, the Trump administration cleared regulatory croft, and the political demand was that the shots would be free, sorted by need — not an Ozempic-priced luxury. Single-payer, if Bernie Sanders had passed his unusually expansive version, would have required an immediate surge of doctors and nurses the United States currently constricts with residency, licensing, and immigration rules.

Rental vouchers in a locked San Francisco market, without more building, are what friends at the Niskanen Center call cost-disease socialism: you increase demand for a good whose supply you have pinched, and the price goes to the moon. Higher education has a version of the same story — Pell Grants and subsidies without enough new slots, so colleges compete on gyms. Thompson put it from first principles. If you wanted American poverty as close to zero as possible, you would do a lot of demand-side policy: child tax credit, earned income tax credit, cash welfare. If housing is the biggest line in the family budget and your only housing policy is more vouchers, prices go up. Demand-side and supply-side progressivism are peanut butter and chocolate. The political problem is that Americans do not believe they work together, because they do not see the spending.

The Biden administration authorized $42 billion for rural broadband and practically none of it was built; $7.5 billion for EV chargers, same story; tens of billions for California high-speed rail you cannot ride. Thompson’s metaphor is Equinox liberalism. People will pay $270 a month for a gym where the equipment is free, the locker room is clean, and there is Kiehl’s lotion after the shower, because they can see the money at work. Swedes and Danes pay much higher taxes and seem happier for the same reason. The liberalism that blocks rather than builds asks Equinox prices and delivers a broken gym: collapsing infrastructure, crime, housing prices on the moon. Pull the supply levers, especially on housing, and people might even support tax-and-spend more — because they would finally be in the club that works.

Housing, Thompson said, is life and even freedom: the freedom to live where you want, to feel that a good life is achievable. A paper called “The Housing Theory of Everything” makes the point that whatever you care about in public policy, housing probably touches it — innovation and density, child-care costs, living near friends and family. Americans are leaving expensive Democratic cities and states for sunnier, cheaper, often Republican ones. Starting in the 1960s and 1970s, zoning, historic preservation, and laws that made it easier to sue to stop a project outfitted a very old human preference — keep the neighborhood the same — with a weapon. One study reported by Thompson’s *Atlantic* colleague Yoni Appelbaum found that a 10 percent increase in a city’s progressive vote share came with a 30 percent decline in houses permitted.

Klein pushed the stakes past four walls. Cities are where wealth is created, not just displayed; they are meant to be escalators into the middle class, not penthouses for the upper class. Almost every frontier AI lab in the United States, outside China, sits in about 50 square miles of the Bay Area — not New York, Dallas, Chicago, or Austin. Spillover is the ancient path of mobility: you make more as a barber near Google than as a barber in rural Arkansas. About a third of twentieth-century American mobility, on some calculations, came from people moving to richer places. In the last twenty or thirty years that convergence has gone into reverse. It used to make sense for the janitor and the lawyer to move to New York; now it only makes sense for the lawyer. Michael Bloomberg called New York a luxury good that costs luxury prices. That is true, Klein said, and it is a terrible inversion of what a city is. Gate the cities and you have closed the American frontier — not of land, but of the economy.

He has a hobby of reading memoirs and noticing when they are secretly housing stories. Moby’s first memoir is a New York cheap-rent story. Lizzie Goodman’s *Meet Me in the Bathroom* is an alt-rock revival that could afford to live there. The Bay Area’s mix of queerness, psychedelic counterculture, and defense culture that became Silicon Valley was a cheap-housing story. If only one type of person can pay the rent, the place becomes a monoculture. Raj Chetty’s research, Klein noted, shows that children invent in the domain of the place around them: grow up in the Bay Area, and you patent in technology, as Jobs and Wozniak did, because you lived around those people. This is not a technology problem. Elijah Otis invented the elevator in the 1850s. California’s permitted housing over forty years is a squiggle down because people wrote laws, filed lawsuits, and judges accepted them — manufactured scarcity. Undoing it is law, mindset, and political courage: a mayor looking at fifty homeowners in a council chamber and representing the 10,000 who could not take the afternoon off.

If you swap the word “regulation” for “rules,” Thompson said, people already know some rules are good and some are stupid. Mid-century America was disgusting: 1943 Los Angeles smog so black people thought Japan had launched a chemical attack; New Hampshire rivers running green and purple from textile dye; the Ohio River on fire. The Clean Air and Water Acts regulated outcomes — pollution levels, tailpipe emissions — and they worked. NEPA and its cousins regulated process, empowering lawsuits and reviews that can wrap any construction in red tape forever. The month after Gavin Newsom signed an end to single-family zoning, San Francisco’s board of supervisors killed a 500-unit building, including 100 below-market units, on a Nordstrom parking lot, because the paperwork under environmental review was wrong. Solutions of one era become problems of the next. The same tools that cleaned the air in the 1970s are now used to block dense housing and clean energy.

