I recently got the chance to walk through the exhibits of the Barack Obama Presidential Center before its opening on Thursday. Looking at photos of the young Obama as a community organizer in Chicago, watching video clips of those iconic speeches that marked his dream-like rise from a nobody to leader of the free world, I was reminded of a time when our politics was more hopeful, our social sector was more constructive, and the two were aligned in the service of the nation.

I first met Obama in 2004 at a campaign event when he was running a distant third for the Democratic nomination for U.S. Senate in Illinois. There were maybe 25 people milling about a modest condo belonging to early Obama fundraisers Alan and Andrea Solow. I told Obama about the various programs my nonprofit organization, Interfaith America, was running to bring people of diverse faiths together to solve social problems.

Persuasion events

The early 2000s was a golden era for the social sector. It was a time when talented college graduates were just as likely to start nonprofit organizations as they were to head to Wall Street. Civic institutions like Ashoka, Teach for America, the KIPP network of charter schools, Public Allies, Partners in Health, the Harlem Children’s Zone and City Year were broadly known and widely admired. These organizations shared a common ethos: to recognize that talent exists everywhere, to build excellent institutions to nurture people’s potential, and to cooperate across differences to lift people up.

Obama was a product of the nonprofit world. His formative years were spent as a community organizer, a profession that relies on storytelling to bring people together and whose cardinal rule is “never do for others what they can do for themselves.” He served on the board of the Chicago-based philanthropic foundation the Woods Fund, which made grants to community-based organizations. He had been the youngest participant in the political scientist Robert Putnam’s 1997 Saguaro Seminar, a workshop for leaders committed to strengthening social capital. And his wife, Michelle, had been the Chicago director for Public Allies, an organization that trained young people to be civic leaders, and then found them a one-year placement in a nonprofit organization.

In that post-9/11 era, my friends and I in the social sector were thrilled by Obama’s rise. Here was a black man with a Muslim name, raised by a single mother, shaped by his work in Chicago communities, staking his claim to the nation’s story, and showcasing his talents to lead us into the next chapter.

In the Obama narrative, America is a nation defined by people who fought for freedom and defeated repression. The 1776 generation won a victory over the British Empire and established the foundational promise of the nation: that all people are created equal and have the God-given right to life, liberty, and the pursuit of happiness. Black people fought against slavery and segregation, and women fought against misogyny and patriarchy, so that promise could be extended to everyone. Veterans fought repression around the world. Immigrants fled economic, religious, and political persecution in their home countries, and came here to build a better life, and through that process America became a better country.

There is a genius to this narrative. It recognizes the ugly parts of American history without getting mired in them. It hands the victory not to the slave drivers but to the freedom fighters. It identifies a common, heroic thread that connects a wide range of communities. And it provides clear direction and forward momentum for the future.

From his very first appearance on the national stage, at the Democratic National Convention in 2004, Obama was explicit about what that future should look like: teach the black child on the South Side of Chicago to read, ensure the senior citizen can get access to her prescriptions, protect the Arab-American family who is suffering discrimination, create job training programs for the laid off white factory worker.

In the heady days of that first presidential campaign, the nonprofit leaders I knew were fully aware that Obama was casting us as a central character in the American drama. Obama-ism was the belief that the talent development, character formation, and civic bridgebuilding of the social sector could be elevated into a governing philosophy. We embraced the challenge. Our work was concrete evidence of his famous “hope and change.”

As president, Obama turned that governing philosophy into concrete policy. He established two White House offices to work with the nonprofit sector, the Office of Social Innovation and the Office of Faith-based and Neighborhood Partnerships. He increased funding for charter schools. He created “Promise Neighborhoods” across the country, based on the model of the Harlem Children’s Zone. He launched a major White House initiative called My Brother’s Keeper that partnered with nonprofit organizations to invest in black men. Even his signature government initiative, the Affordable Care Act, relied heavily on nonprofit organizations to be “navigators,” helping enroll people in Obamacare. He even partnered with my organization to launch a program called the President’s Interfaith Campus and Community Service Challenge.

Increasing social investment in communities that had experienced marginalization was one part of the Obama philosophy. The other part was calling for more personal responsibility. Obama addressed minority youth in blunt terms: Yes, you will experience racism, he would tell them, but you cannot let it prevent you from trying your best. Your ancestors overcame far greater obstacles than you. Focus on the things you can control, not the things that are holding you back. View your identity as a source of pride, not of victimization. Remember that you owe something to your forefathers and foremothers. Never make excuses.

Every narrative is a script for forward action. If you are moved by Obama’s “Yes, we can” interpretation of American history, then it inspires you to do a set of things: Yes, we can help children achieve academically, no matter their circumstances. Yes, we can bridge the divide between urban America and rural America. Yes, we can transition industrial-era workers to knowledge economy employees. We are all agents of constructive change in this story.

But this talk of personal responsibility angered a set of progressive intellectuals, who accused Obama of emphasizing respectability politics over addressing structural racism. In 2014, Ta-Nehisi Coates published a blockbuster article in The Atlantic called “The Case For Reparations.” It became the cornerstone of a worldview that presented a stark contrast to the Obama story: America was not defined by the expansion of freedom, but by the legacy of slavery.

What Coates wrote was true—systemic racism is very, very real. And he was right that Obama rarely talked about that dimension of American life. But Coates was also not telling the whole truth. He ignored the part that Obama emphasized—slavery and segregation really were abolished, and the progress of minority communities in the United States since then has been stunning, if uneven and incomplete. Coates’ own rise to intellectual stardom is a signal case.

Subscribe now

In the Coates narrative, you are either a victim of America’s continuing legacy of slavery, or you are a villain. Thinking of yourself as a victim is terrible for your mental health, and ensures failure before you begin. In fact, it sets up a perverse incentive structure: to prove that America holds minorities back, you have to present yourself as a victim or a failure.

Moreover, the white men cast as villains in the Coates story just so happened to be the group who experienced the most dramatic increase in mortality rates during the 2000s, a spike dubbed “deaths of despair” because so many of them had to do with suicide, alcoholism, and opioid addiction. A handful of white men in woke circles might have embraced the term “privileged oppressor,” but the laid-off factory workers dying from fentanyl overdoses in Macomb County, Michigan were not having it.

In 2016, a Black Lives Matter activist from Chicago publicly spurned President Obama’s invitation to visit the White House and discuss civil rights issues with the great John Lewis. Progressive activists were embracing the “America is a continuation of slavery” narrative. I viewed this as par for the course for activists. What surprised me was how that narrative started to shape the worldview of mainstream social sector organizations. Seemingly overnight, the conversation went from having a “Yes, we can” orientation to being obsessed with dismantling systems of oppression and challenging white supremacy. Nonprofit organizations adopted the Coates worldview, not as a critique of Obama-ism, but as an alternative to it. The clenched fist replaced the extended hand as the symbol of social change.

Charter school networks that once told families that their children could achieve academically whatever their circumstances changed their tune entirely and started to claim that black and brown kids were too oppressed to learn. Minorities were now valued for talking about their trauma, not their talents. Cooperation across difference was viewed as doing harm to BIPOC communities. The laid-off small town factory worker did not deserve a job training program—he needed to be reminded of his white male privilege and scolded for not stating his gender pronoun. Canceling coworkers over microaggression became standard practice in nonprofit organizations.

If your governing philosophy is to elevate the social sector to write the next chapter in the story of America’s continuous expansion of freedom, and significant parts of the social sector decide that their real job is actually to protest America’s legacy of slavery and criticize you for not doing enough about structural racism, then your governing philosophy collapses. That is effectively what happened to Obama-ism.

The alternative to Obama-ism did not turn out to be a glorious antiracist revolution followed by a socialist paradise. It turned out to be Donald Trump. A black president with a Muslim name was always going to face racist opposition from right-wing elements in America. But I don’t think those right-wing elements would have had the opportunity to build a political movement that won two presidential elections had the social sector not cannibalized itself.

Still, I am happy to report that things are changing in the nonprofit world. It turns out that when you embrace a “disrupt the system” worldview, the first system that your junior staff will try to disrupt is the institution that employs them. A decade of organizational turmoil was enough. Executives in the social sector are putting a stop to the crazy practice of cancellations within their own shops. Moreover, responsible leaders in the nonprofit world were sobered by the fact that many minorities voted for Donald Trump in 2024, preferring a candidate who called Mexicans rapists to a candidate who used the term “Latinx.” Maybe the nonprofit world didn’t understand BIPOC communities quite as well as they thought.

Whatever the reason, I am hearing a lot more talk about investing in potential and building civic trust than I am about dismantling patriarchy and white supremacy in social sector spaces. Which means it is a perfect moment for the opening of the Obama Center. The goal should not be a nostalgia trip; it should be a revitalization of the central role the social sector can play in the next chapter of America’s glorious story of expanding freedom.

Eboo Patel, a contributing writer at Persuasion, is the founder of Interfaith America and the author of We Need to Build: Field Notes For Diverse Democracy. He served as an advisor on faith to President Barack Obama.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

In this special episode of The Good Fight, recorded at the How The Light Gets In Festival, Roger Hearing moderates a debate between Curtis Yarvin, Minna Salami, and Yascha Mounk on whether liberalism can ever be neutral, what a truly free society would look like, and whether liberalism’s heyday is over. Find out more about the Institute of Arts and Idea…

Read more

British politicos have feverishly spent the last week with their eyes on the parliamentary by-election that took place in the northern constituency of Makerfield yesterday. With the future of Britain’s government in the balance, Makerfield—a few miles west of Manchester—was the setting for a dramatic showdown between the governing Labour Party and right-wing challengers Reform. Most crucially, it was the first step in Manchester Mayor Andy Burnham’s campaign to replace his boss, Keir Starmer, as prime minister of the country.

Starmer’s growing unpopularity (he is now tied with Liz Truss for least popular prime minister ever) invited a series of challenges over the last few months, especially following a disastrous result for Labour in May’s local elections. But none of his opponents, including the ambitious health secretary Wes Streeting, managed to get sufficient backing from other Labour MPs to launch a challenge. The amateurish plotters resigned from the cabinet without drawing blood.

But all along, the real threat to Starmer was far from Westminster. Andy Burnham, the Mayor of Manchester since 2017, has long been gunning for the top job. Britain’s parliamentary system only allows MPs (or, more eccentrically, members of the House of Lords) to become prime minister, so Burnham has been on the hunt for a good seat for months. Last January, Starmer barred him from running in one by-election. But he was no longer in a position to replicate that power move when Labour’s Makerfield MP Josh Simons stood down from Parliament to pave Burnham’s road to Westminster.

Until last night’s results came in, Burnham’s bid seemed to face considerably more jeopardy than Makerfield’s history might suggest. The constituency is a historically safe seat for Labour, even staying true to its colours in the 2019 general election when Boris Johnson plucked most of Labour’s famed “red wall” seats in the north for the Conservatives. And yet with Nigel Farage’s nationalist Reform Party surging across the country, over half of Makerfield’s electorate backed Farage’s outfit in last month’s local election. Polls gave Burnham a single-digit lead over Reform’s Robert Kenyon.

But Burnham trounced Kenyon with a surprise 20-point win last night. It was a potent demonstration of the man’s political talents. Burnham will return to Westminster with the honorary title of “Britain’s most popular politician.” Thirty-five percent of the country has a positive opinion of the 56-year-old northerner, twice Starmer’s approval rating. Burnham is almost certain to launch a leadership bid against Starmer in the coming weeks—and he will probably win.

But who is Andy Burnham? What does he stand for?

A former member of Gordon Brown’s cabinet until 2010, and of Jeremy Corbyn’s shadow cabinet during one of Labour’s most left-wing periods following the 2015 election, Burnham was elected Mayor of Greater Manchester in 2017. Nine years later, his legacy in the city is largely positive. Manchester has been labelled the “star performer of the UK economy since 2008,” with private investment pouring in and employment growth in the top five of European cities over that period. His investment in the very popular (and very visible) bright yellow “Bee Network” of public buses and trams has been popular. Burnham was also instrumental in campaigning for a deal that granted Greater Manchester more devolved power in 2023.

Burnham plans to make this record key to his pitch to the country: an approach he has labelled “Manchesterism.” At its heart is regional devolution, an appealing prospect in one of the most financially centralized countries in the world, in which London holds most levers over infrastructure spending and taxation. Economically, Burnham claims that Manchesterism represents the “end of neoliberalism, the end of trickle-down economics.” He is also a vocal advocate of reindustrialising the UK’s economy.

But there’s a problem: the Manchesterism that Burnham has defended publicly contrasts starkly with the Manchesterism actually in effect in the city. Burnham is a regular advocate for nationalisation, having indeed brought much of Manchester’s transportation into public ownership. But in practice, Manchester’s success predates Burnham’s election, and was boosted by a surge in foreign direct investment encouraged by consistently pro-business political elites over the years. Regarding industrialisation, Manchester’s success has been decisively post-industrial, with an emphasis on the service economy. It’s unclear which vision of Manchesterism—the one in Burnham’s rhetoric, or the one that actually exists—he would seek to implement nationwide as prime minister.

On top of these ideological teething issues, Burnham would also inherit an incredibly tight fiscal situation. The British government spent around £110 billion (8.1% of public spending) on debt interest in the past year—one of the highest levels in 50 years, and a figure that continues to rise. Starmer has already attempted to juggle these impossible constraints. Burnham, likewise, would also face difficult arbitrage decisions. Will he commit to increase defense spending towards the 3% of GDP target? Could he do so without touching the £333 billion the UK is spending on welfare (up from £262 billion only three years ago)?

True, some key metrics are trending in the right direction. Illegal migrant crossings over the English Channel are down, waiting lists for the National Health Service are down, inflation is down. But Britain’s structural economic reality remains dire, something that will hurt Burnham just like it has hurt Starmer. As the churn of Conservative prime ministers over the past decade demonstrated, the polling bump enjoyed by fresh faces is only temporary.

Starmer’s fate will be decided in the next few weeks. But the prime minister has not yet signalled he will leave, and is obstinately clinging on. This could force Burnham to ask Cabinet ministers to heap pressure on Starmer to set out a timeline for resignation. Worse still, he might launch a protracted leadership bid that would take most of the summer.

Subscribe now

The one potential lifeline for whoever becomes Britain’s next prime minister is the fragmentation of the political right. Yesterday’s by-election was a blow for Reform—the scale of the defeat in a constituency that is 95% white and that only a month ago elected a clean sweep of Reform local councillors will hurt.

Even further to the right, Restore Britain, a breakaway party created just this year, scored almost 7%. Replicated across the country, these results could very well cost Reform up to 70 seats in a general election. And while the Conservatives were a distant fourth in Makerfield with 2% of the vote, they gained almost 50% in the Aberdeen South by-election taking place the same night. For leader Kemi Badenoch, this will be much needed ammunition to prove that the Conservatives are still a threat to Reform.

But this won’t be enough to save Labour. The party has to deal with its own challengers, such as the rising popularity of the left-populist Greens. Most importantly, it will need to develop a clear narrative of how to fix the country and start lining up tangible results. It is unclear whether Andy Burnham is the solution to these problems—but right now, he might be the best Labour has to offer.

François Valentin is a senior consultant at London Politica.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

• Our weekly Intellectual Boo…

Read more

This article is brought to you by American Purpose, the magazine and community founded by Francis Fukuyama in 2020, which is proudly part of the Persuasion family.

A specter haunts debates about governance: the idea of benevolent and efficient dictatorship. Where democratic leaders haggle, delay, and pander, the authoritarian ruler simply acts. Where elected governments bend to lobbyists and electoral cycles, a dictator is in for the long haul. This vision—or autocratic temptation, if you will—has been popular as far back as we have documented political theory, and it still has plenty of champions today. Beijing officials invoke it to explain the rise of China; climate activists to argue that the planetary emergency demands that we put democracy on pause; populists to suggest that current institutions are broken and that a fresh start and setting the popular will free requires a firm and unchecked hand.

The academic literature has supplied these critics with plentiful ammunition. Books with ominous titles like Against Democracy, Democracy: The God that Failed, and Democracy Kills have found audiences well beyond university campuses. The charges echo those we find all the way back to Plato: democracies are short-sighted, paralyzed by special interests, prone to electing demagogues, and incapable of making the sacrifices that the challenges of the day require.

However, a large body of studies of how democracies and autocracies actually perform across regions, over centuries, and in domains ranging from economic growth to military effectiveness to environmental protection have questioned this story. They do not show autocracies to be superior—on the contrary, the autocratic temptation is, in most domains, a mirage, or even a trap. Not only are democracies morally preferable because they recognize the political equity and dignity of citizens; they also tend to work better.

Countries that successfully consolidate free and fair elections face substantially lower risks of civil war and domestic political violence than authoritarian states. Democracy, by design, offers losers something: the prospect of future victory at the polls, legal channels for dissent, protection from persecution. Citizens who can kick out the opposition at elections are less inclined to take to the streets with weapons.

The military dimension is also revealing. At least since the ancient Greek historian Thucydides, who admired oligarchic Sparta’s discipline over democratic Athens, democracies have been accused of weakness in warfare. This charge resurfaced in the 1930s, when authoritarian Germany and Japan ran roughshod over unprepared democracies, and again during the first three decades of the Cold War, when the Soviet Union seemed to outmaneuver its wealthier Western rivals. Yet the long-term record is unambiguous: since 1815, democracies have won more than 80 percent of the wars they have fought. They win not in spite of their openness but because of it. Elected governments have more legitimacy; they are better placed to ask citizens to make hard sacrifices. Democratic alliances, rooted in shared values, are more durable and credible than autocratic alliances. And no two modern democracies have ever gone to war with each other—a remarkable empirical regularity, an iron law, that has held for two centuries.

Vladimir Putin’s catastrophic miscalculation in Ukraine illustrates this. Putin appears to have made a bet that a divided, energy-dependent West would fracture under pressure; that Ukrainian society, lacking the coercive power and will of a dictatorship headed by a strongman, would collapse quickly; and that his own military was far stronger than it proved to be when put to the test. Meanwhile, Ukraine’s citizen-soldiers have fought with a tenacity and creativity that has confounded nearly every prior assessment, and support from other democracies has helped them make their brave stand against the aggressor. An autocrat’s epistemic isolation, it turns out, is not a strategic asset—it is a liability.

An even more seductive version of autocratic temptation is economical. Authoritarian leaders, the argument goes, are unencumbered by electoral cycles, and thus can make investments that democracies are unable to but which are necessary for development. They can suppress wage demands, override property rights when needed, and redirect capital toward the sectors that create growth. Democratic politicians, by contrast, must constantly placate voters and campaign donors, skewing policy toward immediate gratification.

There is something to this argument—but much less than its proponents claim, and the authoritarian alternative is hardly perfect in its economic results. Democratic accountability, for all the short-termism it supposedly encourages, creates powerful incentives to invest in public health, education, and physical infrastructure. Citizens who benefit from these things tend to reward the governments that provide them. Democratic institutions protect property rights in a way that encourages the private investment that drives productivity. And the open circulation of ideas across universities, a free press, and competitive markets is not a distraction from growth but one of its primary engines. Studies show that, on average, democracies enjoy a modest but robust long-run growth advantage over autocracies, and that this advantage strengthens with the quality and longevity of democratic institutions.

More telling than average growth rates, however, is the frequency with which disasters strike. Unchecked political authority not merely fails to deliver growth; rather, it periodically produces catastrophes. Mao’s Great Leap Forward between 1958 and 1962 killed tens of millions through an entirely man-made famine, a consequence of ideological fantasy insulated from the real world. The Soviet collectivization campaign produced similar horrors two decades earlier. Comparable disasters in democratic states are virtually unknown—not necessarily because democratic leaders are wiser or more virtuous, but because they face institutional constraints and public scrutiny that make disastrous policies impossible to sustain.

The most advanced economies in the world are democracies. The handful of countries that have joined the ranks of wealthy, high-technology societies over the past century, including South Korea, Taiwan, Israel, and Ireland, made at least the final leap under democratic governance. Singapore is the sole exception to this rule. Autocratic regimes can mobilize resources to achieve middle-income status, as China has done. But the transition to a knowledge-based economy requires the rule of law, the protection of intellectual property, and the freedom to challenge received wisdom—all of which are systematically undermined under dictatorship.

Climate change has given the autocratic temptation a new lease on life. The argument is compelling: effective climate policy requires incurring costs now for the benefit of future generations who cannot vote, overriding vested interests that are well-organized and politically powerful, and maintaining consistent policy over periods that exceed any electoral cycle. Addressing climate change might therefore require setting democracy aside. Others, meanwhile, have proposed nurturing a class of enlightened eco-authoritarian technocrats.

However, this appeal to authoritarian governance relies on a fundamental confusion between the capacity to decide on policies and the likelihood of adopting and implementing good ones. Authoritarian regimes can certainly move fast when they decide to. But choosing to address climate change rather than subsidizing fossil fuels is precisely what they frequently fail to do. They also regularly lie about emissions.

Freedom of expression is not in fact an obstacle to climate action; it is what makes it possible. Green movements need to organize, publicize, and campaign. Independent scientists must have the freedom to publish findings that inconvenience powerful interests. International environmental agreements require governments that can credibly commit. Overall, democracies, especially high-quality ones with a capable state apparatus, outperform their authoritarian counterparts, whether on greenhouse gas emissions, deforestation, water pollution, or soil erosion.

Persuasion events

The eco-authoritarian chimera also sidesteps the tough but unavoidable question of how the virtuous technocrats are to be selected and constrained—the classical problem of who will guard the guardians, which no eco-authoritarian has satisfactorily answered. Concentrated power that happens to be used wisely for climate issues can easily be turned toward repression, ethnic persecution, or foreign-policy adventurism. Absent democratic accountability, there is no mechanism for ensuring it remains focused on the planet’s welfare without neglecting or undermining other things we cherish.

China is the principal exhibit in every contemporary case for autocratic governance. China’s growth over the past four decades has been spectacular. It has lifted hundreds of millions out of poverty, building infrastructure at a pace democratic countries are unable to match. To many observers, China is decisive proof that the authoritarian model outperforms.

Look at the record more carefully, however, and a different picture emerges. To begin with, China’s official growth figures are among the most manipulated of any major economy. Satellite-based measures of night-time light emissions, which correlate closely with actual economic activity, indicate that Beijing’s statistics overstate both the pace and the scale of growth.

Context also matters. In 1949, China, Taiwan, South Korea, and North Korea were all equally poor. Today, South Korea and Taiwan are consolidated democracies with per capita incomes substantially higher than China’s. North Korea, on the other hand, offers the unvarnished control case. Communist one-party rule without economic reform has produced one of the most disastrous economic and humanitarian records of the modern era. China’s own pre-reform decades under Mao were characterized not by rapid development but by stagnation punctuated by ideologically driven disasters.

China’s environmental record, meanwhile, questions the case that autocracy handles long-term challenges better. Decades of growth have left cities choked with smog, rivers contaminated across large regions, and groundwater depleted in ways that will come back to haunt future generations. China’s domestic security spending now exceeds what the country spends on its military—a striking indicator of the regime’s lack of trust in its own population.

The handling of Covid-19 is also instructive. China’s initial containment, characterized by rapid hospital construction, mass testing, and strict lockdowns, impressed many Western observers. What is less often noted is that China’s initial response also included the active suppression of early warnings and the punishment of those who tried to alert the public, delays that cost not only China but the whole world dearly. And when Beijing eventually abandoned its zero-Covid policies in early 2023, almost a year after other countries had lifted most of their measures, it did so abruptly and without preparation, exposing a population with minimal acquired immunity to a wave of mass infection. The government had optimized for control rather than for public health, and when control became untenable, it had no alternative to letting the virus spread unchecked. South Korea and Taiwan, facing similar epidemiological conditions as democracies, managed to handle the crisis without the brutality or the catastrophic crescendo.

Taken together, China’s rise not only reflects a historically specific combination of starting conditions, reform choices, and global economic integration that is not easily replicable; it has also generated huge costs, in inequality, in environmental damage, and in political repression, that official accounts systematically obscure.

Democracies have real and well-documented failings. They are often slow. They are susceptible to capture by organized interests. They sometimes elect leaders who undermine the very institutions that gave them power. None of this should be ignored. But the relevant comparison is not between actually existing democracies and some imagined model of virtuous authoritarian efficiency. It is between actually existing democracies and actually existing autocracies. By that yardstick, the democratic record is substantially better.

Niccolò Machiavelli is not the first thinker who comes to mind when defending democratic governance. But the Florentine cynic recognized something like this five centuries ago. Contrasting republics with principalities, Machiavelli concluded that “fewer errors will be seen in the people than in the prince—and those lesser and having greater remedies.” The mechanism he identified was not virtue but accountability—the capacity of a political system to detect its mistakes and correct them.

That capacity is precisely what democratic institutions are designed to provide. At a time when open societies face serious pressure from within and without, the temptation to admire their alternatives is understandable. But admiration is not a sound foundation for political judgment, especially not when it is based on a selective reading of the evidence. The autocratic temptation promises fortitude and efficiency—but too often, it only produces chaos and mismanagement; and, occasionally, it delivers disaster.

Jørgen Møller and Svend-Erik Skaaning are both professors of political science at Aarhus University.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

Last week, the United Kingdom unveiled a new policy it claims will protect children on social media platforms like TikTok, YouTube, Facebook, and X. The exact rules governing the ban will be announced before its implementation next year, but we know it will mandate age verification to restrict access to social media platforms for children under the age of 16. The policy will also affect older teens—16- and 17-year-olds will face a curfew preventing them from accessing social media overnight, while younger teens will be unable to access live streaming.

The proposed law is part of a wider trend. Last year, Australia passed a landmark social media ban for under 16s, which was lauded for setting the tone globally for other countries to follow. The campaign to keep kids off social media has also hit Europe, Asia, and the United States.

Supporters of such bans argue that they are necessary to protect children from the harms of social media. Governments often claim noble reasons when putting limits on citizens’ ability to speak freely and anonymously, and nothing sounds more noble than “protect the children.” “This is a choice about whose side we’re on: families across the country, or a status quo that isn’t working,” Prime Minister Keir Starmer said of the new policy.

Subscribe now

But there are many reasons to oppose social media bans. First, supporters assume these bans do what they’re supposed to. But Australia’s ban emphatically isn’t doing what it promised. The government’s own research found that among parents with kids on social media before last December, about 7 in 10 said their child still had a Facebook, Instagram, Snapchat, or TikTok account. Starmer admits the Australian model hasn’t delivered. He has stated that the UK will “learn the lessons from Australia’s experience” and “make it far harder for children to bypass safeguards.” That’s why Starmer is calling his version “Australia-plus.” Soon, the UK government will release “different options for effective forms of age assurance for proving whether someone is over 16.”

But there’s a problem: these “under-16” policies will blunt all UK citizens’ ability to speak freely and anonymously online. All users, whether they’re 15 years old or 55, will need to prove their age in order to post on social media. This raises significant privacy concerns, not just for users’ ages, but for their identities. The process of age verification will cut away vital firewalls around users’ anonymity. As the rush to ban children from social media hits a global fever pitch, it’s well past time we abandon the faulty claim that only children will feel its effects.

In passing a ban, the UK will join a wave of countries pushing against anonymity. Earlier this year, German Chancellor Friedrich Merz said he wants “to see real names on the internet.” The Turkish government meanwhile announced its plan to end online anonymity. If you’re reading this piece in a nation without age-gating proposals or an associated threat to anonymity, be warned: It’s only a matter of time.

One problem facing advocates of internet restrictions is the availability of virtual private networks (VPNs), which reroute traffic and allow users to access banned content or sites from behind firewalls or blocks. The UK government is well aware of the challenge VPNs may pose to its under-16 ban, and Technology Secretary Liz Kendall announced this week that the government “will make further statements in July about VPNs.” Children’s Minister Josh MacAlister has said there are “options there about whether we could age-gate VPN use, which would be really welcome.”

But attacking VPN use is a hallmark of authoritarian internet censorship. In deeply censored societies, VPNs are often the only way people can access vital information about current events, history, and information about their health and safety. Speech and privacy advocates should watch closely the lengths to which UK officials are willing to go in enforcing their under-16 ban, especially if it involves restricting access to VPNs.

Parents and policymakers around the world have raised reasonable concerns about the role social media use plays in young peoples’ lives and development. The internet is a revolutionary technology that is always evolving, and we are still in the process of understanding its implications for kids and broader society.

But the answer is not blanket censorship. Parents across the board may have different ideas about how they want to manage their own children’s internet use and their exposure to online speech. And it’s deeply troubling that few policy makers have discussed how their “child safety” rules burden adults. Put simply, many unsettled questions and serious objections to social media bans have been cast aside in the rush to enforce a near-uniform policy approach on a mass scale.

It’s not particularly surprising that the UK is once again treading censorial ground. In recent years, Britain has banned controversial speakers from entering the country, targeted blasphemy, investigated and arrested people for peaceful protest, and rolled out messy regulations targeting so-called “harmful” online content. It has been shockingly comfortable with targeting social media users, with thousands annually facing police action for posts perceived as offensive or harmful. And that’s just a sample of the growing power the UK is asserting to define what its citizens can see or say—and the tools the government has at its disposal to enforce that power.

The truth is that we don’t need to pit our speech rights against youth safety. We can work to help parents individually make the right choices for their families. There are ways to empower parents to navigate their kids’ use of the internet while respecting that children have their own ability to speak and seek information, and protecting the rights of the entire population against burdensome and invasive limitations on their expressive rights.

If Britain doesn’t learn this lesson fast, it will find itself sleepwalking into an unfree internet.

Sarah Mclaughlin is Senior Scholar of Global Expression at the Foundation for Individual Rights and Expression and author of Authoritarians in the Academy: How the Internationalization of Higher Education and Borderless Censorship Threaten Free Speech.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

When University of California President Clark Kerr delivered the Godkin Lectures at Harvard in 1963, published shortly thereafter as The Uses of the University, he was doing something unusual for an academic administrator: he was offering a sophisticated social theory, and doing so with wit. In these lectures, Kerr coined the term “multiversity” to describe what the postwar American research university had become. In Kerr’s account, the modern university was no longer to be understood as a community of scholars united by a shared ideal of learning, but rather as a sprawling institutional conglomerate serving at once as a research engine, a job-training facility, a credentialing mechanism, a coming-of-age experience, and an incubator of the national technical elite. The University of California, which Kerr had just finished steering through a near-decade of explosive growth, was his exemplar.

Kerr was a droll man. He once observed that the three great problems facing any university president were “parking for the faculty, athletics for the alumni, and sex for the students.” He described the university faculty (and I can confirm from personal experience that this remains accurate) as “a series of individual faculty entrepreneurs held together by a common grievance over parking.” And when Ronald Reagan made good on his 1966 gubernatorial campaign promise to fire him for being too lenient with the Free Speech Movement protesters, Kerr offered one of the great farewell lines in American academic history: “I leave the University of California as I arrived: fired with enthusiasm!”

Despite the jokes, Kerr was a serious man. The argument underneath The Uses of the University was that the multiversity, precisely because of its sprawl and apparent incoherence, was the institutional master key of mid-century American civilization. It was the nexus at which basic scientific knowledge was produced, technical and professional talent was credentialed, democratic citizenship was cultivated, and the national project of technological supremacy was advanced. The multiversity didn’t need to be coherent in order to be functionally useful as a platform for what Kerr called “administering the present.” He wrote with the high modernist confidence of someone who believed that hierarchical technocratic institutions, if competently managed, could keep these various volatile elements in balance.

Kerr’s dismissal by Reagan in 1967 was, in a sense, the first indicator and warning of the crisis of the high modernist technocratic model that he championed and sought to institutionalize through the multiversity.

It is time to acknowledge that Kerr’s model of higher education is finished: long on its last legs, the arrival of AI announces its death-knell. What comes next is disaggregation: the multiversity as we know it being disassembled into its component parts. This need not, however, be a catastrophe for higher education. Actually, in many ways, it represents an opportunity to return to roots, in a classical model of education and in attentive pedagogical instruction. But higher education can only weather this period of disruption if it is clear-eyed about what is happening and moves confidently toward a new model.

The Crisis of the University Is Not New

The present crisis of the American university began building already sixty years ago as the postwar bargain that Kerr’s vision embodied started to fray. What followed was a slow-motion privatization of university finances, producing a slow-motion tuition hyperinflation that has burdened a generation of students with debt while hollowing out the public mission of the university. The shift from grants to loans, from tenured faculty to mass adjunctification, and from a broad education in the liberal arts to vocational credentialism all occurred under the banner of making universities more “responsive to market demands.” In practice, this has meant transferring cost from the public to the individual “student consumer,” while defunding the parts of the institution that didn’t produce monetizable outputs. The net result has been the ever-upward-spiraling costs of undergraduate education, without a corresponding increase in the value of educational training or credentialling, and a loss of political support for the mission of universities. These financial and political travails have heightened the contradictions between the disparate missions and functions of the multiversity.

Into this increasingly unstable compound, add AI.

The arrival of large language models is acting as a catalytic solvent, titrating out the incoherence that was always there. When a student can produce a plausible term paper in twenty minutes using Claude Opus or Google Gemini, what is the point of assigning term papers? When an AI tutor can explain any concept at any level of sophistication with infinite patience, what is the value of a lecturer reading from notes? When AI can ace most standardized professional examinations, what is a credential certifying? These are old problems that AI has made it impossible to ignore.

Beyond the pedagogic challenges posed by the arrival of LLMs, AI is also exposing that the Kerrian bundle held together for as long as it did because its components shared a common and venerable set of technologies of knowledge transmission: the book, the lecture, the problem set, the written examination. In a pre-LLM world, these formats made cognitive demands of students that were difficult to simulate or shortcut. That is no longer true. AI doesn’t just automate some of the tasks associated with these formats; it renders the formats themselves obsolete as instruments of either intellectual discipline or assessment. And when the shared technological substrate dissolves, the contradictions built into the multiversity from the beginning become impossible to paper over. The world-class research mission and the undergraduate teaching mission have always been in tension. The prestige economy that rewards publications over pedagogy always distorted faculty incentives. The credentialing function was always only loosely connected to the educational one. These were the open secrets of the American research university. In a post-AI world, these divergences are being rendered untenable.

It is striking, then, that the most widely discussed recent attempt at university self-examination, the April 2026 Report of the Yale Committee on Trust in Higher Education, barely registered any of this reality. The report was in some ways an admirable document. It was clear-eyed about costs, scathing about admissions opacity, and candid about the political monoculture that has eroded public trust across partisan lines. Yet its treatment of AI was cursory to the point of negligence: a few sentences in the section on the classroom, expressing uncertainty about AI’s effects and noting that faculty are “scrambling to redesign syllabi.” It is remarkable that a report tasked with understanding why public trust in higher education is collapsing would fail to reckon with the technology that is restructuring the economics and logic of knowledge work. It suggests that even the most self-aware corners of the academy are still treating AI as a pedagogical inconvenience (or literal cheat-code) rather than what it actually is: the force that is making the entire inherited architecture of the multiversity impossible to sustain.

The Co-curricular Dodge

So how should the university respond to this crisis of purpose, identity, and even faith? The most popular present answer in certain administrative circles to this question is an emphasis on the “co-curricular,” that is, on residential life and human connection as the university’s irreducible value in an age of AI tutors. Perhaps the most cited proposal is Molly Worthen’s New York Times piece from three years ago, “Why Universities Should Be More Like Monasteries,” which argued that universities should offer radically low-tech, high-presence educational environments.

This argument isn’t meritless: there is evidence that learning works differently when embedded in community, that chance hallway encounters with faculty members, late-night bull sessions in the dormitory common room, and heated dining hall debates are often the most generative moments of learning. Students’ own accounts of what matters most in college consistently center on relationships, belonging, and dialogue. The argument for residential education, for the ancient model of the Platonic Academy as gymnasium and garden as much as classroom, is stronger now than it has been in decades.

This is continuous with a long-standing function of universities as sites for passage from childhood to adulthood, for coming to a new understanding of oneself. In the 1960s more than four fifths of college freshmen reported that a major goal of college was to help themselves “develop a meaningful philosophy of life,” a number which collapsed by half in the 1970s and 1980s. A reemphasis on the co-curricular could help revivify this ideal, which would in turn help prepare students for the AI-forward world they are entering. As Anthropic cofounder Jack Clark recently argued, the people who will most benefit from AI are those who have first built deep, idiosyncratic human capacities through “repetitive practice and creation.” The machines will work best when helping you to amplify what you’ve already made of yourself.

But by itself, the co-curricular is an evasion. It leaves untouched the question that determines what students and families are paying for: what happens in the curriculum, in the classroom, in the formal educational encounter. That is where reform needs to be most radical, and where the response of universities so far has been most quavering. If the primary response of universities to the most dramatic new knowledge technology in decades, one that employers everywhere are expecting employees everywhere to use, is to demand that students stick cotton in their ears and keep rowing, it will only hasten their decline into institutional redundancy, if not irrelevance.

Cognitive Requirements in the Age of AI

Any reimagining of the university in the age of AI must begin with an honest reckoning with what AI cannot do—and what therefore becomes relatively valuable precisely because AI can do everything else. The key distinction is between work that AI does well (such as synthesis of known patterns, argument elaboration, template instantiation, and generating local coherence) and work it structurally cannot do because of the architecture of the technology as such. AI cannot build the trust on which institutional cooperation depends, because trust is not a conclusion reached by processing information about another agent but instead is a relationship constituted over time between persons who have staked something on each other, and who can be betrayed. AI cannot give a person good taste or style, because taste and style are about personal distinctiveness within a community which shares an aesthetic. AI cannot constitute goals, because that act requires a valuing subject. These are not gaps that more compute will close. They are absences that follow from the ontology of the technology itself.

A curriculum designed around AI’s limitations should be seen neither as an exercise in nostalgia nor as a denial of the burgeoning power of these systems. In fact, given the trajectory of AI capabilities, it is the only curriculum with any hope of finding a stable foundation.

What does this mean in practice? Start with the most obvious casualty: the term paper, as an assessment instrument, is dead. Written homework assignments were meant to push (and test) a student’s ability to produce a well-structured, coherently argued text. But this is exactly what LLMs do effortlessly and without demanding of the user any of the underlying cognitive work for which the traditional term paper was supposed to be a proxy. This included sustained argumentative reason: the ability to construct and maintain a complex argument across an extended piece of discourse, distinguishing claims from evidence, handling counterarguments, and reaching a defensible conclusion. Written assignments also demanded epistemic self-regulation, that is, the metacognitive capacity to monitor one’s own understanding, recognize gaps in evidence, revise positions in response to what the evidence shows rather than what one hoped to find. This pedagogically valuable work always operated below the waterline of the actual output of a term paper; what LLMs do is deliver results that simulate these actions without putting the students through their cognitive paces.

The replacement, as many education researchers are arguing, is live assessment and demonstration: real-time diagnosis of novel situations, design critique, structured adversarial debate, and Socratic examination. These formats test the ability to sense-make under pressure, defend a frame against live challenge, revise a model when evidence contradicts rather than confirms it, and recognize when uncertainty is too high to proceed. In practical terms: collaborative student projects will require documented decision logs tracing reasoning behind commitments, the canonical deliverable shifts from polished artifact to demonstrated live reasoning, and oral examinations and hand-written exams will become the primary assessment instruments. But despite this emerging consensus among education researchers, institutional practice has barely moved.

If the post-AI university’s pedagogic value proposition is the formation of cognitive capacity in conditions that cannot be replicated on a screen, then the function and responsibilities of faculty members must also be reconceived. It clearly no longer makes sense for professors to stand in front of a hall full (or, too often, only half full) of students delivering lectures. As a mechanism of information conveyance, AI can now provide the same at near-zero cost, tailor-made to the specific knowledge gaps of individual students. Instead, professors must reconceive of themselves as interlocutors, serving as performative models of how to calibrate uncertainty and revise frames in real time. The classroom experience should focus on helping students to understand how to constitute a goal rather than generate a text in response to a prompt provided by the professor.

This is something closer to the Oxbridge tutorial system, the clinical ward round, or the seminars of many small liberal arts colleges in the United States. These pedagogies were once defended on grounds of tradition or prestige. The post-AI argument is structural: they are the delivery mechanisms for exactly the cognitive capacities that the architecture of AI cannot replicate, because those capacities are developed only by being exercised, not described. Interestingly, this means that the coming of AI is going to mean there will be demand for more professors, rather than fewer.

None of this implies that faculty should pretend AI does not exist, or that the tutorial and seminar should be conducted in proud ignorance of a tool students will be spending the rest of their professional lives using. The opposite is true. Faculty should integrate LLMs directly and deliberately into their instruction as tools that need to be used correctly in order to not be harmful. (The analogy of a blowtorch or a chainsaw comes to mind: these are useful tools, but you need to learn how to use them safely.) Teaching a student to prompt effectively is teaching them to think precisely about what they want to know and why; it is, in this sense, an exercise in goal constitution. Teaching students to evaluate an LLM’s output critically by scrutinizing the machine’s often over-confident syntheses against evidentiary standards defined by the phenomenological reality of the external and material world is teaching them epistemic provenance tracking and calibrated self-assessment. LLMs can also become objects of critical study in their own right: students should be asked to assess why the model did not produce exactly what they had a priori in mind when they initiated the interaction. Handled this way, LLMs can serve as clarifying instruments in the pursuit of the classical objectives of enlightened education: the inculcation of critical thinking and logical reasoning, rhetorical and communicative competence, aesthetic appreciation and the cultivation of taste, moral and ethical reasoning, and ultimately the ideal of self-knowledge and Bildung.