Klein reached for Mancur Olson’s *The Rise and Decline of Nations*: affluent, stable societies grow bargaining organizations that entrench themselves, and they select for the skills their economy rewards. China in its developmentalist phase wants civil engineers and semiconductor people. America, as a price of affluence, became a country where negotiation is the game, so it selects for lawyers, finance, and management consultants. Patrick Collison’s line, quoted in the book: it is tough to be a high-speed rail engineer in America, so people who want to build go into bits and bytes, the frontier of last resort. Donald Trump’s real-estate career, in Maggie Haberman’s telling, was a relational New York business, not a national homebuilding firm. Between Walter Mondale and Tim Walz, Klein said, citing Michigan law professor Nick Bagley, there was not a single person on a Democratic presidential ticket who did not go to law school. Lawyers see legitimacy as following process. Bagley’s rebuttal is that government attains legitimacy at least in part through outcomes. You tell voters you are the ones who know how to run government. They do not see it. Eventually somebody comes through the wall like the Kool-Aid Man, and wins.

Fridman asked them to steelman Elon Musk and DOGE, noting that some people read *Abundance* as the anti-DOGE. Thompson’s easy version: if government is inefficient, a Department of Government Efficiency sounds like a godsend. The book’s theme is how badly government can fail at its own ends. Klein’s line about deregulating government itself would, in that world, make DOGE a gift — especially for liberals who believe government matters. Klein offered a different steelman, closer to what the operators say. The bureaucracy is an interest. This was never about efficiency. It is zero-based budgeting as demolition: fire huge parts of it because the Department of Education is not outcompeted by a three-year-old Agency of Education, and the only way to make radical change possible is to bulldoze what exists, then rebuild it as a neural link of Trump himself. Read the OMB memo that froze funding: the government must reflect the will of the people as expressed through their choice of president. Read Russ Vought. That is unitary executive theory. Klein does not share it. He thinks it is the honest steelman of what they are doing.

Then the other side. Efficiency only makes sense yoked to a goal. Musk has goals at Tesla, SpaceX, SolarCity, even an expressed one at Twitter. DOGE’s goal, if you listen, is Trump. Musk serves at his pleasure and has said he would chisel Trump’s face into Mount Rushmore. The “what did you do this week?” email to some two million federal employees is not zero-based budgeting: DOGE does not have a staff that can read those replies and check whether the work was good. They gutted USAID, including PEPFAR — a George W. Bush program Klein called maybe the highest-value life-saving thing the U.S. government has ever done through foreign aid — without anyone coming in to justify it. They cut the Education Department, one of the leanest agencies by overhead, without saying what a half-size department should do. A judge had just ordered about 26,000 fired workers rehired. Klein supports civil-service reform, and at the beginning Ro Khanna and other Democrats were willing to work with DOGE; they had voted for the Laken Riley bill. The other side did not want a deal, because the point was political control.

What Musk did at Twitter, Klein said, was not make it a better product. He made it controllable: he broke a place where he thought people disagreed with him until they would do what he wanted. That, plus Chris Rufo, plus Vought’s word “traumatize” for civil servants, plus Musk on USAID — “it’s all worms, no apple. We’re gonna feed it to the wood chipper” — is the language of conquering, not of seeing who is doing a good job. Thompson’s concrete example was the FDA: laying off probationary employees does not speed Phase III approvals; it spreads the same work over fewer people. At Tesla or SpaceX, staff could tell you Musk’s goal for the rocket or the robot. Around DOGE, Thompson felt the opposite of a mind meld — not creative destruction, “destruction destruction,” an ideological purge of progressivism performing the job of efficiency. Klein’s through-line: Musk has made it known that any House or Senate Republican who defies Trump on a key vote will face $50 million to $100 million in a primary. That is why, in Klein’s account, Joni Ernst voted for Pete Hegseth. The death star of primary money is how you centralize power in Congress the way DOGE tries to centralize it in the agencies.

Fridman wondered whether, if democracy holds, aggressive cutting is an interesting experiment in unclogging bureaucracy — destruction as the first step toward an abundance vision. Thompson said they are speed-running executive power with the checks taken off, and the early feedback is already in: markets, consumer sentiment, the ten-year yield pointing toward a recession that did not have to happen. Tariffs announced at 9:00 a.m. and taken off at 3:30 p.m. are not a theory of the perfect rate; they are a New York real-estate personality forcing world leaders onto the phone. A 25 percent tariff on Canadian lumber and Mexican drywall, the National Association of Home Builders warned in a March 7 memo, would push housing costs up — the opposite of the grocery-and-rent election Trump says he won. This is not right-wing abundance. It is a personal agenda.