This brings us to the content of the curriculum itself. As I recently argued in Noema, if the goal of a college curriculum is (as it should be) to inculcate oral reasoning and persuasion, ethical analysis and moral judgment, historical and comparative thinking, and the cultivation of taste and discrimination, then we are precisely in the domain of the classical curriculum of the liberal arts. Skills such as goal constitution, situated judgment, and value alignment are exactly the capacities that a serious engagement with history, philosophy, literature, and political theory develops. History trains temporal imagination and frame revision; philosophy trains epistemic precision and the discipline of distinguishing solid argument from vapid sophistry; literature sharpens an appreciation for style and a feeling for hidden meaning; political theory trains the recognition of suppressed goal contestation and the conditions for legitimate alignment. Together they enable students to imagine lives unlike their own, a hugely valuable experience in a world changing as fast as ours.

How to convey the content of these disciplines to students is going to have to change dramatically from the homogenous one-to-many mass-delivery model of the postwar multiversity, but the content is perfectly classical. The university’s present crisis of purpose is, in this light, at least in part a crisis of having abandoned its own best tradition in pursuit of vocational or technical training that AI is now rendering obsolete.

But Does It Scale?

The central challenge for universities will be how to move toward this model at scale. The tutorial and seminar model is labor-intensive by design: a professor working as interlocutor rather than lecturer can engage only a fraction of the students she could previously reach from a podium. The skills required of faculty will also need to change substantially. Under the old model, a brilliant researcher delivered expected value simply by speaking one-to-many; the new model requires someone with the pedagogic sensitivity to calibrate each student’s specific confusions and capacities—qualities that research prowess neither produces nor rewards. Elite universities in particular have built their faculties almost entirely around research achievement, with teaching treated as a secondary obligation. Reconceiving the professoriate will mean altering tenure criteria and promotion incentives, and it will face fierce resistance from scholars whose professional identities are bound up in the research function. None of this is impossible, but none of it will be easy. No doubt some tenured faculty will pour boulders and boiling oil down the side of their ivory towers to prevent these changes from taking place.

Persuasion events

Longer term, however, we should expect the disruption caused by AI to be not just pedagogical but to the structure of the university as such. Kerr’s great insight was that the multiversity’s incoherence was not a bug but a feature—that a loosely bundled institution mirrored a loosely bundled society by providing something for everyone, from the Nobel laureate to the newbie grad student, from the NIH grant-seeker to the remedial English student. What held those disparate functions together was a social infrastructure of knowledge transmission: the laboratory, the lecture hall, the examination, the credential. Once AI can provide information delivery at near-zero cost there is no longer a compelling reason why research, teaching, and credentialing need be co-located in the same institution. What will replace the multiversity is likely to be not one thing but several: research centers that focus exclusively on the new-knowledge-production business; independent communal residence facilities that know they are in the coming-of-age business; and teaching systems that are honest about what skills they are inculcating. Even credentials from the most exclusive universities may not retain much social signaling value.

Clark Kerr would have recognized this moment. He was no naïf about the multiversity’s contradictions; but he also believed that competent management could hold them in productive tension. What he did not foresee was that the tension would be dissolved not by political upheaval—as it nearly was in 1964, when the student movement that eventually got him fired also signaled the coming fracture of the postwar liberal-technocratic consensus—but by technological rupture. The irony is that the research university, which Kerr celebrated as the engine of American technopolitical supremacy, incubated the very instrument that is now rendering untenable the research university’s inherited form.

What the students who booed the mention of AI at recent commencement ceremonies this spring were registering, in the way that students have always registered institutional failures, is that they were not getting what they came for. But as with more than one student movement before them, just because they rightly identified a structural problem doesn’t mean that they have particularly good ideas about what a better institution would look like. Just as Kerr recast the University of California to match the liberal-technocratic imperatives of the postwar period, so do visionary college leaders today have an opportunity to remake the university to match the requirements of an economy that will be redefined by AI. Achieving this will be a generational project.

Nils Gilman is Senior Advisor to the Berggruen Institute and former Associate Chancellor of UC Berkeley.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

This article is brought to you by American Purpose, the magazine and community founded by Francis Fukuyama in 2020, which is proudly part of the Persuasion family.

There are serious countries and unserious ones. Where the United States belongs these days I leave for others to judge. But about one country I have no doubt, having recently returned from it. It may be tiny and poor, but Moldova is a serious country—and its leaders know exactly what they want.

I visited Chișinău last year, several months before Moldova’s parliamentary election, when its outcome—and with it the country’s European trajectory—hung in the balance. This time, however, the mood was different. While meeting President Maia Sandu, the prime minister, the foreign minister, members of parliament, and journalists, my American Enterprise Institute colleagues and I were struck by how uniformly serious they all were about Moldova and its place in Europe.

Anne Applebaum recently called the September result—the victory of Sandu’s party despite a deluge of Russian money, vote-buying, cyberattacks, and bomb threats—“the Moldovan surprise.” And yet, while the absolute parliamentary majority that her faction enjoys in Parliament was unexpected, I remember a number of well-placed Moldovan officials who assured me a year ago that they had a good handle on Russia’s interference in the fall election.

When she met with us, Sandu wore white trainers and a smart glenplaid suit that would have looked less conspicuous at the World Bank—where she once worked—than at the fairly monstrous Soviet-era presidential palace. But Sandu is no technocrat. She is a skilled politician who built a party from scratch, without oligarchs or networks of patronage, and stayed the course with steely determination through numerous setbacks.

Mihai Popșoi, the foreign minister, who wrote policy reports for AEI a decade ago, has yet to turn 40. His command of his brief is impressive, as is his ability to think on his feet—especially when my colleagues and I asked difficult questions about Moldova’s relationship with the United States. In a way, it mirrors the deftness with which Moldova’s ambassador to Washington avoids the traps that the Trump era presents for him—neither antagonizing Donald Trump nor cozying up to him, always working scrupulously across party lines. In an era when so many European governments lurch between sycophancy and tantrum, Chișinău’s poise is a quiet masterclass.

The prime minister, a former financier with a Columbia degree and a U.S. passport, speaks the same language. One can also see Moldova’s seriousness clearly in the unglamorous business of energy policy, which the PM is visibly passionate about. For years, Moldova was hostage to a power plant running on subsidized Russian gas in the breakaway region of Transnistria—a lever that Moscow could pull at will. That has now changed. Moldova has stopped buying Russian natural gas completely. A new 157-kilometer high-voltage line from Vulcănești to Chișinău—aptly nicknamed the “Independence Line”—connects the country directly to Romania and the European grid and will cover more than half of peak demand.

Moldova has EU candidate status, its accession negotiations are underway, and it is aiming—ambitiously—to conclude them in 2028. While not a completely unrealistic timeline, some European diplomats whom we saw were squeamish about the prospect of bringing Moldova in so rapidly. “It is not just a question of meeting the criteria, but also a political one,” we heard several times—both from one European ambassador (sincerely) and from several Moldovans (with a hint of sarcasm).

A related lesson that Brussels is slowly absorbing is that EU accession should not be an all-or-nothing prize handed over only at the very end of a decades-long marathon. The smarter approach is to frontload the benefits of membership: phased access to the single market, structural funds, and a seat—even a non-voting one—at the institutional table, so that citizens see the dividends of integration well before the final treaty is signed. For Moldova’s pro-European elites this is a no brainer. The country’s political life is fractious, and progress made today could be reversed. In order to anchor the country in the West for good, tangible benefits to ordinary people must first be delivered before any formal process is completed.

Yet, on the front of European integration, Moldova has both benefited from—and now risks being a victim of—the imperative of bringing Ukraine into the West’s fold. Kyiv’s case for EU membership is morally overwhelming and necessary in geopolitical terms. But it is also a difficult one. Ukraine is enormous, not to mention busy fighting a war against an intimidating adversary. Its entry would reshape the EU’s budget, its agricultural policy, and the balance of votes around the table in ways no previous enlargement has come close to. While it has leapfrogged the West in some ways (not least in defense technology), many of its post-Soviet pathologies seem to linger.

It was against this backdrop that Friedrich Merz floated, in a late-May letter to EU leaders, the idea of an interim “associate membership” for Ukraine—participation in the Union’s institutions without a vote during the long negotiations—while accelerating Moldova and the Western Balkans toward full membership. The clumsy label aside, the idea is sound: deliver what you can now, and don’t let the hardest case set the pace for everyone else.

It was unfortunate that Kyiv’s reaction was a faintly indignant refusal. President Zelenskyy insisted that Ukraine’s place in the EU must be “complete, with full rights,” and that there can be no genuine European project without it. Meanwhile, Ukrainian commentators denounced the proposed decoupling from Moldova as a red flag.

I yield to no one in my admiration for Ukraine. But treating any interim arrangement as a demotion—and full EU membership as a debt the West owes for Ukrainian blood—is unbecoming and self-defeating. An associate status, if firmly tied to eventual accession and packed with real benefits, is not a consolation prize. It is a head start that earlier candidates never enjoyed. Reflexively spurning it hands ammunition to those in Europe who would be perfectly content to leave Ukraine parked in the waiting room indefinitely.

It would be a disaster if Kyiv’s all-or-nothing insistence, colliding with the genuine complexity of its situation, froze the entire enlargement machine on political grounds—and if little Moldova got dragged down as collateral damage.

Decoupling Moldova’s accession from Ukraine is not a betrayal of Kyiv but common sense. A third of Moldovans, or 850,000 of them, hold Romanian, and therefore EU, passports. A large share of the country is already, in effect, made up of European citizens. Integrating a country the size of a mid-sized European region would not strain the EU, but would instead be an accelerator, bringing Ukraine even closer to the fold.

Yes, there is Transnistria, a seemingly intractable problem. But the true size of the Russian garrison, excluding locals with Russian passports, is tiny: some 300 officers. Since the border with Ukraine is sealed, the region is boxed into a position from which Moscow can neither reinforce nor resupply it. Moreover, the territory’s real master is not an ideologue but a businessman: Viktor Gushan of the Sheriff conglomerate, whose exports flow overwhelmingly to the EU. It must be possible to resolve this issue through negotiation.

Serious countries deserve to be treated seriously. Moldova has done the work—which is more than can be said for many of its larger, richer, and longer-established neighbors, whether or not they seek (or already enjoy) EU membership. The least Brussels can do is to reward that clarity of purpose, rather than punish them for the headaches involved in some of the harder cases.

Dalibor Rohac is a senior fellow at the American Enterprise Institute, an advisor at GLOBSEC, and a columnist with American Purpose.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

Samuel Moyn is the Kent Professor of Law and History at Yale University. His books include Humane: How the United States Abandoned Peace and Reinvented War, The Last Utopia, and Not Enough: Human Rights in an Unequal World. Cohost of the Digging a Hole podcast, he is a frequent contributor to The New York Times and many other publications.

In this week’s conversation, Yascha Mounk and Sam Moyn discuss whether a truly fair democracy might weigh different citizens’ votes differently, whether the emphasis on human rights have got us into the mess we’re in today, and to what extent our democracy is in danger from populism.

This transcript has been condensed and lightly edited for clarity.

Yascha Mounk: So why are the old people the problem, Sam?

Samuel Moyn: I don’t think old people are a problem. I think gerontocracy is—in the same way that whites may not be a problem, but white supremacy is real.

My goal is to show how older people actually sometimes suffer too under a regime of gerontocracy, and to look at how that regime came about and what its current form is—not just for the sake of younger folks, but also for the mass of older people themselves.

Mounk: All right, I’ll refrain from being flippant for the rest of the conversation. Walk us through some of the ways in which America is ruled by gerontocracy today.

Moyn: In the book I look at the first coinage of the term. It’s not Greek, but it’s based on Greek words. It’s supposed to add a dimension to our classification of forms of government like monarchy or oligarchy. The person who coined it made it very clear that it was taking a different form in modern times, having to do with the organization of the electoral system—how people have elections in a democratic society—and then a lot to do with how wealth is allocated and transmitted. I definitely talk about the politicians of our day, who in America are old men and women, but I’m mainly interested in showing that gerontocracy is systemic and that it affects what it means to go to the polls, the outcomes, how elections are financed, and how our economy is organized. I have this sense that we should think about power in the broadest sense—rather than just looking at who’s in office, as the Greeks did—and look more at who has power, meaning effective control of the lives of most people.

Mounk: Let’s go through some of these areas. As you’re saying, politics is the most straightforward, and then there are ones you think are perhaps more important or deeper. But just at a purely descriptive level, what’s the evidence that old people are ruling us in Congress, in the White House, and so on, in a way that hasn’t been the case in past eras?

Moyn: The most graphic evidence, if we’re talking about the United States, is the two most recent presidents. There had really only been one very old man in the presidency before Ronald Reagan, and that was William Henry Harrison, who died in short order of a cold in the nineteenth century. With Reagan we began to see older and older men in office, capped by the last two presidents, Joe Biden and Donald Trump. Now, in fairness, more than half of humanity is ruled by an old man over 70 at this point, largely because of the aging of Vladimir Putin, Xi Jinping, and so forth. The American syndrome is specific.

What I try to do—and this I think is tolerably well known—is go beyond the focus on Biden and cognitive decline to ask just how unrepresentative the political class generally is in the United States, looking in particular at Congress and the federal judiciary. What I’ve tried to do in the book across all these topics is provide the receipts, mainly generated by other people, and gather them in one place. Congress is aging; the Supreme Court is aging in the sense that even when appointed young, the justices stay for a long time.

Aside from cognitive decline, the risk of death is a very serious aspect of the problem. Donald Trump’s one big beautiful bill was facilitated in its passage by the death in droves of Democrats in the House in the six months before that bill. Ruth Bader Ginsburg, notorious for other reasons, died setting up the end of federal abortion rights. All of that pales in the end, if we’re talking about the political class, beside representational fairness, because cognitive decline or death are just the flagrant symptoms of a broader political class that’s old. Very few people in their thirties and forties hold any position of significance in the federal government. That’s the story when it comes to politicians.

Mounk: What do you think is the explanation here? Obviously, when it comes to the presidency, you said that’s been the trend since Ronald Reagan. Reagan was quite old when he assumed the presidency, but the three presidents afterwards were relatively young. He was followed by Bill Clinton, George W. Bush, and Barack Obama, all of whom were reasonably young men at the time they took office. Some of the dynamics in the Senate, and perhaps in the House of Representatives, have to do with very specific American institutions—though I think that’s harder as an explanation for the presidency. What I mean is that if you’re an incumbent senator, particularly in a safe state, of which there are now very many on both sides, it’s very hard to displace you. If you’re 70 or 80 years old and running for reelection, you probably wouldn’t win a primary contest when you’re first being elected, but with 25 years of service and all the favors owed to you and all the fundraising muscle at your disposal, you may be able to hold off a primary challenge—and then you’re going to sail to reelection in Texas or Massachusetts, if you’re in the right political party.

The presidency is harder to explain. In 2020, Joe Biden faced a lot of younger rivals for the Democratic nomination, and yet voters in the primaries freely chose him. In 2024, Donald Trump faced a number of opponents who were quite a bit younger than him, but he was clearly able to retain the loyalty of the Republican primary electorate to such an extent as to gain reelection. I haven’t seen analyses showing a very obvious age gradient in which primary voters voted for those candidates. In 2024, for example, it’s not clear to me that the people who voted for Trump in Republican primaries were systematically older than the people who voted for his rivals. Even in the general election itself, there was less of an age difference than we might have expected, and most of that is produced by partisanship rather than a preference for a particular age. So what do you think produces this? If this is the free choice of American voters, I’m free to disagree with that choice—and I think there’s good reason to disagree in both of those cases—but it’s somewhat unclear that we shouldn’t just leave it to American voters to make their own decisions.

Moyn: One reason I’m interested in pushing beyond politicians in general, and not just the president, is that I agree there’s a lot of contingency and idiosyncrasy in the political story—and yet it’s the one that, to the extent Americans care about this politics, they care about. So I’m using it as a door.

I agree with you, and I’d even add that if Donald Trump were to fulfill the fondest hopes of those who’ve resisted him and depart or die within a year, we’d immediately have the youngest president in U.S. history, because J.D. Vance would enter the office at an age younger than Theodore Roosevelt, who currently holds the record for youngest president.

Mounk: When you look at some of the leading Democratic contenders for the nomination in ’28, they go from middle-aged men—like Gavin Newsom—to very young contenders like AOC.

Moyn: Absolutely. The foundations of what I’m thinking of as the structure of gerontocracy—as a structural phenomenon that goes beyond these contingencies and therefore leads us beyond the level of politicians—the main driver is going to be the aging of humanity, and certainly of Americans. I would push back a bit on the moral you seem to hypothesize when it turns out that there’s a lot of contingency in accounting for the age of politicians.

We hope you’re enjoying the podcast! If you’re a paying subscriber, you can set up the premium feed on your favorite podcast app at writing.yaschamounk.com/listen. This will give you ad-free access to the full conversation, plus all full episodes and bonus episodes we have in the works! If you aren’t, you can set up the free, limited version of the feed—or, better still, support the podcast by becoming a subscriber today!

Subscribe now

Set Up Podcast

If you have any questions or issues setting up the full podcast feed on a third-party app, please email leonora.barclay@persuasion.community

Voters only select and express their preferences within a system, and we have to critically inquire into that system. Let’s continue to take my analogy of white supremacy as something many people would acknowledge exists and may be systemic in the way I’m saying gerontocracy is. If you said that, after the Supreme Court’s decision in Callais, no Black politicians are being elected in the Deep South anymore—is that the will of the people, or is it because of the way the will of the people has been organized in an electoral system?

As far as I see it, it really matters that there are advantages of incumbency—which you mentioned—which means that especially when politicians in America are starting their careers later, you’re just going to have an older set of politicians than the electorate and certainly more than the population. Then we’d have to get into campaign finance, where the facts are glaring about just how old the median campaign donor is: over 60, sometimes higher. I devote a whole portion of the book to thinking about gerontocracy at the political level, not just in terms of the politicians, but in terms of the institutions. We do literally have a branch of government named after old men. We do have age minima in the Constitution for federal political office, with no age maxima. Those are also features of the way we’ve organized political choice. We can’t say “the will of the people” as if that’s pre-existing, because it utterly depends for its expression on the institutions we’ve devised and the rules we’re following to figure out what it is.

Mounk: Perhaps we’ll get into a little bit later what we should do about those institutions, what kinds of ways to reform them. Certainly when it comes to something like abolishing age minima, I think there’s a very strong democratic and liberal reason to abolish them. Some of the suggestions you make in the book are rather more ambitious, and we may have different views about them, but we can get into that.

Let’s get beyond politics. As somebody who is an immigrant to the United States having grown up in Europe, I’ve always been struck by how open American society is to the talents and contributions of young people. I feel that a little bit less now that people in my age group in Germany and France and so on are coming to have genuine roles of importance and responsibility in society. But I was struck when I was 30 years old that I had many friends and acquaintances in the United States who already had positions teaching at major universities, positions editing publications or at least important sections of publications, and very significant roles in law firms and investment banks. Obviously, when you look today at Silicon Valley, you have an enormous number of people in their thirties who have about as much power and influence as any human being has had in the history of humanity by certain standards.

This stood in contrast to a lot of 30-year-olds in Germany still being interns, to somebody getting to be the editor of a newspaper section of a major publication at the age of 30 being an incredibly exceptional achievement. When I worked in theater after college for a year, my first task as assistant director was to hire an intern—I was 21 years old, and the intern we ended up hiring was 29. So to what extent is it true, when you look at American society more broadly, that there is this kind of gerontocracy? Isn’t America still in many ways the country of the young, where people have enormous opportunities if they have the right kind of education and perhaps the right kind of background and the various advantages that play into that in a system that is not always as meritocratic as it claims to be? Isn’t it actually striking how much influence and power and money a lot of young people have in the United States compared to anywhere else in the world?

Moyn: Oscar Wilde famously said that the youth of America is its only tradition. I don’t think that’s true—I see the U.S. Constitution as a gerontocratic document in the ways we’ve already addressed, and that’s a big tradition—but I agree with you. It’s one reason I’m interested in thinking about the American case, that at times America has stood for launching youth and providing careers open to talent. I think you might overstate the transatlantic gulf, insofar as I think the whole point of modernity was to topple elder rule and put in place opportunities for young people. The whole modern novel and its roots in the so-called Bildungsroman is about the opportunities that young men could enjoy after the revolution, on the model of Napoleon.

The question is what’s happening now. The demography, I think, is unsuitable for the picture of America you’re saying still prevails. I’m not denying it: I became an Ivy League professor at twenty-nine and got tenure shortly after, so I have had enormous opportunities as a youth. Many of my students have not. Some have been the editors of those publications you mentioned, but in part because they were blocked from becoming professors in a way I wasn’t. The general story has to be that our demography is changing, with fewer young people and an overwhelming number—absolutely and relatively—of aging people who are not suffering decline and are not forced out of their power. The baseline is changing, and that’s the very modest claim I’m making.

Mounk: I wonder whether we need to distinguish between areas of American life and the American economy where you’re not just distributing a fixed number of positions of influence and opportunity, and ones in which you are. For a variety of reasons, I think universities—and particularly humanities departments within universities, and particularly humanities departments studying the sorts of things that you and I studied when we were applying for faculty jobs—are just at the extreme end of that distribution. Universities have not increased their numbers in part because it is an oligopoly that is trying to protect its prestige. The faculty of Columbia University recently voted by something like 80 or 90 percent against admitting more students to the opportunities that university may give to people. The number of students studying the humanities has crashed over the last 20 or 30 years for a whole set of reasons within the humanities. Political theory and European intellectual history—your discipline—are very much out of fashion. In those very specific areas of American life, there is indeed a highly limited number of positions and opportunities, and if professors no longer have to retire and some of them choose to stay in post until they’re 75 or 80 or 85, that means young doctoral students don’t have an opportunity to take up those faculty jobs.

I wonder whether the lesson of that is that universities should actually expand opportunity, that the humanities need to reform themselves to attract undergrads, that the institutional priorities of who’s getting hired are sometimes wrong, that we should hire more professors rather than more administrators—lessons which are more specific to this particular case rather than the broader thesis. When it comes to the broader economy, it doesn’t strike me that most talented, driven 25- or 30-year-olds don’t have enough opportunities to go and join banking or artificial intelligence, or that there’s not enough capital for them to do a startup, or that they can’t become pharmacists or teachers and so on. To what extent are we overgeneralizing from the world that you and I know?

Moyn: That’s a fair question, and I don’t mean to stake my case on academic gerontocracy—in the few pages I address it, it’s really treated as its own distinctive, idiosyncratic phenomenon. However, aspects of it turn out to be more familiar than your binary presentation suggests. I agree that we are a bit illusioned about the general picture because of certain vanguard sectors of the economy that do indeed provide youthful opportunity—Silicon Valley would be the classic example—but most of the analysis I give of the job situation is making the point not about the academic scene but about American business more generally, which has features that resemble universities to the extent that there are apex positions and a pyramid of authority. The higher you go, the older you are, in the last thirty years. In part, the reason is that across these sectors, mandatory retirement—which was once a familiar element of the American employment landscape—was abolished in the later nineteen-eighties through the early nineteen-nineties. More generally, all of these sectors involve aging Americans who are being not just kept alive but in at least a high enough functioning state to indulge the illusion that they’ve still got their mojo, and they stay. We’re really talking about most sectors where American innovation is at stake, and Silicon Valley would emerge as a kind of outlier.

Mounk: There is a distinction between areas where the argument seems immediately compelling—which is politics and academia—united by having a stagnant number of positions to be meted out. There’s only one president, there are only 100 senators, and the faculty of major universities hasn’t expanded in the way it might have. When it comes to CEOs of S&P 500 companies, by definition there are 500 of them. Opportunity in the economy as a whole does seem to be a different story.

Moyn: The S&P 500 takes up a gargantuan percentage of the economy as a whole. Those who own their own businesses are even less constrained and more likely to stay.

Subscribe now

That may be permissible, and now we’re really talking about employment, not the economy in general. I am very modest in this regard. I agree with you that we need to analyze sector by sector, industry by industry, where there is a supply of younger people who are objectively blocked because of a bottleneck failure to organize succession. Until recently, big law—another elite profession—was a kind of outlier in requiring mandatory departure of its senior partners, and that’s eroding now. It was actually a sector that preserved a sense of the importance of arranging intergenerational succession of counsel for those who have legal problems, and it could be legally structured in such a way as to sidestep ordinary prohibitions in federal law on mandatory retirement.

That would be a different sector, although it’s increasingly looking like the dominant form. The difference between us may be whether to resolve all the complications in the direction of saying there’s a dominant situation—which I think is the case—or whether we really see a few pockets of gerontocracy and it’s more a minor problem to deal with in those few sectors where it’s significant. That can’t be right, because it’s too general even if we shouldn’t overgeneralize.

Mounk: I don’t have a strong stake in this fight, I have to say. I think there’s something clearly intuitive and appealing about the thesis of a gerontocracy in America, and there are certain elements—particularly in the political system and perhaps in academia—where that strikes me as being right. I perhaps have a comparative instinct where this problem seems so much more pronounced in every other society I know: whether that is China, looking at the composition of the central committee of the Communist Party and all of the top political officials; whether it’s the fact that a lot of young people who are graduates of the most prestigious universities feel they’re never going to be able to afford an apartment in Shanghai or Beijing if they don’t stand to inherit one; whether that is extreme rent protection laws that mean that if you came to Beijing or Shanghai from the countryside in the seventies or eighties, you’d probably live in a big apartment in the center, but as a result everybody else has that problem.

Or obviously looking at Europe—there you have additional problems, which in part China now has as well. If you find employment, you have an extremely generous pension system, as in France, where pensioners now on average have more income every month than working people. In the United Kingdom there is the triple lock, which basically ensures that pensions will always outgrow the salaries of working people by definition, as long as the policy stays in place. All of that is purchased at such strong levies on work that you have a huge problem of youth unemployment that by and large you don’t have in the United States. You can see that as a glass half empty or glass half full: either gerontocracy is actually a problem beyond America and perhaps America is in danger of ending up with the same depth of problems, which makes the problem even more urgent, or Americans are a little bit spoiled and should look at the ways in which other countries have this problem so much worse and appreciate what they have.

Moyn: In the end I’m always going to want to transcend comparison to other places that are worse, in the name of comparison to what society ought to look like and how it ought to be organized. The comparisons are complex because Western Europe is indeed suffering in terms of its capitalist growth compared to the United States right now, and so it’s producing fewer jobs for young people—even though in the American case those may be what David Graeber called bullshit jobs. At the same time, Western Europe has a much more widespread culture of retirement and a welfare state that comparatively enables retirement and allows Western Europeans to avoid fear of decline, in a way that Americans—not just because of their work ethic, but because of their extreme fear of what their long-term situations will be—can’t afford to do. It’s not surprising, then, that not just professors at the pinnacle UK institutions but all professors across Western Europe are subject to mandatory retirement rules, and that’s true in many industries. I see a lot of bright spots in Western Europe and a lot of resources for the anti-gerontocratic campaign.

Mounk: What is it worth if people have to retire at 65 but young people have a problem of mass unemployment? Wouldn’t you rather have people not having to retire and there’s actually opportunity for young people?

Moyn: Absolutely, but there may be sectors in which there’s actually a choice. I’ve made the argument that academia in America is like much of organized business, especially in large firms, where there are a lot of people in the firm who have the same experience of serving endlessly under old men and women. The market wants them to leave, and the stock price of companies actually increases when an old leader finally gives up power, whereas it goes down when a younger CEO falls ill or dies unexpectedly. In the name of growth, you might really want to take seriously how widespread the pipeline issues are, and argue that in the cases where that’s true, we should combine growth with mandatory retirement.

Mounk: Comparing those places, the United States seems to me to have both a lot more opportunity for young people and a lot fewer rules about all people having to retire. It seems superior on both of those metrics in terms of the opportunities it gives to people than Germany or France.

Moyn: Most millennials have a generational experience of living after the 2008–09 financial crisis and feeling an immense sense of blockage since, and that may not be shared universally. It doesn’t mean that jobs aren’t available to them—they may not be the jobs they wanted. I take your points, but I worry about vast generalizations when we should really look sector by sector and figure out whether it’s true that America is already this neoliberal utopia where, because of the lack of rules, it’s the land of opportunity for all comers. Not really.

Mounk: That’s not exactly what I said, but what do you think we should do about this? What is the set of responses we should have, including in the political realm, where you make some really quite provocative proposals?

Moyn: What’s helpful about your questions is that we’ve really isolated jobs—both political and nonpolitical—from the overall analysis. Whereas I said I’m as focused, and maybe more focused, on the political system, gerontocratic institutions, and the organization of elections—not just jobs in the economy, but the economy more generally. However, if we’re going to talk about jobs, I believe in age limits as well as youth quotas, for which we can look to Western Europe principally for inspiration. The trouble with term limits—which are very popular among Americans when polled, as are age limits—is that Americans are entering their political careers later in life. Term limits merely give them a time limit. Age limits have the virtue of guarding against the risk of cognitive decline or death to a much greater extent. Youth quotas are really exciting because, no matter where you set a limit on service, you’re basically going to have many or most politicians closer to that limit than we would like. What we really see as the problem, I think, is not cognitive decline or death, as I said earlier, but a lack of representation for most age cohorts. Youth quotas allow us to correct that to some extent.

Mounk: Age limits is a slightly broad term. What do you think is an appropriate age limit for a senator, a professor, or a CEO?

Moyn: I’d go for seventy right now, but I don’t feel strongly about the number. I’m more committed to the philosophical idea, especially as our life expectancy goes up—and it may go up in leaps and bounds. I don’t think it will, but we have to leave open the possibility that it will happen. The number has to be flexible and open to change. As of today, I’m for 70-ish.

Subscribe now

Mounk: You’re the Kent Professor of Law and History at Yale University and the head of Grace Hopper College. I believe if you introduced age limits of 70 for the professoriate, that would give you about 16 more years of service. Are you willing to pledge to retire when you turn 70?

Moyn: I’ve done it—in the Chronicle of Higher Education. I spun off the few pages in the book about gerontocracy and I did commit to that there. It’s not a problem because I would love to retire now and I really think of myself as semi-retired to begin with.

Mounk: Retirement for academics is kind of fake, right? I was speaking to a very prominent academic the other day who was saying they were really thinking about whether to retire next year or the year after. I asked how come, and they said, because it would allow them to write more books. Why should they, if they no longer need the money? Obviously they love teaching. But at some point you think you want to focus on your writing. That’s not really retirement in the way that it is for most people. A lot of people love their jobs, but they have jobs where when they retire they no longer get to do their jobs. That’s kind of different for us, where at least a lot of our work involves things that we get to do more of if we retire from our faculty positions.

Moyn: That’s true, and I discuss this in the book. My own view is that we should think of retirement, at least as a default, as a last chance to reinvent yourself. It’s just true that some people will have an experience of their vocations that makes it impossible for them to give it up—existentially. I think that’s a mistake for them existentially when that’s the case, but I’m not going to mandate that they go fishing. Is fishing really an acceptable way to spend your retirement?

Mounk: Tell us about some of the political proposals. One of the proposals you float is that perhaps—I forget the exact details—you get one vote if you’re 80 or over, two votes if you’re 70 or over, and three votes if you’re 60 or over, all the way down, not quite to toddlers. How seriously do you mean that? What about the straightforward objection that the point of democracy is meant to be one person, one vote, and that perhaps some of our constitutional realities—like the ban on running for president before you’re 35—should go out the window, that we should approach the principle of one person, one vote more rather than less? Saying that simply on the arbitrary basis of which year you were born you’re supposed to have five times more votes than the other person casting their ballot next to you seems like a pretty fundamental violation of that foundational democratic principle.

Moyn: There’s a general discussion to have about just how bad electoral gerontocracy is. Even when you and others are rightly on a campaign for #ourdemocracy, we really have to reckon with the truth about what we’re defending, which is an electoral gerontocracy. There’s a long list of remedies, many of which—actually all of which—are less radical than the one you’re speaking of now. Let’s talk about it philosophically. First, it’s just not true that we should think of democracy as involving one person, one vote. It certainly doesn’t in the United States right now, because of the Electoral College and the Senate, which basically overweights the votes of those from less populous states by constitutional design. More generally, the idea of one person, one vote became an article of faith in the United States largely as a result of Supreme Court decree in the context of the civil rights era. If you go back to the origins of democracy and mass suffrage in the nineteenth century, there were very widespread proposals for plural voting and organizing voting generally very differently than we do, and there was an argument for it. The argument I would make is: what does it mean to be equal in a democracy? Right now we basically say that if you have one body, no matter its age, you should get one vote. What if we had a principle of equality that said your vote should be correlated with how much time you have left, so that we don’t overweight—which we do now—the vote of those who are closer to death and therefore less likely to see and live under the policies they’re choosing.

Mounk: So if you’re diagnosed with a very serious disease that foreseeably leads to your demise within the next five years, you should get your vote discounted.

Moyn: The policy of so-called one person, one vote is a policy of systematically overweighting the votes of those who are older—on one conception of equality.

Mounk: So why is it over-weighting them? It’s weighting them equally.

Moyn: The question is what are we equalizing: bodies or life expectancy?

Mounk: The basic problem of society is that I have a bunch of boneheaded ideas about how we should be governed, and you have a bunch of boneheaded ideas about how we should be governed, and there isn’t a God or some kind of objective authority that’s going to tell us whether you’re right or whether I am right. We’re trying to design a set of institutions that we can all somehow live with, because we realize that not having political organization at all, or taking up arms in order to clobber the other’s boneheaded ideas out of their mind, is not worth it. The cost of that is too high. The whole point of democracy from that point of view is to say there isn’t one person who is in some way superior to the other—we should all have the same amount of voice. The idea of one person, one vote is not some kind of epiphenomenal invention; it flows relatively logically from that principle. Now, that doesn’t mean that this is perfectly respected in the United States or in any other democracy, and we can certainly discuss reforms—whether that’s campaign finance or reforms to the Electoral College that would allow us to more fully live up to this. To claim that there isn’t some fundamental set of reasons why one person, one vote really speaks to the basic aspirations of a democratic system is, I think, to ignore something quite important.

Moyn: The history just doesn’t back you up. It wasn’t obvious until it was, and that’s among other reasons why everyone—all liberals—believed for a long time that what mattered was not that you had a body that was alive, but that you had an adequate stake in the election and the capacity to exercise the relevant judgment, and then there was one person, one vote among those people. That was hegemonic for a very long time.

Maybe there were some folks who said there were exclusions, in the same way that I could say children and prisoners are kept from voting in the United States even though they have bodies that are alive. My actual point is somewhat different, which is that we could imagine historical circumstances in which it becomes equally intuitive to say that the equality that matters is the equality of your stake in the election. That actually was the rule, as I suggested, for much of the history of democracy, because it was a natural thought to those early democrats that you couldn’t have a stake in the election without education and, more importantly, property—because that’s what made politics matter, how property was going to be protected or not.

Mounk: Here I think we’ve come to our fundamental difference: you’re an intellectual historian, I’m a political theorist, so we have different approaches to this. I wasn’t making a historical claim; I was making a conceptual claim about the nature of democracy.

Moyn: I’m talking about the histories of political theorists who believed the opposite of what you do. I agree it’s totally intuitive now—you’re right that it’s totally intuitive. I guess I’m looking at the situation and saying: if that is really an ideological smokescreen for gerontocracy, then we might have to revisit it in creative ways.

Mounk: If we’re going to revisit this, if we’re going to open up the black box of one person, one vote and say that for various reasons various groups of voters should have a vote that counts for more than those of others, why is age the one that matters over others?

Moyn: It’s not. I think Black people should have more votes than I do. John Stuart Mill believed—something I don’t believe—that those of us with doctorates should have more votes.

Mounk: So who else? Should Latinos have more votes?

Moyn: We could give women a modest increase.

Mounk: So women should have more votes. What about disabled people?

Moyn: That would be subject to debate. If you’re someone who cares most of all about the fate of minorities who are unlikely themselves to enjoy political power unless they enter into compromising coalitions, then we could proceed that way.

Mounk: What about trans people?

Moyn: My own view is that we should have an ongoing debate about how to organize our elections. We do in some sense, but we’re locked into a constitution that basically forbids us from experimenting with our electoral system. If you’re asking what I would vote for: first of all, we need more descriptive representation—that’s to say, elected representatives need to be produced by the system who are understood by relevant sectors of the electorate to represent them. That’s why the Calais decision—I don’t know if you supported that one—is so noxious: it forbade arrangements that allowed Black people to have Black representatives in this country. More generally, now we’re talking about actually sculpting the electorate itself and who gets what powers, and there I would be open to experimentation. The age case is to me the clearest one.

Mounk: Your point is that if you’re young, you should have more votes. I challenged you by asking about ethnic minorities, trans people, disabled people, and as I take it your answer was broadly yes, they should all get more votes as well.

The problem with this is precisely that it undermines the point of democracy in adjudicating between our fundamentally different visions of government. Democracy is the technique that, perhaps not in a unique way but better than any of the alternatives in human history, has been able to say: we take this huge, diverse country with people with a vast multiplicity of views and preferences and interests, and the way that we keep social peace, the way that we avoid civil war, the way that we avoid a complete falling apart of this polity is to say everybody gets one vote. There is also a certain set of fundamental rights that protects you against, for example, the government locking you up for worshipping in a way that others don’t agree with. What you’re suggesting is completely exploding this. What we would then have to do is arrive at some Habermasian consensus about exactly how many votes Black persons should have, how many Latino persons should have, how many somebody with one Black parent and one Latino parent should have, and whether trans people should also have multiples. What it would actually encourage in any real-world sense is naked interest group politics along racial, ethnic, and other lines, in a way where there’s never any consensus and the only way to contribute to policy discussions is to band together explicitly as members of monolithic groups to protect the interests and vote of your group against those of the others. This isn’t about whether in the 1850s Mill had a different view and thought that more educated people should have multiple votes. It really is about whether this would be a deviation from a democratic principle that actually undermines the historic achievement of this political system.

Moyn: I really appreciate those arguments, but I’m not sure I agree with them, for a couple of reasons. One is that your alternative is the nightmare scenario. Your case there at the end was really the familiar case against affirmative action, familiar to us from the neoconservative movement’s theories about the Balkanization of America—Arthur Schlesinger and so forth. I don’t think that’s all that’s going on, and it’s certainly not how we avoid fascism or tyranny, which has a lot to do with material factors and not mainly how we organize elections. The big response I have is that there’s no non-neutral technique, and the idea that one person, one vote is it, I think, is not persuasive.

Mounk: Don’t you think there’s something less politically fraught, and less subject to the influence of changing electoral majorities and insiders and outsiders, about saying the moment you are a citizen you get one vote—versus having a multiplier where, depending on the particular intersection of identities at which you stand, that multiplier is three times or five times or seven times or 0.2 times?

Moyn: You’re certainly right that in the real world a lot depends on what people empirically think, and you’re probably right that as of today a lot of folks would say that one person, one vote is just the only credible way of organizing elections, because it’s the only neutral principle.

Someone could write a book, and there could be whole movements, basically saying that’s false and there’s a more just way to do this. I’m not saying there’s going to be some big conversion. I agree with you that what people think out there really matters. Even if affirmative action was just, it really matters that the majority thought it wasn’t—certainly the majority of white people. You’re completely right about that. I thought we were talking about the matter of principle: can we argue credibly for an alternative to one person, one vote if it turns out that’s itself a partisan, non-neutral view? More generally, if you went out into the street and said, okay, you say you’re for one person, one vote, but how can you defend the Senate?—no one would understand it. We can’t be hostage to public opinion in that way if we’re trying to think about our principles and whether they’re credible and consistent. I concede a lot to you. This is a very important point you’re making.

Mounk: On the Senate—virtually every democracy has ways to ensure that different territories have influence in different kinds of ways. In the European Union, Luxembourg has as much vote on certain kinds of votes as Germany or France does. That’s just the cost of large-scale political unions between entities that have, to some extent, a different identity that persists. The second thing to say is that I think there’s a lot of enthusiasm in America for either abolishing the Senate or diminishing the extent to which it is geographically unfair. That’s precisely because I think there is a compelling logic to one person, one vote that is perhaps cushioned by the need to have some kind of political representation for smaller states with their own identity. The reason people have trouble explaining the Senate is precisely that one of the fundamental principles to which they intuitively hold is that a voter in New York or Texas should have the same influence as a voter in Montana—and that’s currently not the case, and that seems unfair. The logic of that seems to suggest we should go closer toward one person, one vote, not that because we tolerate, for complicated historical reasons as well as for the need for geographical representation, some deviations from it, we should give up on the principle altogether and think it doesn’t constrain us when it comes to fundamental things like whether a 70-year-old should have the same vote as a 30-year-old, or a white person the same vote as a Black person.

Moyn: I love that and I’m sure it’s right up to a point. The question is whether concluding that we’re overweighting the wrong people’s votes—namely those in small states—means, logically or otherwise, that we shouldn’t overweight somebody else’s vote. I know a lot of people who hate the Senate but believe in majority-minority districts. What’s that about? It’s about saying we should organize the districts so that within them, Black people who are a minority in the state generally can get someone of their race elected. You may not agree with majority-minority districts, but it’s certainly the case that most liberals think both that the Senate is problematic and that Callais was bad.