Klein refused to cut tariffs out of the DOGE story, because the through-line is leverage. A stable 10 to 20 percent tariff on all imports and 65 percent on China would at least be a new cost structure firms could invest against. What Trump is doing is on-and-off, because trade deals need Congress and tariffs are a unilateral knob. He used that knob to reshape supply chains and to tell Colombia to take deportees in chains. He used the Justice Department to get the case off Eric Adams, a Democratic mayor who is not his ideological ally, in order to have power over New York. Trump, Klein said, is a very nineteenth-century politician: relational, zero-sum, always looking for the angle. DOGE, tariffs, the Adams deal — you can think it is good or bad, but it is coherent. The remaining question is your normative view of Trump having that much power. Fridman’s was no, because Trump denied an election result; he likes people who can walk away from power.

Klein’s personality diagnosis: the right’s problem is that it is autocratic now; the left’s is that it is bureaucratic. DOGE demonstrated that the room for movement inside the American state is much wider than either party admitted — probably not as wide as the courts will allow, which is why they are losing cases, but wide, and you could also pass statutes. There is no abundance agenda if Democrats remain as process-obsessed as they were for the last decade. Thesis: institutions are great, Democrats defend them. Antithesis: the government is a ball of worms, feed it to a dictator. Synthesis: reform the process at a fundamental level, lawfully, with respect for the lives it touches, more committed to good government’s goals than to the procedures it has grown. Thompson put meat on it with BEAD: $42 billion in the bipartisan infrastructure law for rural broadband, a 14-stage gauntlet of maps, challenge periods, letters of intent, five-year plans, and equity matrices. By November 2024, three of 56 jurisdictions had cleared all fourteen stages. The money died on contact with Earth. A DOGE they would run would name the goal, find the bottlenecks, subtract the blockers, and add accelerators. Liberals, he said, have lost the ability to talk about government failure, which is why they are losing the communications war in an anti-establishment age.

The same logic, for Thompson, applies to science. The NIH is a postwar crown jewel; almost every U.S. biomedical breakthrough bears its fingerprints. It is also a bureaucracy. Surveys suggest scientists now spend about 40 percent of their time on grants and paperwork — as if a virus gave American researchers chronic fatigue every January through June. He does not want the NIH torn down or grants slash-burned. He wants high-risk, high-reward science, not incrementalism that funds older researchers and wastes the rest. If liberals will not speak the language of institutional reform, institutions grow sclerotic, Americans get angry, and they vote for the wrecking ball.

Pressed, as a fan of both men, about an old fight with Sam Harris, Klein would not re-litigate 2018. He had introduced that debate by saying he disagreed with Harris on Charles Murray and still thought Harris was good on meditation, psychedelics, consciousness, and later AI. What he admires is independence: Harris did not follow the “intellectual dark web” into conspiracy, and did not let anger at the left blind him to the right. Fridman asked whether the Overton window on intelligence research had moved. Klein’s ambient sense of the topic had shifted toward the Flynn effect — IQs rising over decades as societies put on “scientific spectacles” — and toward evidence that the rise has stopped. A recent *Financial Times* piece asked whether we have passed peak human intelligence, with literacy and test scores falling. His working line: we spend a lot of time talking about how to get smarter and not enough about how to avoid getting dumber. Phones full of social media feel like information and act like distraction.

He is optimistic about parts of AI and more worried about another part. The easy automations — summarize the reading, write the first draft — are, in his view, where intelligence actually happens. No summary is equivalent to doing the reading and sitting with what it brings up. ChatGPT cannot write the first draft of his book, and the first draft is hard because it is often completely wrong in a way the model will not tell you: you are trying to do the wrong thing, or you have not reported enough. He prints everything and reads at a table, because attention on paper is different. Fridman said the same about his own programming: the tools work, and he can feel himself becoming a worse programmer, handing over the little creative leap. Klein’s worry for his kids is that we will make it incredibly alluring to automate the laborious part of creation, then hope the models can innovate autonomously, because we will have stopped being able to.

What gives them hope about America? Thompson answered as temperament: he is a profound optimist. He pointed to mRNA cancer vaccines, CAR-T, and GLP-1 drugs synthesized from lizard venom for type 2 diabetes that now seem to cut body-wide inflammation and might touch dementia — an accidental medicine for Alzheimer’s from a lizard’s tongue. He wants AI for cross-domain synthesis, Edison’s 10,000 materials at machine speed, CRISPR meeting cancer science. Housing and energy are in the book because they are how you live a good life; health is the rest of it. Institutions still have to be right, or the discoveries will not be instantiated. Klein is less temperamentally sunny. He would usually give Thompson’s answer, and instead gave a realist’s one: this is a more liquid, high-variance period than the 1990s, more like the late nineteenth and early twentieth centuries. That is peril and possibility. What if AI is a boon, clean energy delivers genuine abundance and desalination, cultivated meat ends the annual killing of tens of billions of animals, and a thesis-antithesis-synthesis actually reforms government? “It’s not hope. It’s like that’s a future you have to create, a future you have to call into being, a future you’ll have to fight for.” He is not hopeful about 2030 as a forecast. He believes enough remarkable outcomes are possible that the present is worth being in, and worth the small part of bending the arc toward something more just.