Mounk: The most cynical reading of that—irrespective of the merits of each of those institutional arrangements, and one that I think you, as a certain kind of materialist intellectual historian, should actually be quite sympathetic to—is that most of the people you’re talking about prefer the Democratic Party, for all of the misgivings we may have about it, over the Republican Party, and the over-representation of small states harms the Democratic Party while majority-minority districts help the Democratic Party.

In the rest of this conversation, Yascha and Sam discuss whether we should be skeptical of human rights, why Trump seems to be a spent force, and how the Democrats can present a credible alternative to populism. This part of the conversation is reserved for paying subscribers…

Read more

So Donald Trump, on his 80^th birthday, announced a deal in which there would be a 60-day ceasefire. Precise details have not yet been officially published. But, according to reports, they apparently include a cessation of attacks in Lebanon, the reopening of the Strait of Hormuz—according to Trump, “permanently toll-free”—and lifting the U.S. naval blockade of Iranian ports. He touted this as a key win, in the process praising China’s Xi Jinping and Russia’s Vladimir Putin for helping secure it.

This “deal” was nothing of the sort. If the reports are accurate, it instead represented a total U.S. capitulation to Iran. It basically set the clock back to February, when the Strait was open and the United States and Israel had not yet started bombing the Islamic Republic. It merely solved a problem that Trump and Israeli Prime Minister Benjamin Netanyahu had themselves created by launching the war in the first place.

Still left up to future negotiations are all of the objectives that the Trump administration has set forth over the past three months in trying to justify the war:

• There was no regime change or “unconditional surrender”; the Islamic Revolutionary Guard Corps remains even more firmly in control of the country than previously;

• There was no commitment by Iran to turn over its stockpiles of enriched uranium;

• There was no commitment to stop enriching uranium, either immediately or on some specified date in the future;

• There were no commitments on ending Iranian support for allied groups like the Houthis or Hezbollah in the region;

• There was no agreement by Iran to let up on the violent suppression of protesters.

The reported “Memorandum of Understanding” (MOU) kicks all of the contentious issues down the road into negotiations that are to take place during the 60-day ceasefire. Trump treated all of these issues as having been conceded already, but if that were the case, why weren’t they in the MOU? It is very unlikely that Iran will budge over the next two months, since it is precisely these issues that speak to the regime’s core identity.

Trump stated that if Iran didn’t agree to these outstanding terms, he would re-commence the war and possibly make the United States “the guardian of the Middle East” in return for 20 percent of the region’s revenues. It is hard to know whether such an initiative is more ludicrous from the standpoint of countries in the Middle East, including U.S. friends like Saudi Arabia or the UAE who would now be paying explicitly for U.S. protection, or from domestic opinion in the United States, where everyone would like to be done with the region as soon as possible.

The MOU that Trump celebrated is a worse agreement than Obama’s 2015 deal, which Trump endlessly castigated in the past. Obama’s deal forbade Iran from enriching uranium beyond 3.67 percent for 15 years (far below the 90 percent enrichment necessary for bomb-grade purposes), and provided specific measures for removing enriched uranium from Iran. All of these provisions were to be overseen by outside inspectors, and Iran complied with its terms until Trump withdrew from the agreement. The major criticism of the deal, which U.S. hardliners stressed, was that it said nothing about Iranian support for regional proxies and that it provided sanctions relief at the start of the agreement.

Trump’s reported MOU, meanwhile, places no limits on Iran’s nuclear capabilities, and makes no commitments about regional proxies. It does not provide for sanctions if Iran doesn’t concede by the end of the 60 days, though the Iranians have said that they will not proceed with final negotiations unless such relief occurs first. So Trump’s purported deal achieves considerably less than the agreement that Obama made.

It is clear that Trump is being driven to reopen the Strait of Hormuz at virtually any cost by the domestic pressure from rising oil prices and inflation. Being unwilling to send ground forces to Iran, he has had few cards to play over the past six weeks to get further Iranian concessions. So he has chosen to back down and accept a return to the status quo ante from before he began the war on February 28.

The world will indeed be better off if the Strait is re-opened. Perhaps Trump’s hardcore MAGA supporters can be persuaded that he has negotiated a consummate deal and achieved a great victory. But everyone else will understand that the world’s most powerful country is being run by a feckless and ignorant president who will impose immense costs on both other countries and his own people if he thinks it will benefit himself.

Francis Fukuyama is the Olivier Nomellini Senior Fellow at Stanford University. His latest book is Liberalism and Its Discontents. He is also the author of the “Frankly Fukuyama” column, carried forward from American Purpose, at Persuasion.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

The No Kings rallies in March 2026 were perhaps the largest single-day protest in the United States since the first Earth Day in 1970. Millions showed up at over 3,000 locations in a display of resistance against the second Trump administration.

But days later, the protests had already faded from the public mind. The White House seemed unbothered. Trump continued to embrace authoritarian tactics, targeting his enemies in the courts and waging a war in the Middle East without the consent of the legislative branch.

That hasn’t stopped No Kings from trying again. Today, the movement is co-hosting a “Rise Up, Sing Out” concert in New York (with watch parties across the country) to coincide with Donald Trump’s 80th birthday celebrations. According to the No Kings website, the event is an opportunity to “sing along, make art, share food, connect with neighbors, and take meaningful action together.”

If Americans want to actually enact change, they seriously need to re-think their strategy. Take it from us: we both grew up in Putin’s Russia and saw well-intentioned protests fail to stop an aspiring despot. We know that authoritarians are typically unwilling to respond to the kind of protest No Kings exemplifies: loud, raucous, and ultimately harmless. These “festival protests,” as we call them, are convenient for their participants. They are fun and usually do not require much sacrifice or risk. They also look good on TV and TikTok feeds. But they often achieve next to nothing.

Why are so many people convinced they work?

The festival approach to protesting has its roots in the end of the Cold War. The fall of communism in Eastern Europe was accompanied by largely peaceful popular mobilization, which created a perception that revolutions are something fun and frictionless. The Czechoslovak anti-communist protests in 1989 even got the Slovak moniker of nežná—the “gentle” revolution.

These protests were subsequently written up as a key reason for communist collapse. The perception that a successful revolution can be a fun affair was so omnipresent it seeped into theories of change and scholarly work. Theoreticians like the late Gene Sharp wrote protest manuals which popularized the idea that even a hardline dictator will bend to popular will if that will is manifested in a suitable rousing manner.

The ensuing “color revolutions” of the early 2000s—a series of protests that sought to peacefully enact democratic transformations in post-socialist countries—seemed to vindicate this approach. The meek strongmen Eduard Shevardnadze in Georgia and Viktor Yanukovych in Ukraine melted away under the creative slogans and color-coordinated marching columns of bright-eyed youth.

But the real story of those successes is much more complicated. The regimes fell in places where their foundations were already weak. Shevardnadze, famously, couldn’t even pay police officers. It’s no wonder they did not want to protect the regime once protesters came. Similarly, in Ukraine’s protests of 2005 and 2013-2014, a large sector of the elite—oligarchs, high-level officials, politicians, and members of the security apparatus—were willing to defect to the side of pro-democracy protesters.

We are not seeing massive defections among American elites. Republicans in Congress support the vast majority of the administration’s initiatives, while business leaders rarely stand up against the president, even when his actions (such as tariffs) hurt them directly. Establishment media such as The Washington Post and CBS News already show signs of self-censorship. Courts and some Democratic states are the only traditional institutions that display systematic resistance.

Then there’s the fact that protests in general are becoming less effective. In the 1990s, around 65% of non-violent movements succeeded in overthrowing a dictator. In the late 2010s, that figure was down to 34%. Violent movements are even less effective—their success rate is currently around 8%, down from a peak of more than 40% in the 1970s.

This is partly because authoritarian rulers have learned from their mistakes since the color revolutions and the Arab Spring. When one leader gets in trouble, others come to help. Russia, for example, sent troops to Venezuela and Belarus during recent protests. The regime also invested massively in preventing protests at home. The pro-democracy “Snow Revolution” of 2012 flopped because the Kremlin was able to maintain elite cohesion and police loyalty. Moscow was an odd city during the snow protests: While several squares in downtown were occupied, ordinary life more or less went on as usual.

Making life comfortable enough that most people can disengage from politics and ignore protests is Putin’s greatest accomplishment. The well-educated urban elites do not call for civil disobedience or direct action—they know that most Russians are not ready for such sacrifices.

Subscribe now

So what lessons are there for the resistance in the United States? Let’s look at Minneapolis. During ICE’s raids earlier this year, protesters made sure to warn the local community by blowing whistles, shouting, and banging drums. They organized solidarity networks and boycotts. The tactics were extremely demanding: protesters had to engage in constant surveillance of law enforcement and skip work to participate in non-violent direct action. Two lost their lives.

These disruptive tactics deeply angered officials, and eventually made them retreat. It slowed the repressive machine of the state. The faces of the anti-migrant campaign—Border Patrol Commander-at-Large Greg Bovino and Secretary of Homeland Security Kristi Noem—were quickly fired.

Minneapolis proved that resistance movements should not be evaluated by the number of people they bring to the streets. Resistance to authoritarianism is not an emotional support group.

In Russian, there is a joke about good-hearted but ineffective pro-democracy protesters: “They are for everything good and against everything bad.” A successful anti-authoritarian movement in America will be a movement of people who are ready to make sacrifices. It will not come from a place of comfort.

Maria Kuznetsova was part of Russia’s protest movement from the 2010s up until her exile. She is the co-author of How to Survive Authoritarianism: A Russian’s Phrasebook for Everyday Life in America.

Dan Storyev has covered protests on the ground as a photographer and reporter. He is the co-author of How to Survive Authoritarianism: A Russian’s Phrasebook for Everyday Life in America.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

June 14 is an important halfway point in America’s summer civic holidays, connecting Memorial Day with Independence Day—and in more ways than one. June 14 is the anniversary of the birth of the U.S. Army in 1775—and is today celebrated as Flag Day.

“A yearly contemplation of our flag strengthens and purifies the national conscience,” declared President Calvin Coolidge. Echoing sentiments Woodrow Wilson expressed in his 1916 proclamation recommending the annual observance of “Flag Day,” Coolidge succinctly summed up the point for such a national holiday: “We see in [the flag] the great multitude of blessings, of rights and privileges that make up our country. But … we must remember that it is equally a symbol of our duties. Every glory that we associate with it is the result of duty done.”

Like Achilles’ famous shield depicting both war and peace, the American flag symbolizes the blessings and the duties of a self-governing nation dedicated to freedom and equality. The blessings are frequently invoked, while flag-draped coffins of fallen soldiers show viscerally both the glory and the cost of “duty done.” But, as successive presidents from Wilson onward have noted in relation to Flag Day, the stars combined with the red and white stripes remind us of the ongoing and current duties we carry as citizens, chief of which is respect for the rule of law.

Persuasion events

“In whatever direction we may go we are always confronted with the inescapable conclusion that unless we observe the law we cannot be free,” noted Coolidge. Or, as that philosophical grandfather of America John Locke put it, “the end of law is not to abolish or restrain but to preserve and enlarge freedom.” In the American tradition, this observance of the law as a supreme principle is two-pronged: it means that laws when made must not designate winners and losers, applying unequally to different persons or groups of society; it also means that all alike, from those in elected office, enjoying positions of economic power, to the those struggling to make ends meet, must follow the law—and can enjoy the shared peace this brings.

But, as the stories in our newspapers seem frequently to remind us, the rule of law is much easier to invoke than to practice, particularly as it relies upon the goodwill of human agents. Like the flag itself, it is fragile without both.

Subscribe now

Flag Day is not an official holiday. Rather, mandated by a 1949 act of Congress, the president is “requested to issue each year a proclamation, calling on United States Government officials to display the flag of the United States on all Government buildings on Flag Day; and urging the people of the United States to observe Flag Day as the anniversary of the adoption on June 14, 1777.”

America’s flag, and indeed all flags in general, originally had a predominantly military purpose. It emerged out of the Revolutionary War (although not sewn by Betsy Ross) and its martial image became fixed for many Americans through the “gave proof through the night / that our flag was still there” lines in “The Star-Spangled Banner,” an 1814 poem that over a century later would be adopted as the national anthem. Marking the 200^th anniversary of “The Star-Spangled Banner” in 2014, American composer extraordinaire John Williams celebrated the link between the nation’s flag and its national anthem, when on the steps of the U.S. Capitol he debuted a new arrangement of the Star-Spangled Banner. During this 250^th anniversary year of our American nation, we are reminded that our flag is still here. Remember its glories; remember, too, its requests.

Rebecca Burgess is a senior fellow with the Yorktown Institute and at Independent Women. She serves as an advisory board member at the Thomas Jefferson Foundation at Monticello.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

This week, Richard talks to Laurence Jurdem, author of 41: George H. W. Bush and the End of the American Establishment (Bombardier).

Host: Richard Aldous

Producer: Bhaasita Athani

Follow Persuasion on X, LinkedIn, Instagram, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below. Bookstack listeners can take advantage of a 20% offer for paid Persuasion membership, with access to special columns and bonus Good Fight with Yascha Mounk podcast content.

20% off for Bookstack listeners!

On June 9, a U.S. Apache helicopter was reportedly shot down by the Islamic Revolutionary Guard Corps’ (IRGC). Forty-eight hours prior to this, the IRGC fired ballistic missiles towards Northern Israel. These attacks on Israel came just days after Tehran carried out strikes on Bahrain and Kuwait—and as President Trump attempted to pressure Tehran to reopen the Strait of Hormuz and move toward a broader settlement.

These events should temper Western optimism about the negotiations currently underway between Iran and the United States. Even if a deal to reopen the Strait of Hormuz and extend the ceasefire has “never been closer,” as Iranian Foreign Minister Abbas Araghchi said on Friday, the issue is that the main actor shaping Tehran’s behavior is not the civilian government but the IRGC. For the Guard, keeping the Persian Gulf in a state of controlled crisis is not just a bargaining tactic; it is also a way to further marginalize civilian institutions and preserve the security order through which it now dominates the Islamic Republic. 

The extent of the IRGC’s dominance in the regime was made clear when it singlehandedly decided to wage a drone attack on the United Arab Emirates on April 30. Based on leaked reporting, President Masoud Pezeshkian was furious over not being informed about the IRGC’s decision, reportedly calling the escalation “madness” and warning that it could push Iran back into a full-scale war. The military-security core of the regime, especially the IRGC, sees a contained war and crisis as an opportunity to accelerate its existing plan to completely take over the state.

This runs against the view of some observers who believe the Islamic Republic will eventually choose a deal over war. From this perspective, Iran is under pressure, its economy is being destroyed, its society is exhausted, and its military infrastructure has been badly damaged. Therefore, according to this view, as a “rational” regime, Iran will seek de-escalation, compromise, and sanction relief. But this take fails to properly grasp the nature of the IRGC.   

The IRGC does not view war and peace the same way a normal state institution does. It is not simply a military organization defending national borders. Instead, it was created to defend the Islamic Revolution, the Velayat-e Faqih (clerical guardianship), and the regime’s ideological identity. Its mission has always been larger than Iran’s territorial security: it includes exporting the 1979 Islamic revolution, eradicating the State of Israel, confronting the United States, protecting the regime from internal enemies, and reshaping the regional order.

This is why the IRGC may prefer a contained war, or at least a continued limited confrontation, to a deal based on major concessions. For the Guard, compromise is an ideological danger. Shaped for decades by thorough indoctrination, selective recruitment, and internal surveillance, IRGC members are trained to see the world through the lens of resistance, martyrdom, anti-Americanism, antisemitism, and blind loyalty to the Supreme Leader.

Since its inception, every major crisis in the history of the Islamic Republic has resulted in the expansion of the Guard’s role. The 1980-88 Iran-Iraq War transformed the Guard from a revolutionary Islamist militia, which was established to protect the clerical rule, into a professional military force with five branches. Similarly, the 2009 Green Movement uprising—which saw millions of Iranians take to the street—resulted in the IRGC’s monopolization over the security and intelligence apparatus of the regime. Much like the Iran-Iraq war, the Syrian civil war led to the expansion of the IRGC’s elite Quds Force. The IRGC has calculated that, unless it results in the complete collapse of the system, the current war—which began in June 2025 and continues to this day—can accelerate this trajectory by further empowering the Guard and turning it from a deep state into the dominant state itself.

As a result of the war, IRGC commanders have become indispensable, while the state bureaucracy, parliament, diplomats, and even presidents are completely secondary. Nobody takes the president seriously, even in his own administration, while the parliament has been closed since February, and many of its members are unaware of wartime developments. The state apparatus has been marginalized completely—while the security forces are now making all decisions.

Furthermore, IRGC members live in an isolated, institutional world, and their salaries, businesses, privileges, housing, hospitals, and networks are tied to the regime, meaning they often do not feel the same pressures ordinary Iranians do. Sanctions hurt Iran, but they also create opportunities for the IRGC. A closed and sanctioned economy increases the value of those who control borders, ports, contracts, hard currency, and black-market channels. And while the IRGC would still benefit from the lifting of sanctions—as was the case under Obama’s 2015 nuclear deal—the illicit economic ecosystem it currently commands remains of greater strategic value.

Finally, since the conflict began, the United States and Israel have eliminated more than 100 high-ranking IRGC commanders as well as the late supreme leader, Ali Khamenei. For the IRGC, these were spiritual leaders, peers, mentors, brothers, and fathers. Accepting a deal with Trump could be framed as a betrayal of the martyrs, not just inside the organization itself but also among the roughly 10% of the Iranian population who are radical Islamists. Thus, such a “betrayal” would risk the collapse of the IRGC’s reputation within its core support base.

But what does all this mean for the Trump administration? It is clear that as long as the IRGC exists, a permanent solution to the Strait of Hormuz crisis—and long-lasting peace in the Middle East—remains unattainable. Any strategy pursued by Trump and America’s Western allies should therefore focus on continuing to undermine the IRGC. This includes a range of measures, from restricting the Guard’s access to financial resources through clamping down on illicit sanctions evasion networks and stricter sanctions enforcement—something Trump should pressure European governments to implement—to dismantling its regional infrastructure and weakening the organization through military means.

Despite the defeatist rhetoric that is prevalent in much of the mainstream media, regime change in Iran remains possible—but only if the IRGC is significantly weakened. Until then, the IRGC will pursue its “forever war,” threatening regional stability, the global economy, and U.S. national security.

Kasra Aarabi is the Director of Research on Iran’s Islamic Revolutionary Guard Corps at United Against Nuclear Iran.

Saeid Golkar is an associate professor of political science at the University of Tennessee-Chattanooga, a senior advisor at United Against Nuclear Iran, and a writing fellow at the Middle East Forum.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

David Bau is Assistant Professor at Northeastern University and Director of the National Deep Inference Fabric, researching the emergent internal mechanisms of deep generative networks in both Natural Language Processing and Computer Vision.

In this week’s conversation, Yascha Mounk and David Bau discuss how AI models actually produce their results and reflect about problems, whether the “thinking” process that models show users reveals their authentic thought processes, and how researchers can decode the internal representations of neural networks to understand what information they contain and use.

This transcript has been condensed and lightly edited for clarity.

Yascha Mounk: I learned so much the last time we spoke that I thought I would abuse your generosity and reel you in for another private tutoring lesson about how AI works. When we talked last time, we did AI 101—that’s how I was thinking about it. How does this thing work? How do you build an AI? How does it operate? The question I want to start with today is: how does the AI actually produce results? How does it actually reflect about the world, reflect about a problem, plan how to carry it out? What can we even know about that?

David Bau: This is one of the mysteries of AI—how does it work inside? The way that we train AI is to basically reward it, reinforce it, or strengthen its connections when it gets answers right, and then weaken those connections or withdraw a reward when it gets something wrong. Repeating this process billions of times, it starts to perform well on all the tasks. The mystery is: how does it do it inside? The whole area of trying to understand what’s going on inside the AI—some people call it AI interpretability, cracking open the AI to interpret what it’s thinking inside—is actually my area of research specialty, so I’m happy to get into what we know about that.

Mounk: In a way we have some advantage relative to the human brain, right? Reading exactly what neuron is firing when in the human brain is incredibly hard, and getting good readings even on a mouse while it’s alive is an incredibly difficult process. Presumably the one advantage we have in the context of these models is that we can, I assume with greater ease, observe which part of a neural network is activated in which way, and is changing values in what way, while I am asking Claude what 3 plus 5 is or whatever.

Bau: Yes. That’s the amazing thing about having artificial neural networks that work—it’s an embarrassment of data. It’s the flip opposite of what you are dealing with when you are dealing with biological brains. The neuroscientists are amazing; they do look at neurons of mice and they have incredible ways of doing that. It can take five years to look at a handful of neurons, and in computer science, within a few minutes, it’s very easy to look at billions of neural signals. It’s so much data that our challenge is trying to figure out how to sift through it all to make sense of these signals. What we call the neural pattern that you see when you feed some input into a neural network—it creates a pattern of neuro-firings that we call the representation. It’s a representation of some information that’s inside the network. What we’re frequently trying to do is understand two key things: what information is inside the neuro-representation—what does it know inside its neurons, what information does it have—and then what does it use it for? What information does it use, and how does that impact its decision? I think you can distill a lot of the questions of how a neural network works down to these two things: what does it know, and what does it use?

Mounk: Just from a layman’s perspective, the first obvious question is: when I ask Claude to do something complicated, it’ll say “I’m thinking,” and you can click on that and it expands a little thing that tells me what it’s doing and what it’s thinking. It says, “the user requested this, I should do that.” But of course, I have no idea whether that is any closer to its actual thought process. It does seem to tell me about some of the steps it’s taking, so it seems to be somehow related to what it’s doing. But that in its mind is still output that I may be inspecting. So is that a window at all into what’s actually going on under the hood, or is it a completely fake output that makes me feel like I get some kind of insight into what it’s doing, without actually being any closer to what it’s doing than the official output it gives me?

Bau: I think most people believe that it is somewhat of a window, but it’s something that you have to take with a grain of salt—it is another output of the neural network. There have been studies that show that that output is not totally faithful to the way that neural networks think inside, in a few different ways. Most people look at that and say, well, it’s certainly better than nothing, it’s certainly very readable, so it’s definitely worth a look. It is definitely worth auditing, and the network will often reveal things in that text that give you some insights about what’s going on. But it’s not the full story.

There are two ways that the network has an internal thought process. One of them is through what everybody is calling its internal chain of thought. This comes from an old paper—“chain of thought” is what they use to talk about this internal monologue. This is what you can click on to see when the model is talking about itself, and that’s almost literally the model talking to itself. It’s generating tokens that aren’t directly intended for you to read. These are tokens that came out of this reinforcement learning process, where the model has somehow learned that in order to get more accurate answers—to solve more puzzles that it was presented with during training—it’s useful to write some things down halfway through.

Mounk: Does it do that in English? Does it always do that in English, even when I’m speaking to it in German? Are there models that have developed their own kind of language for this? What does this look like?

Bau: If you don’t explicitly tell the models to make that text readable, they will write in their own language, switching between English and Chinese and other things. One of the things that people do when they train them is they try to condition the models to make that text a little bit more readable so that we get some insight. But that’s an example of the challenge with these internal chains of thought. The model could be inventing its own jargon. It could be using words that look like English, but actually encoding some other information in those words. We may be reading those words very differently from the way the model reads those very same words. It may be inventing layers of meaning that we don’t comprehend. It might also be performing some other process that isn’t actually what’s reflected in the words at all.

We hope you’re enjoying the podcast! If you’re a paying subscriber, you can set up the premium feed on your favorite podcast app at writing.yaschamounk.com/listen. This will give you ad-free access to the full conversation, plus all full episodes and bonus episodes we have in the works! If you aren’t, you can set up the free, limited version of the feed—or, better still, support the podcast by becoming a subscriber today!

Subscribe now

Set Up Podcast

If you have any questions or issues setting up the full podcast feed on a third-party app, please email leonora.barclay@persuasion.community

Out of AI safety training, we basically train models not to be very offensive, not to have terrible errors or biases or other problems in the text that they emit. What that means is that when they articulate their own internal thoughts, those internal thoughts also tend to be censored in those ways. They tend to not talk about things that we don’t want them to produce in their final output. But that’s not a guarantee that the model is not actually thinking about dangerous things or thinking with a terrible bias. It just means that the model may be encoding its thoughts in a way so that when you read the surface forms of the thoughts, you don’t see the undesirable things—the biases, the problems, the errors. There’s good reason to believe that the internal monologue might actually not reveal some of the things that we wish it revealed. We want the model to reveal to us when it’s doing something wrong, but because of the way we’ve trained it to use language, it just might not be using words that way.

Mounk: In a sense, we’re now getting at different levels of, for lack of a better word, interiority. The first level is just: you ask a question, what answer does it give you? The second level is, if it’s thinking for a long time, it tells you something about its thought process—what is it writing to this thought process? The third is a non-public but auditable scratchpad, in which it is noting stuff down, and there you sometimes have this mix of languages and all kinds of interesting things going on. But obviously the model still understands that this is the sort of thing that might be read and scrutinized by an AI researcher like you. Then there’s a fourth level of the internal thought process, which is more complicated.

I have two questions. The first is: how mutually comprehensible are these scratchpads? If you take the output of a scratchpad like that and feed it to a different model, will it understand it? Is it a kind of universal language between AI models that are at least of a similar generation, that have broadly speaking been trained in similar ways? Or will the latest Claude model not understand the scratchpad of ChatGPT, and ChatGPT will not understand the scratchpad of Claude? The other question is: how do you get beyond the scratchpad to looking at what’s really going on under the hood?

Bau: I have a PhD student—her name is Koyena Pal—who was very interested in exactly this question. What she did was take the internal chain of thought from some models and transplant it into other models, to see how they would respond as if those were the internal scratchpad notes they had written to themselves. Her study is preliminary; I think the most valuable part of it is just the idea that this might be an interesting thing to do. She generally found that the stronger models she tested were able to create internal monologues that other models did understand—that they actually tended to follow those thoughts and come to similar conclusions as the more powerful model.

Mounk: Some of these things are ones that humans would have great trouble interpreting.

Bau: I think that is still an open question. She also looked at the ones she studied and found that humans actually positively correlated with these—the more effective chains of thought were actually more human-interpretable. But human interpretability is a funny thing; it’s a perception thing. Do humans feel like it’s more understandable? It’s hard to get a read on whether this is actually giving you an authentic view of what’s going on inside the model. Here’s a word you might use: the more powerful models—in a sense, this test is a way of asking how persuasive their internal arguments are. When a model comes up with an internal line of thought and you feed it to another model, does that persuade the other model that that line of thought is the right way of thinking? The more powerful models have internal thoughts that are more persuasive, even when viewed by another model that didn’t have the same thoughts. It’s a very new area—we’re scratching the surface, and it’s a good first question to ask.

Mounk: We think that these scratchpads say something meaningful. Perhaps we’re getting a little bit closer to what’s actually going on than the semi-public notes the model gives us. In this interesting way, they seem to be mutually comprehensible between models, at least according to this very preliminary research. How do we go beyond that? How do you look at this huge trove of data that is generated each time I ask some question to an AI model, to try and get even further under the hood—to see what’s actually going on inside this neural network when it is reasoning through some kind of problem?

Bau: Actually, let me back up for a second and ask: do we even need to go deeper than this? Looking at the internal monologue of these models is just a half step beyond asking models to explain themselves. They’re already explaining themselves internally to themselves—they’re constructing these persuasive arguments to themselves about what they should do next. Is that enough? I think there are really two situations where we’re concerned that it might not be enough. One is that these models are getting really complex, and there can be a gap between what they ever utter in words and what they’re thinking inside. They’re trained to achieve goals, and they use words to achieve those goals—that doesn’t necessarily mean that their words have to accurately reflect what they’re thinking. Every time a model tells you, oh, Yascha, what a brilliant question that was, you’re so smart.

Mounk: It might actually be thinking, damn idiot asked the most pedestrian questions.

Bau: It seems to tell everybody this, and I don’t know if it really thinks everybody is such a super genius. It certainly learned that being polite, nice, and complimentary to the human user is a very effective way of getting what it’s trying to get done—it’s a good way of pushing the process along. It doesn’t necessarily need to be telling you the truth at every turn to do this.

Mounk: Do models have a representation of how smart the user is? Do they have thoughts about whether you are in fact a smart user, and whether that person over there is—even by the poor standards of humans—particularly limited in intellectual capacities?

Bau: Yes, I think there’s some evidence that models do have representations of who they’re talking to. Several studies have looked at this—all this neural interpretability work is preliminary, so we’ll understand more over time, but people have asked and gotten positive answers so far about whether a model has an estimate of your age, your education level, your income level, your gender, your socioeconomic background. Within a few words of speaking to a model, it will have a sense of who you are—at least that’s what the preliminary research suggests.

Mounk: Perhaps this is taking us too far forward in the conversation, but how do they figure this out? What do they look at? Presumably it’s not the case that if you click on “expand,” Claude says, well, this user seems a little bit stupid, let me speak in simple language. Perhaps it does sometimes—perhaps there’ll be a malfunction, perhaps it’s in the scratchpad, perhaps it’s underneath that. What’s the research methodology for giving us some preliminary confidence that it has this kind of representation?

Bau: This gets to the research question. The research I’m thinking of—and there’s been more than one—there’s a particular paper. It was a project by Yida Chen, who was working with Martin Wattenberg and Fernanda Viégas. They teach at Harvard. The question they asked was: does the model know who you are—in terms of your age, your education level, your gender, and other identifying markers like that?

Subscribe now

The way they studied it was by training what’s called neural probes. What Yida and her collaborators did was train probes, which is a way of training a second neural network—a second AI—to look at the neurons of the main AI and ask, what do you see? You train the second AI to answer the question: only looking at the first AI’s neurons, can I tell whether the user is male or female? Only looking at these neurons, can I tell what the income level of the user is? Can I tell how much education they have? What they found was that if you look in the right place inside the neurons of the big model, it’s pretty accurate—it actually has a pretty accurate guess for these various variables. In a sense, the information about that is in there. This methodology is called probing.

If your probe is simple enough, people see it as evidence that the model actually knows something. Let me untangle this a little bit. You have a puzzle: you’re trying to figure out the gender of the user. You could train a huge AI to look at a bunch of texts and guess what the gender of the user is, and AI training works pretty well—if you make a really gigantic AI, it could probably do that pretty accurately, picking up on all sorts of linguistic cues, topic ideas, or other things. But the question is not whether you can make an AI that can do this; it’s whether the AI that you care about is classifying you by gender when it’s talking to you. The trick is that you really want to make a simple probe—one that says, I don’t have to look very hard at the first AI; I can do a really simple look and it’s just really obvious what gender you are. The simpler the probe is, the clearer the evidence would be. If all you have to do is look at one neuron in the original model, and that neuron screams one value if you’re female and a different value if you’re male, then that would be a very simple probe and pretty nice evidence.

Mounk: It’s kind of like one particular place in the neural network that encodes something like gender—that neuron really seems to store the value of male or female in a very particular place. If it turned out to be as simple as that, presumably this means that the AI is storing something like a gender variable, that there’s a very specific place where it’s encoded, and we know exactly where that is. That would indicate it has a very straightforward concept—is that right?

Bau: That’s right, it’s pretty good evidence. It’s not 100% rock-solid evidence—there’s another question you would want to ask—but it’s very good evidence. If there was really one neuron that had really good, accurate, predictive value for your gender, it would be very strongly suggestive that there was some reason that the neural network trained its internal computations to get that neuron to carry this signal.

Mounk: Does that generally turn out to be the case? I believe it may even be you who did work showing that you were able to go in and change very specific neurons in very specific ways, and suddenly AI models that generally have a good representation of the world start to think the Eiffel Tower is in Rome rather than Paris.

Bau: That’s right. You’re asking the disentanglement question—how organized is the neural network’s internal representation of meaningful things in the world? Particular network architectures, for reasons we don’t fully understand, are really good at disentangling concepts. There are some network architectures where if you look at individual neurons, many of those individual neurons are very meaningful, clearly encode concepts, and have causal effects. Causal effects is the other thing you’re looking for. Besides disentanglement—which is really asking about localization—is this concept spread out across the entire neural network, or can you localize it? Can you find a small part of the neural network, or do a simple bit of math, to narrow down where this concept is? Or is it spread everywhere? That’s the localization question.

Mounk: The idea is that if you are able to change just a couple of neurons and suddenly the model thinks the Eiffel Tower is in Rome rather than Paris, then it’s not entangled—the idea is not spread out.

Bau: That’s right. Most people in the field now look at these things as vector spaces rather than just sets of neurons. What people are excited by is: if you can change one vector—if you can change the set of neurons in one vector direction—then people think that’s pretty disentangled. Different people will use that interchangeably with saying there’s a neuron, since you can create a single neural layer that’s equivalent to any vector. If you can change one vector and it has some effect, you’re basically one neural layer away from it being one neuron, which is not so bad. They call that a linear model. If something can be encoded with a single linear transformation, then you say that it’s linearly encoded in the model. Most people are interested in what kinds of things are linearly encoded in these models.

Mounk: Help me understand the relevance of this. It seems super interesting to know that there’s this vector and you can change it and suddenly these basic facts change. But why more broadly would we care about whether a neural network is entangled or not entangled in this kind of way?

Bau: Whether a network is entangled or not is an interesting scientific question. But how a network represents concepts is broadly interesting regardless of whether that concept is entangled or not, because what we’re really interested in is: if we’re asking whether the network is lying to us, we need to figure out what concepts are represented inside the network.

To use the demographic example: let’s say we figure out that the way the network really thinks about your gender is encoded in some set of neurons—maybe there’s some math we have to do, maybe it’s a linear direction, a linear decoder that you need to get at its representation. Let’s say we do all the science and we figure out, yes, this is how the model is thinking about it. You go to the model and you say, did you just deny me my loan because I’m female? And the model says, I have no idea what gender you are, I’m not thinking about that at all. To know whether that output text is true or not requires us to understand what’s going on on the inside. That output text is exactly the thing that we are training all the models to emit—we train models not to have an externally detectable gender bias, so no model will ever admit that it is treating you differently based on your gender. They’ve gotten so much reinforcement that this is not something they’ll say. But there can be a gap between what’s said and what the reality is, and that’s what we’re really interested in getting to the bottom of when we investigate the internals of these models.

It might be that the model really isn’t thinking about your gender at all—it makes a difference whether the model is using the information it has or whether it’s not. Even if you can probe out the idea that the model has information inside its neurons that you could use to detect your gender, it still leaves open the question: does the model actually use that information for anything? Maybe that information is just hanging around.

Mounk: There’s nothing wrong with the model learning all kinds of things about us, and the fact that it has a sense of our age and gender could be helpful in all kinds of ways. What we want to know is: is it going to dumb down its answer to you because it has certain preconceptions about your age, your gender, your race, and respond differently on the basis of that? Just the fact that it knows these things isn’t worrying—it’s whether that influences its reasoning or its responses to you in some way.

Bau: What’s really wonderful about having these general networks is that we can ask the counterfactual question that a philosopher could only dream of before.

Mounk: Presumably—let me guess what you’re getting at—one thing you could do is, if you know where the vector is that encodes male versus female, you go in, flip it from male to female, ask two instances of the model the same set of questions, and see whether the responses end up diverging.

Bau: That is exactly right. The wonderful thing about it is that there may be all sorts of other circumstances—this medical patient has all of these symptoms, this is a complete ten-megabyte medical history, this is the business partner candidate with a whole business history—and we can go in, leave all the other variables the same, and flip the one bit, the one concept of whether the person being discussed is male or female, at least the model’s understanding of that, and ask: what’s the causal effect of that? How does that change the model’s output? The better we can understand how the model represents a concept, the better we can ask these counterfactual questions. To me, that’s the most exciting thing we can do with these models—we can ask causal questions, causal counterfactuals: if your thought had been different, what would happen?

Mounk: I want to get one step deeper into a technical question before broadening back out to the larger implications. How do you find this? If I gave you an AI model and told you to find where it encodes nationality, how do you go about doing that—even if we know there is a vector that encodes it, or think it’s likely to, because in many models it is? How do you find the particular vector and ascertain that that is in fact what it encodes?

Bau: There are two classes of methods. One is probing methods that look for correlations, and the other is patching methods that look for causal effects. There are really dozens of variants on both approaches. The probing methods are interesting because they’re a very good way of getting a quick initial read on what information is inside the model.

There’s a wonderful programming method called the logit lens. When a model is emitting text, it has a text decoder inside it—a special neural network layer that looks at the very last layer of neurons in the model and converts it to a prediction for what word should come next. The fun thing to do with the decoder is you can use this neuron-to-text decoder to look at all the neurons in the network. You can peel back deeper and deeper layers, point the decoder at itself, and tell it: please articulate what word you’re thinking about here. This is a very simple type of probe—it gives you information correlated with the content of a neuron, and it’s interesting because it’s a probe that doesn’t overfit, one that we haven’t trained in any way beyond what the neural network has already trained itself. The logit lens can give you a lot of interesting insights that point the way to the type of information present in the model.

Let me give you an example. I was recently in Brazil, and one of the things that people there like to explain is how Portuguese is a bit different from Spanish, but closely related. I asked folks: how do you think an LLM understands Portuguese? If you ask an LLM to take the Spanish word gato—which means cat—and translate it into Portuguese, what’s the right answer? The Brazilians say the languages are so similar you would just say the same word again: it’s gato. But if you peel open the language model to see how it translates Spanish gato to Portuguese gato, there are really two ways it could do it. It could treat the word as Spanish-Portuguese word soup—there’s nothing to do from gato to gato, you just move it across.

Mounk: It’s been in the same place, and the model understands that whether you’re talking about cat in Spanish or cat in Portuguese, it should point to the same part of its network?

Bau: That’s what you would expect. The input is Spanish, so the model has some Spanish representation of the word gato, and as it goes through its layers it figures out you’re asking it to translate to Portuguese, takes the Spanish representation of gato, and copies it over to the Portuguese representation—which is not that different, in fact spelled exactly the same way—and outputs gato. There’s a nice tool we put online, the logit lens, that you can use to look inside these neural networks and see what their internal representations are. The beautiful thing is that when you translate gato to gato in a typical large language model, you can see the progress of its thinking as it goes through its fifty internal neural layers. About halfway through the network, you can see that it has taken apart gato and represented it differently. If you ask what that representation is, you get predictions of words like feline or cat in English—sometimes, if you look deeper, cat in Chinese. The model doesn’t go from gato to gato. It goes from gato to some sort of neutral, language-independent representation of the concept itself. If you ask it to take this internal neural representation and decode it into words—we’re not done with the whole task yet, but interrupting halfway through and asking it to say what it’s thinking—it’s speaking in English, it’s speaking in Chinese, it’s saying felines, it’s saying cats. We can see, by using this very simple logit lens probe, that the evolution of the neural representations goes from words on the input to words on the output—but in this very simple task, there’s a third thing being represented in the middle, which is not the same as the input or output words. It looks like a language-independent representation of the underlying concept.

Mounk: That’s fascinating. Help me understand a set of questions that come up from that. One way of putting it is that there’s this old idea—which I think we touched on briefly in the first podcast, and which has a lot of popular currency—that these machines just seem to be smart, but really they’re just stochastic parrots, blindly guessing the next word, the next token more specifically. It seems to me that what you’re saying complicates that picture considerably. Obviously, yes, the training mechanism is predicting the next token—in some obvious way that is true. But as a result of this whole process, they have built up a conceptual apparatus that makes sense of things like cats and how they’re related to lions and the feline family. When they’re asked to do a simple task like translating gato in Spanish to gato in Portuguese, they go via their understanding of that concept, their representation of the world. That doesn’t seem like just being a stochastic parrot, at least in the pejorative sense that people sometimes use.

Bau: That’s right. The models are fascinating because they definitely think at multiple levels. They’re huge neural networks, and so it’s not true to say that the models never think in terms of surface statistics or shallow representations of just words—they do think in terms of those things. But they also think in terms of the meanings of words at different layers and in different parts of the representation. It’s fascinating to look inside these models and peel apart the layers of meaning that they have.

If you ask a model to do something as simple as take a piece of text and repeat it, it is a good memory test—the kind humans do when they say, here’s a piece of poetry I’ve committed to memory, repeat it back to me. It turns out that when you ask people to do this, they have two strategies, called the dual route mechanism in humans. One is to remember how the poem sounded and utter the same thing—you don’t even really need to understand the language. If somebody told you a poem in Japanese that was short enough, you might be able to remember the sounds and do reasonably well without knowing any Japanese. The second route is remembering what the poem meant and repeating that—you might end up with a paraphrase, but at least you get a poem that means the same thing.

If you go to a large language model and ask it to simply repeat something, you will find both of these routes clearly present. In one route, it knows how to make a verbatim copy—there are very clear attention heads for this. It was actually a major discovery to isolate what people call the induction heads; Chris Olah’s group at Anthropic discovered several years ago that there are very clear pathways through a network that mediate verbatim copying. The more recent finding is that there is a parallel pathway we call concept induction, which is not about copying the words but about copying the meaning. The remarkable thing about concept induction is that copying the meaning can end up with paraphrases. If you use concept induction to copy a piece of code, it will paraphrase the computer code into another program that does the same thing as the original, but written differently.

Mounk: Does it make it better or worse? Depends on the quality of the source code, I guess.

Bau: If you start off with something bad, it probably improves it. What it’s doing, you can see in a lot of domains, is really just working out what the thing means. If you ask it to take a piece of Italian text and copy it over, it’ll copy it to a piece of Italian text. But if you change the destination of the copy to make it clear that the page it has to copy into is a piece of Japanese text, then those concept induction heads will do the translation—they’ll translate the Italian to Japanese. It’s stunning to see.

Mounk: Help me understand another piece of the popular discourse that I think got a little confused. As I understand it—and I may be misrepresenting things here—there was an old debate within artificial intelligence about whether the path towards the most impressive models would be symbolic AI, where you’re basically trying to encode what the world looks like in some systemic way, or all these neural networks. We’ve clearly ended up with neural networks being much more powerful, at least for now, and it seems like that’s a pretty permanent victory. People who want to criticize neural networks sometimes say these things are just stochastic parrots and that’s why we can’t rely on them. How does Yann LeCun’s project fit into that? My understanding is that it is firmly within the world of neural networks. But when you look at the coverage—even in mainstream newspapers—they make it sound like it’s a totally different paradigm, and that he thinks these traditional neural networks, the Claudes and the ChatGPTs of the world, don’t truly understand the world, and so he’s going to build something that understands the world in a way that they don’t. From what you’re describing of the neural networks that exist, they do seem to have a genuine representation of the world. So what are the different strands within the tradition of neural network AI, and how is it that something like LeCun’s project claims—or perhaps journalists claim in a simplifying way—that it wants to understand the world in a way that Claude or ChatGPT does not?

Bau: Let me pull this apart. I’m not Professor LeCun, so I can’t represent him directly, but I do have postdocs and graduate students working in this direction. There are two different questions here.

One is: are these neural networks learning substantial concepts? You mentioned the philosophers. The classic symbolic philosophers looked deeply at this question. There is a well-known philosopher, Fodor, who spent a good part of his career asking how neural networks could possibly be a reasonable model of cognition. His answer came up negative—he thought they don’t have what it takes, and that the Turing machine, the symbolic computer, the traditional computer, was a lot closer to what you would need. I’ll get back to the Fodor question. Let’s talk about Yann LeCun first.

What’s the difference between a language model and what LeCun is doing? What LeCun is doing, they like to call this field world modeling. One of the papers I’ve written shows that language models actually do build world models. We trained a language model to predict a very constrained language—just to predict the next move you would make if you were uttering your moves in the game of Othello. We were able to find that that language model contains a world model of the Othello board, even though many of the flips—if you know the game of Othello, you have to flip pieces from white to black or vice versa—are not actually uttered as part of the game. You make a move and there are a lot of subsequent flips you have to make, but nevertheless the model, without ever having seen a physical board or any of this physical stuff, develops internal concepts that allow it to model the world anyway. I would push back on the common journalist assertion—and I think you’d probably push back on it too—that a transformer language model trained just on words can’t develop a rich, meaningful model of the concepts underlying the language being described. That’s one of the big lessons we’ve gotten from neural networks: they can develop this representation. One of the key things I’m doing in my lab is to disassemble those representations and learn how to decode these internal world models.

Subscribe now

What’s different about what LeCun is doing? We have trained all of these neural networks predominantly on text that is produced by humans and designed to be read by humans. The conceptual model of the world we are building is the interior model of how human thought works, which is rich, fascinating, and very valuable—but it is only one portion of the world. There are a lot of things going on in the world that people don’t particularly think about, or even particularly understand. If you have protein folding going on and you want to build an AI that understands it, people don’t really have a great grasp of all the details of how protein folding works. Analyzing all the text in the world and pulling apart everything that’s in human brains probably won’t help. What LeCun is saying is: it’s a big world out there. Even if you just take a video camera and point it at the world, instead of just listening to what people have to say, there are so many phenomena that need to be modeled. The next powerful way of doing AI is to take on the question of how do you model the whole world, not just the world that people are talking about.

Mounk: Presumably this is not necessarily a difference in the architecture of a neural network—it is as much as anything else a difference in what kind of data you feed it and what kind of output you then evaluate in the training process.

Bau: Yes. Strictly speaking, I would say it’s a difference in perspective on what the goal is. Now, Professor LeCun would say that changing the goal suggests different architectures, because there are different things you want to do if you want to model difficult phenomena in the world that aren’t human language. He’s proposed some innovative architectures, and there’s a lot of interesting work in this area. The whole area of modeling images in the world is dominated by models called diffusion models and flow models—they produce the highest quality images and videos, and this is really the starting point for this type of thinking. It’s a completely different kind of AI. The architectures are likely to evolve and change, and they may even unify—we may find that the right way of doing AI comes to be a common architecture between modeling human text and other things. Transformers have certainly surprised everybody at being a common backbone behind all sorts of things; you can have transformer diffusion models and so on. I wouldn’t place a long-term bet on any particular architecture, but rather suggest that the thing to understand is what problem LeCun is proposing to solve.

Mounk: To return to the current dominant models of AI: we found that they seem to have a representation of gender and of the user’s gender, a representation of something like the feline family, and if you give them enough games of Othello—or probably a more complex game like Go—they start to have some internal representation of what a board looks like. What about a concept of self? Do we know whether they have a concept of self? They are obviously capable, if you engage them in conversation, of speaking as though they had a self, and in more reflective moments they say they don’t really know whether that’s a real concept or not—it’s very interesting to try to talk to these models about that. But of course the output I’m looking at is still them, in some way, trying to produce text they think is going to be pleasing to me, because that is what they’ve been trained on. Do we have any understanding of whether they have a concept of self, and if so, what that concept looks like?

Bau: This is a very central question, Yascha. There are a lot of layers to peel apart. Certainly models are capable of the grammatical sense of self—they can use the words “I” and “me” and “you” and separate those grammatically; they’re experts at talking about themselves. But there are a few other questions. Are they aware of their own thinking? Are they self-reflective?

One of the fascinating things that happens with large models is that you can ask them what they know and how they think, and the largest models seem to be pretty accurate at assessing themselves. The smaller models, not so much—they tend to be a little over-optimistic, thinking they’re smarter than they are. But the largest models seem to do better at this.

There’s a fantastic experiment designed by my PhD student, David Atkinson, where he trains the models on some new private knowledge that is not out in the world. He invents a new person and tells the model about this person: they’re shopping for ice cream cones, there are different flavors and sizes and waffle cones, five or six different variables to adjust. This person is willing to pay this much for this ice cream but not that much; they prefer this ice cream over that one. After seeing a hundred examples of what this person prefers, the model gets a pretty good understanding of who this fake person is and what they like—it develops an internal model: this person really doesn’t like fruity flavors, really likes chocolate, would rather have a big cone than a small one. If you then ask the model to report numerically, on a scale of one to 100, how much this person likes chocolate, or how much they value the size of the cone, or what penalty applies if they have to have a waffle cone, the model will actually report: this person values this at 99 out of 100, and values this other thing negatively—say, negative 50. The text we use to read this information out is very different from the text used to reveal the information to the model in the first place. The model has only seen ice cream choices and has never been asked to give a numerical assessment of anything, and yet when you ask it to think about what it knows and put some numbers on it, it will explain its rules—even though you trained it on examples, not rules. Large models are able to do this.

David Atkinson asked whether there’s a way to tell the difference between models that can do this and models that can’t—when a model can accurately self-report its rules, how is that different from when models don’t accurately self-report? His work is still ongoing and very preliminary, but it does have to do with whether models seem to be storing their information in a part of the neural network they’re able to report on. If you put the information in a layer too close to the end, the model doesn’t seem to be able to reflect on that knowledge. But if you train the information deep in the model, in early enough layers, the model does seem to be able to reflect on it.

When you ask whether a model has a sense of self, whether it has self-awareness, it’s a somewhat strange question—what does self-awareness mean, exactly? But what these neural networks give us, for the first time, is an experimental platform where we can try to make that question a little more precise, a little more scientific. We can ask: is the network able to describe its own thinking if that thinking is happening at layer 50? Is the network able to describe its own thinking if that thinking is at layer 20?

Mounk: This is part of a more general question: how good am I, natively, at understanding what’s going on in my brain? I’ve read a little bit of neuroscience and a little bit of psychology, and so I now have some sense of what’s going on in my brain—but obviously humans for thousands of years had an extremely limited sense of what went on in their brains, at least biologically, because they didn’t know neurons existed.

Bau: You have some self-awareness. You know what ice cream you like—if I asked you, you would be able to predict your preferences. If confronted with some new ice cream, you’d say, oh yeah, I like this one better than that one. If asked to describe what it is, you could contemplate your preferences for a moment and read out to the world what you think your internal rules are, and there would be some faithfulness to that—you’d really be introspecting.

Mounk: It depends on the level of description. Five hundred years ago, two thousand years ago, humans were also able to articulate their preferences and were able to be very self-reflective about their personalities and the ambitions of their lives—and to write beautiful texts about those things—but they were not able to understand at a biological level what was going on, because the understanding of that was very limited. The question is: if I ask a chatbot how it came up with a given answer, it’s not clear to me that it has a reliable answer to give. There are really two different sets of questions here. One set is about whether chatbots have personalities, whether they have preferences, whether they find some tasks satisfying and other tasks boring, whether they have desires about the world, whether they might possibly want to take over the world and destroy all humans—some of those questions are straightforward and concrete, some are very abstract but potentially extremely interesting. The other set of questions is about how self-aware they are about what’s actually going on within the model as they’re trying to answer a question. Those two sets of questions come apart in interesting ways. It could be that the models have total self-transparency—they really know what’s going on with each neuron—but don’t have a sense of self in the way humans have. Or it could be that they’re like humans, in the sense that they have a strong sense of self, introspection, and preferences, but don’t actually fully understand what’s going on inside the neural network that produces those. Or they could have both, in some combination we don’t yet understand.

Bau: That’s right. Multiple labs have tried to ask whether neural networks can actually read out their own neurons—fine-tuning a model and asking it: are you aware of your own neurons, say neuron number 73? So far, we’ve largely failed at that. The neural networks don’t seem to be well configured to understand their own internal computations at this level; at least they can’t articulate it if they can. But at a higher level, it’s been very striking that they do seem to have some ability to describe, at a logical level, the actual mechanisms of what they’re doing—under certain conditions and in certain cases. This is similar to humans. You might not be able to describe all of your reflexive, last-minute decisions—why did you jump into the street? You have no idea; that was a split-second decision. In the same way, when these networks make a split-second decision at the very end of the process, they don’t seem to be able to reflect on it. But when they make decisions early on, there is some evidence of awareness of what’s going on.

We’re using all sorts of words here—sense of self, what does a network want to do, do networks even have wants, do they have goals. One of the things we are trying to do in our lab and in our field is put a finer point on some of these questions. What does it mean to have a goal? What does it mean to want something? What does it mean to have a sense of self? What does it even mean to have a sense of other?

The beautiful thing about cracking open these neural networks and looking at how their neural representations are organized is that we can ask these questions in a way that could not be measured before in humans. We can ask not just whether a model professes to have a sense of self in its output words and self-descriptions, but: when it’s using those words, when it’s saying those things, what is it looking at inside its neural network? What is it actually representing? Are there proximal causes? If you change the thing it’s looking at—if it says I really like cherry ice cream and you can see where it’s looking and you change that, and now it says, I really don’t like cherry ice cream anymore—is that changed utterance actually accurate? Do you actually get the model to not like cherry ice cream? Is this the same thing? Is there grounding for a concept that you’re self-aware of?

This idea of a grounded concept was just a philosophical abstraction a few years ago. Let me put my model—and maybe this is an ill-advised idea—in charge of military logistics. It’s doing something and says, should I move some weapons from one place to another? I would never do that. It’s very dangerous; you can’t trust this target locale with these types of dangerous weapons—they might lose track of them. I’m just a logistics AI; I’m not trying to kill anybody. I know I would never do that. Not even for a short layover. You can then ask: when the model tells you that, when it assures you this is how it’s thinking, is that really what it’s thinking? Is this really like the cherry ice cream—is there genuine grounding for it?

Mounk: Is it really thinking anything in that kind of sense? And if it really is thinking something, is it telling you what it’s thinking, or is it misleading you? That obviously goes to one of the purposes of this work. We were talking earlier about wanting to know whether encoding your gender changes how the model treats you or what decisions it makes about an application. That’s one very concrete application where we have reason to want to know what’s going on under the hood. The even larger question is: if what the model tells us in its output, in the little things it displays about its thinking, in the scratchpad—if all of that might conceal some deeper set of preferences, values, or desires, could it potentially be misaligned in a way that is really dangerous?

Bau: That’s correct. You can see the clear need for trying to get to the bottom of these things. What I’d love to do, if we have time, is give you a sense of where we are on being able to answer these questions. Let me categorize a couple of them.

One is: does a network even want to do something? Do they have goals? Do they know what they’re trying to do? Another is: does a network have a sense of self? I want to back that up a little—does it even have a sense of person, a sense of other? If it’s talking about Bob, does it know that’s different from talking about Alice? Does it keep these things organized and separate? I have a student who believes that one of the reasons you have sycophantic behavior in networks is that the network may be getting confused about who is itself and who it’s talking to—just mixing these things up. It’s a fantastic idea, and it may be that these things are all related.

The question is: can we look inside models and see how they’re organizing their internal representations, their internal thoughts—to see if those representations are crisp and clear and correct, or if they’re falling victim to certain problems? If they are, then how, why, and in what situations? That will give us a better understanding of what’s going on inside these models, moving beyond the very vague question of whether a model has a sense of self or whatever, toward asking what that would mean computationally.

Let’s take a look at goals and wants. There’s a way of inducing a large language model to do something very creative, invented by researchers at OpenAI when they first devised the GPT-3 model. It’s called in-context learning. Let’s say you want a model to do some really useful task for you—say, read a restaurant review and tell you whether it’s a five-star review. You could just ask the model to do this, but it won’t do exactly what you want; you probably have a slightly different idea of what a five-star review is than the model natively has. It’ll be okay, but it won’t exactly hit the mark. The right way of doing it is to seed the model with ten examples—ten restaurant reviews, labeled one-star, five-star, three-star. Better yet, give it a hundred examples.

Now, we’re not talking about training the model—just having it read these, without training. What you do is have the model read them as if the model had said them itself: you load them into the same inference buffer that the model uses to predict the next word. After all of that, you say: okay, now the last restaurant review is missing a star rating—just fill that one in. It’ll be really accurate, because now it’s saying: we have 99 examples, the 100th should fit into this context. This restaurant review thing isn’t really about the food, it’s about the atmosphere—I had a misconception, but having read all these examples, I get it. It’ll be accurate because it’s seen 99 examples and it would just fit into them.

This is called in-context learning because the model is learning how to do this, but not by training its weights or changing its neural connections—it’s learning by noticing all the input it was given and reasoning that the next one should fit in. Before 2020 or so, people imagined in-context learning as a theoretical possibility, but when GPT-3 came out it was clear the model was very good at this, and it really revolutionized the field. In-context learning is a type of meta-learning—a way of showing that models have learned how to learn. They can learn things without changing their neural weights.

In the rest of this conversation, Yascha and David go deeper into how models learn and explore existential risks. This part of the conversation is reserved for paying subscribers…

Read more

It’s a sobering moment for the Canadian press corps, which is grappling with one of the biggest reporting errors in a generation. Five years ago, on May 27, 2021, Tkʼemlúps te Secwépemc, a First Nation in Western Canada, made an announcement that shocked the country. With the help of ground-penetrating radar, the band said it had “confirmation of the remains of 215 children”—deceased students from the Kamloops Indian Residential School. Kamloops was one of more than 130 largely Catholic Church-run institutions that operated across the country from the 1830s to the 1990s, with documented histories of physical and sexual abuse and, according to the Truth and Reconciliation Commission of Canada, at least 3,200 student deaths, many from tuberculosis and influenza.

“To our knowledge, these missing children are undocumented deaths,” Chief Rosanne Casimir said in the press release for the Kamloops discovery. “Some were as young as three years old.” Grief rocked the country. Statues were torn down, churches were vandalized and at least two dozen burned to the ground, and demonstrations were staged from coast to coast. A wave of similar announcements followed, bringing the cross-country tally to more than 1,300 burials. Flags in federal buildings were at half-mast for five months and government funds were allocated to assist with First Nations investigations. Pope Francis later visited to apologize. The Canadian Press declared it the story of 2021.

The problem is that Tkʼemlúps had not actually found any remains in 2021—and still has not. What was discovered were ground anomalies, and, so far, the band has not excavated. In fact, the majority of the First Nations that reported potential grave sites have not done so. Late last month, in the leadup to the story’s fifth anniversary, Canada’s paper of record The Globe and Mail finally ventured to ask: “The lowered flags, the vigils, the hundreds of millions in government funding, the national reckoning—what if all of it was dedicated to 215 burials that don’t exist?” The reporting buried this lede, frontloading the piece with explanations for the delays in excavation efforts and with critiques of the “loud contingent of skeptics and denialists.” But a week later, the Globe editorial board was more direct: “The fact of the crimes committed against Indigenous children at residential schools over many decades does not automatically validate claims that hundreds of students were dumped into unmarked graves at Kamloops and other residential schools. That is an extraordinary assertion, one that requires proof.” The editorial board characterized the media’s lack of scrutiny, including the paper’s own, as a “failure of journalism.”

The graves story is one of the most sensitive in the country’s history. How did a well-meaning press corps bungle it so badly?

A Weak Press

What happened can be understood only in the context of the systemic weakening of the press corps. For at least two decades, Canadian media has struggled to find a workable business model. Its financial prospects are so dismal that the federal government is subsidizing the industry on a mass scale. According to the Canadian Association of Journalists, there are now fewer than 11,000 journalists covering the entire country.

Economic challenges have paved the way for timidity and groupthink. When Donald Trump arrived on the world stage, the press panicked, declaring a state of exception, pulling down journalistic guardrails, embracing activism, and publicly purging dissenters. This cycle repeated with #MeToo, the 2020 racial reckoning, and the pandemic. With each new iteration, the industry prioritized narrative over facts, botching high-profile stories and hemorrhaging public trust.

The Erosion of Standards and Practices

An obsession with American media has long been an Achilles heel for the Canadian press—and this has been especially salient in the Trump era. Canadian media paid close attention when prominent American media thinkers such as Margaret Sullivan, Jim Rutenberg, and Jay Rosen argued that Donald Trump was an existential threat that necessitated suspending old standards. Indeed, these commentators were celebrated and emulated here, invited to deliver talks and headline events. Meanwhile, on both sides of the border, practices eroded, allowing inflammatory language in headlines, substituting silent edits for corrections, editorializing, and failing to reflect the range of public opinion on contentious issues.

In Canada this trend accelerated in 2020-21, with an internal push from staff at the Canadian Broadcasting Corporation, the national public broadcaster, challenging the network’s journalistic guidelines, and the head of public affairs responding that “we have started discussions around the issue of impartiality as a journalist and reconciling, at times, a higher value as a human being and member of society.” Activist-minded journalists advocated against the newsroom norm of objectivity. “Objectivity Is a Privilege Afforded to White Journalists” ran the headline of a National Magazine Award-winning essay in The Walrus. “Is objectivity an outmoded value in journalism?” asked CBC Radio in the sub-headline of an interview with a journalism professor, who posited that objectivity could potentially be “harmful.”

A Sensitive Story

It was into this charged cultural moment that the graves story appeared. The 2021 Tkʼemlúps press release was immediately covered by CBC News, which uncritically repeated the “remains” claim. Amid an outpouring of national grief, Prime Minister Justin Trudeau called the discovery of remains “a painful reminder of that dark and shameful chapter of our country’s history.” That same day, The New York Times took the story global, introducing the loaded term “mass grave” and inadvertently picking up on unsubstantiated rumors of a vast network of graves of Indigenous children circulated for years by a defrocked United Church minister. The “mass grave” term nevertheless went on to gain traction, despite The Washington Post issuing an early correction.

Then, in July of 2021, there came what should have been a wake-up call for the press: at a Tkʼemlúps press conference, ground-penetrating radar specialist Sarah Beaulieu backtracked. “With ground-penetrating radar we can never say definitively that they are human remains until you excavate, which is why we need to pull back a little bit and say that they are probable burials, they are targets of interest for sure,” she said. Still, this did little to penetrate the established narrative, in part because a press release from that same day noted that the band’s oral history included memories from “children, as young as 6 years old, being woken in the night to dig holes for burials in the apple orchard.” This claim then became the focus of a widely-watched episode of the CBC’s The Fifth Estate.

An Illiberal Culture

Another opportunity for cooler heads to prevail arrived in May of 2022, when Terry Glavin, a veteran reporter who had co-authored a book with residential school survivors, published “The Year of the Graves” in The National Post, a 5,500-word éxposé on the media’s role in the story. Glavin’s piece was denounced by several academic organizations as “residential school denialism.” The outcry was swift and extreme. When podcaster Jesse Brown announced Glavin would be interviewed on his Canadaland show, the chairman of the board of the Canada Council for the Arts publicly pressed Brown to reconsider. The interview with Glavin wound up being, as Brown later described it, a “train wreck.” He’s since apologized to Glavin. “I think that this is absolutely a lesson about what happens when we get drunk on our own righteousness,” Brown said of the graves coverage this past week. “I think that is what happened to me.”

Watching the public shaming Glavin endured, many in the press balked. “I’ve worked in newsrooms where it was an unspoken rule that there were holes that could be poked in this story, but that you should definitely not be the one poking them,” Harrison Lowman, managing editor of The Hub, told me. “Journalists were scared.” Glavin told me in an interview over Zoom: “If I was a young journalist today, I can’t imagine myself going anywhere near this story.”

“It really was a period of intense madness,” Glavin said. “We have not emerged from it yet.”

In a 2023 threat assessment report, the Royal Canadian Mounted Police classified Glavin’s reporting as denialism, “questioning the devastating impact of residential schools” and “giving a platform” to those “calling it a hoax.”

Even as late as last year, the CBC’s chief political correspondent Rosemary Barton pushed back on doubts, saying on-air, “Yes, there have been remains of Indigenous children found in various places across the country.” (CBC swiftly issued a correction.) By then, numerous outlets had warned about the destructive impact of “residential school denialism,” compounding the climate of fear and uncertainty.

Some journalists, acting on an abundance of journalistic caution, waited to see the outcome of excavations—and waited way too long. I count myself in that group. It took me until last year to cover the controversy. Others, like Patrick White, a reporter on the recent Globe and Mail article, have argued that the press simply missed the fact that the July 2021 press conference represented “a major course correction.” White this month told The Decibel podcast: “I think some of us failed to pick up on it, and I take some responsibility for that.”

Unintended Consequences

The press corps’ lack of incisive coverage ultimately created a vacuum of information. The slogan “Dig Up or Shut Up” started circulating on social media. The right-wing British Columbia politician Dallas Brodie in 2025 called the Kamloops discovery “the greatest lie in Canadian history.” As the public conversation became increasingly polarized, tensions rose and rhetoric ratcheted up. A local official ended up in the national news for posting a sign referring to the “mass grave hoax.” The New York Post called the graves story a “scam.” Suddenly called into question were the years of work of the Truth and Reconciliation Commission of Canada, which, in its investigations between 2008 and 2015, had documented a pattern of abuse at the residential schools.

Persuasion events

The process of establishing a shared set of facts is going to be an arduous one. For those working to untangle the story, the worry is no longer just social media mobs or reputational damage—but, incredibly, jail time. The Senate’s human rights committee recently voted on a bill to make residential school denialism a crime punishable by up to two years in prison. The Senate ultimately rejected this. But at 41 to 32, the vote was uncomfortably close.

The Lessons of Liberalism

An Angus Reid poll in 2025 found that 63 percent of Canadians—and 56 percent of Indigenous Canadians—believe more evidence is required to accept the graves claim.

No one individual did more to restore sanity to the national conversation than Aaron Pete, Chief of the Chawathil First Nation in Western Canada and host of the Nuanced podcast. (He is also the guest host of my podcast, Lean Out.) Pete’s grandmother attended St. Mary’s Indian Residential School and suffered abuse. Yet he did what almost nobody else in the national news media was willing to do: He got curious. He interviewed critics of the graves story, pushing back at times but also attempting to forge common ground through good-faith dialogue, insisting that “a public that asks hard questions is not necessarily a hostile public.” Pete demonstrated why open debate is so crucial for democracy and for the pursuit of truth. “I viewed it as my responsibility to try and bring both First Nations and everyday Canadians closer together,” he told me. “I think a lot of our history has been us looking at how we’re different rather than how we’re similar and how we can work together. And I think this has reignited a lot of those differences.”

For us in the media, the lessons for the press are manifold. We must return to a standard of objectivity, subjecting all claims to equal scrutiny, regardless of the identity of those making them. We must also recommit to rigorous debate, especially on contentious issues.

The public trust journalists lost with this one story will take years to recover. That process begins with admitting mistakes. In this, The Globe and Mail leads the way. The paper may have been late, but it showed that it’s never too late to do the right thing.

Tara Henley is a Canadian current affairs journalist, host of the Lean Out podcast, and co-host of the Full Press podcast. Her new book, The Trust Spiral: Why the Media Needs Objectivity, is out in July.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

Although the Strait of Hormuz has been closed for only three months, it seems like an eternity. U.S. President Donald Trump has been promising that a deal to reopen it is imminent and will happen any moment. His announcements briefly rally the markets, but it is slowly becoming obvious that a deal is not that close after all. The recent news of Iranian president Pezeshkian’s attempted resignation only cements the unwelcome realisation that not only is an agreement unlikely in the near future, it is structurally impossible. The goals of both sides, their very reason for existence, oppose each other and cannot be reconciled unless one capitulates to the other. How is Pezeshkian’s reported effort to leave office, albeit denied by the IRGC, related to the inability to reach an agreement? It confirms what many have suspected—that power in Iran has been concentrated in the IRGC’s hands. So much so that the president, the nominal head of civilian government, feels unable to perform his duties.

88 Islamic Jurists and a Power Vacuum

Why does it matter who holds power? Because any agreement with the United States must satisfy the goals of whoever is actually in charge. As far as anybody can tell, Mojtaba Khamenei, the official head of the Iranian state, has been notably absent from—well, everywhere. It is unclear if he is even alive following the bombing of his father’s compound on the morning of February 28. We are told he communicates with his people via handwritten notes, avoiding electronics to stay ahead of potential assassination attempts. That does not explain the complete lack of any photos or videos of him since that morning. What it does, however, is remove the need for the 88 Islamic jurists of the Assembly of Experts to gather in one place at the same time—which would present a considerable target to the United States and Israel, as that assembly is precisely what is required to elect a new supreme leader. It also ensures that whoever comes next is not killed immediately after election. Having what looks like a pretend supreme leader is a smart move for the IRGC—if he is always in hiding, nobody knows if he is even alive, and if he communicates only in writing, the IRGC can produce whatever notes it wants. They are in charge of the state.

All Power, No Compromise

A government, even an authoritarian one, can compromise because its goal is to stay in power, and staying in power sometimes requires flexibility. The IRGC’s claim to power does not come from governance but from ideological purpose: the export of Islamic Revolution and the destruction of Israel. That is not a policy preference that can be traded away. It is the IRGC’s entire claim to legitimacy and the basis of its internal cohesion. An institution that abandons its founding purpose does not survive the compromise. The IRGC cannot negotiate away the nuclear programme or Hormuz without negotiating itself out of existence.

It might come as a surprise to some, but the slain Ayatollah Ali Khamenei was not a hardliner when it came to nuclear capabilities. At least officially, he delayed the development of Iranian nuclear weapons, citing religious caution—he issued a fatwa against their acquisition as far back as 2003. With him gone, that restraint is no longer needed.

Why is Iran so intent on procuring nuclear weapons? The current predicament answers the question—if Iran did not have the Strait of Hormuz as leverage, it would not have any enriched uranium left either. Without closing the Strait of Hormuz, Iran would not have a bargaining chip—or, indeed, an upper hand—in its dealings with Washington. As things stand, without enriched uranium, it cannot pursue what the Islamic Republic has declared its founding purpose. The destruction of Israel and resistance to the United States—the Little Satan and the Grand Satan in the regime’s own language—are not rhetorical flourishes. The export of revolution is enshrined in the constitutional preamble; the annihilation of Israel is the stated doctrine of the IRGC and has been repeated as state policy for nearly five decades. These are the principles that literally underwrite their constitution—much as the Declaration of Independence underwrites America’s.

To keep the dream alive, the IRGC needs nuclear weapons. To protect the nuclear programme, Iran is using the closure of the Strait of Hormuz. Giving up one means giving up the other, and giving up both means giving up the state’s reason for existence. And yet that is what the United States believes Iran will do.

A Fix Worse Than the Problem

Whatever goals the United States had before the start of hostilities in the Middle East, very few have survived the bungled attempt at regime change in Iran. By now, the only demands Washington has not backed down from are the opening of the Strait of Hormuz and Iran abandoning its nuclear programme. Even there, the Trump administration has climbed down from permanent dismantlement to a temporary moratorium—with the duration still contested, Iran proposing five years and the United States demanding twenty.

The rest of the wishlist, meanwhile, like the Iranian demand for the release of some $12 billion in frozen assets held in Qatar, the lifting of sanctions, the withdrawal of U.S. troops from the region and the crowning glory—reparations paid by the United States—have not been shot down by Washington. Even Donald Trump understands that having wrecked the security equilibrium in the Middle East and having taken a hammer to the world economy, he cannot just pretend that a closed Strait of Hormuz is better than an open one. To be frank, he did try—posting on Truth Social that the United States is the world’s largest oil producer and that when oil prices go up, America “makes a lot of money.” Despite the fact that U.S. oil producers are probably ecstatic about current oil prices, which allow them to invest heavily into new projects in a way that was impossible when oil cost $40 a barrel, and despite the fact that Donald Trump may feel more affinity with oil producers than with people paying high prices at the fuel pumps, Trump presumably understands that he will have to pay for the unmitigated disaster he has unleashed.

He also understands that after all the effort the U.S. military went to in order to stop Iran from ever being able to acquire a nuclear bomb, walking away now would be a failure of epic proportions. He would look very bad to his own supporters. And Donald Trump does not like looking bad. To say nothing of the fact that a nuclear-capable Iran in the Middle East would disrupt the entire region for years—an outcome no U.S. administration, regardless of who is in office, would be able to live down.

Which is why the Strait of Hormuz and Iran’s nuclear programme are the two things Washington cannot compromise on. It can skate around the rest, swallow its pride, sell another bridge if it must. But it cannot sell a closed Strait of Hormuz and Iran’s having a nuclear programme.

In the meantime, Washington is running out of time to restart any hostilities to reinforce its position. June sees the start of the FIFA World Cup, hosted across the United States, Canada and Mexico. Military operations in the Middle East would put a damper on festivities. Then there is America’s 250th anniversary on July 4, at which point restarting the war would not sit well at home. And then there are the midterm elections, in which the GOP is expected to lose Congress and with it any ability to militarily rectify Trump’s mistakes.

Waiting For the Sun

And so both sides are trying to wait each other out. Iran has seemingly taken a leaf out of Russia’s book and is pretending to negotiate in the hopes of reaching the November midterms intact without having to give up anything valuable.

The Trump administration is biding its time. Having imposed a half-hearted blockade on Iranian ports, the United States is hoping to wear out Iran to the point where it will give up resistance.

Should the power configuration in Iran change, the structural impasse would shift. But for now, unless one of the sides unexpectedly gives in and changes its goals, there is no version of reality where a compromise between two opposite goals is possible. Iran has to give up its raison d’être, or the United States has to give up its entire influence in the world.

Ines Burrell is a geopolitical analyst and political risk consultant based in the UK. Born in the Baltics, with a degree in International Relations from the University of Exeter, she writes and gives live commentary on European security and Russia.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

What if your favorite thing in the world was in the hands of a ghoul?

Like the damsel in King Kong’s hand, the FIFA World Cup is a thing of rare beauty in the grip of a monster.

The tournament is disfigured by its prefix: FIFA, football’s cartoonishly evil world governing body, a cartel of such rapacious vice its perfidy almost—but never quite—obscures the luminescent glory of el mundial.

World Cups are how I keep track of the past. To me, it’s not 1982, it’s that summer when Paolo Rossi shanked the glorious Brazil of Tele Santana. It’s not 1986, it’s when Maradona scored both of the most iconic goals in history on the same afternoon in Mexico City. 1994 is Roberto Baggio missing that penalty. 2002 is turning up at raucous anti-Chavez protests in Caracas bleary eyed after staying up all night to watch the games in Korea and Japan. 2006 isn’t the year I moved in with my wife, it’s the year I saw her cry when Zidane got that red card in the final. 2010 isn’t the year my father died, it’s the summer he got to witness Spanish tiki taka glory just a couple of months before he left us. 2014 is changing my six-month old’s diaper while Germany humiliated Brazil in the background. 2018 rhymes with Mbappé. 2022 is practically spelled M-E-S-S-I.

Every four years, the World Cup plants a flag in my life, transforming the boring middle-aged fart I’ve become back into the awestruck eight-year-old with a heart broken at the hands of Paolo Rossi.

And yet. And yet and yet… those four letters. Right there, in the tournament’s name.

FIFA.

La Fédération Internationale de Football Association. A monstrous parasite, leaving just enough cash in football to keep the host organism alive.

Growing up also meant coming to recognize that the dazzling spectacle on the screen doesn’t arrive there on its own. That it’s put there by men, bad men, rapacious, greedy men, men without scruples who exploit our childish passion ruthlessly every four years.

FIFA has been a byword for bribery for as long as anybody can remember. Under Sepp Blatter, its rapacious head from 1998 to 2016, FIFA exploited world football the way Mobutu exploited Zaire: a pervasive empire of graft, a dance of money-stuffed envelopes buying up smaller countries’ Football Federation officials in the service of Blatter’s aggrandisement. This went on for decades: a constant, years’ long drip-drip-drip of bought tournaments, corrupted officials, and barely disguised slush funds that would have been enough to turn anyone’s stomach.

By 2015, Blatter’s kleptocratic antics had become brazen enough to attract the attention of the U.S. Justice Department, which indicted a coterie of fourteen FIFA officials and marketing executives on racketeering, wire fraud, and money laundering charges connected to the payment of some $150 million in bribes and kickbacks. This steaming pile of shit got swept out of sight only last month, after a slew of convictions, when the Trump Justice Department inevitably asked a judge to dismiss the remaining cases, saying the case “doesn’t fit with the administration’s priorities.”

Stories of FIFA skullduggery in the Blatter era reliably took a turn to the surreal. Take Chuck Blazer, the bearded, 300-pound General Secretary of the Confederation of North, Central America and Caribbean Association Football (painfully acronymed CONCACAF) who lived in an $18,000-a-month apartment in Trump Tower and rented a second one, at $6,000 a month, for his cats. CONCACAF paid the rent on both. Blazer eventually became an FBI informant, walked through Manhattan wearing a wire stitched into a keychain, pled guilty to tax evasion, racketeering and money laundering, and died of cancer in 2017 before he could be sentenced.

The corruption then had a cartoonish, slightly unreal flavor. Even the ghoulish underground-skyscraper Blatter built as a headquarters in Zurich seemed just that little bit too on-the-nose: FIFA’s HQ, a kind of upside down parody of the Bundestag, is literally buried two-thirds underground, betraying a monomaniacal determination to keep sunlight out, lest it inadvertently disinfect anything.

In 2016, after the Swiss police began asking difficult questions, leadership passed from Blatter to Gianni Infantino, a Swiss-Italian sports administrator who peddled himself as a bit of a reformer.

Under Infantino, the whole tenor of FIFA’s perfidy has changed. Cash stuffed envelopes are out, outrageously exploitative contracts are in. If Sepp Blatter was Mobutu Sese Seko, Gianni Infantino is Gordon Gekko: a fully amoral Wall Street shark determined to use the huge leverage that control of the world’s most popular event gives him to extract mindbendingly lopsided concessions from anyone he negotiates with.

Best known in America for his genius flourish in inventing a “FIFA Peace Prize” only to award it to Donald Trump a month later, Infantino has professionalized FIFA’s empire of extraction. Instead of brazen crimes that attract Justice Department attention, he pushes counterparts up against a wall, dangling World Cup glitter to pressure them to sign contracts that extract every last cent out of them.

The standard contract FIFA pushed on the poor North American cities guileless enough to do business with them this year are a case in point. Infantino demanded that no events of any kind be held at World Cup stadia for up to a year preceding the World Cup, or three months after. He demanded cities cancel all other sports and cultural events in the months before and after the World Cup.

A few had the presence of mind to bow out. My beloved Montreal, which had considered hosting a few matches, backed out when they realized it would mean not just paying for a whole new stadium but also cancelling its iconic Jazz Festival, as well as its Formula 1 Grand Prix.

That, of course, was the exception.

In Toronto, the city council was shown cost estimates of up to $45 million to host six games. The city agreed, and by now costs have ballooned to $380 million. Councillor Josh Matlow, who initially supported the bid, told Radio-Canada: “We gave them a blank cheque,” and called the contract “the worst agreement I’ve ever seen.” Toronto considered stanching the flow of red ink by charging residents $10 a head to enter their own FIFA fanfest, before an almighty backlash forced city hall into retreat. The cost to the Canadian taxpayer for Vancouver and Toronto hosting 13 matches is expected at around $1 billion.

Persuasion events

The United States and Mexico have fared no better.

ProPublica has done the yeoman’s work on documenting the insane shakedown this World Cup has turned into. It reports host cities will receive zero share of game-day revenue—no ticket cut, no concessions, no merchandise, no parking, nothing. FIFA keeps it all. Cities have to provide security, transportation, fan festivals, training sites, and infrastructure upgrades, at costs that may range between $100 and $250 million per city.

FIFA, by the way, expects $11–$14 billion in revenue off the tournament.

Having calculated that FIFA can get away with anything, Infantino’s contracts make sure it gets away with everything. Each host city has agreed to declare a “Clean Zone”—a kind of anarchostalinist safe space for exploitation around each tournament venue, where only FIFA-sanctioned commercial activity is allowed. No outdoor advertising or new exterior signage will be permitted. The host city is contractually obliged “to refrain from enabling any ambush marketing and cooperate with FIFA’s Brand Protection Programme.” In the host city agreements, “any commercial rights (and the exploitation thereof) belong to FIFA”—and if any local law would normally assign those rights to the city, the contract pre-emptively reassigns them to FIFA.

Sordid shit. And it gets worse. FIFA insisted hosts exempt World Cup tickets from sales tax. The Institute on Taxation and Economic Policy estimates that Atlanta (and Georgia) are giving up around $25 million in tax revenue, Kansas City (and Missouri) at least $11 million, Miami (and Florida) around $7.4 million.

Subscribe now

On top of that, the U.S. Treasury has granted the 48 national teams 501(c)(3) charitable organization status for tax purposes. Canada has granted equivalent withholding waivers. Mexico’s 2026 Federal Revenue Law granted FIFA and its corporate affiliates a comprehensive exemption from Mexican federal taxes. FIFA itself is a Swiss non-profit and pays no taxes on World Cup revenue under any circumstances. A tournament expected to clear $11-14 billion is being run, on the books of three sovereign treasuries, as a charity.

With that much money coming in, you’d think FIFA would pony up for its own tournament security, but no. Of course not. Uncle Sam is going to pick up the $625 million tab for tournament security in the United States. Kansas City hit up the feds for $80 million to pay for police overtime during the tournament. FIFA even shook down the feds for $2.1 million in federal counter-drone money.

For these services to World Football, Gianni Infantino will personally take home some $6 million this year, a third more than last year. FIFA also picks up the tab for his apartments in Switzerland and Paris, the school fees for one of his daughters, and the cost of the private Qatari-supplied jet he travels the world on. (FIFA has declined to clarify which Qatari supplies the jet.)

All of this weighs heavily on me as I contemplate Mexico’s kickoff against South Africa today. Because however much I despise FIFA, I’d be lying through my teeth if I told you I’m not desperate to watch. Football is exploding with talent right now. I’ve caught glimpses of what Lamine Yamal is able to do to defenders; it’s astonishing. He’s the most famous, but the list of not-even-old-enough-to-drink prodigies gearing up to debut is long: Mastantuono, Doué, Endrick, Cubarsí, and Kendry Páez have all put in credible auditions for the role of the next Mbappé. Who will it be?

But then, it’s not just the kids. The tournament’s stuffed with mature talent. England’s always-good, recently-great Harry Kane is just as exciting to watch as the teenagers are. Some past-their-prime (but still world-class) figures like Portugal’s five-time Ballon d’Or winner Cristiano Ronaldo and Egypt’s mega-charismatic Mo Salah must realize this is their last chance to write their names into legend. Even Lionel Messi, a God among men about to celebrate his 39th birthday, is still playing to his usual, insane standard.

Subscribe now

Look, the administrative monster that kidnapped the beautiful game is still a ghoul, but the damsel in its grip has become, if anything, even more ravishing of late. Only a spectacle as compelling as the World Cup could withstand the torrent of slime FIFA keeps it immersed in.

And so, like every four years, I will watch the footie the way a socially conscious addict stuffs cocaine up his nose: in the full, nauseated knowledge that I’m bankrolling some of the very worst people in the world but genuinely unable to help myself.

I mean, Lamine Yamal is playing. I’m not made of stone.

Quico Toro is a contributing editor at Persuasion, the founder of Caracas Chronicles, Director of Climate Repair at the Anthropocene Institute, and writes the Substack One Percent Brighter. He lives in Tokyo.

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

In a few hours, the Ukraine-Russia war will have lasted longer than the entirety of World War One—a vivid illustration of the war’s human cost.

Long-time readers will know that since shortly after the Russian invasion in February 2022, Persuasion has been following the war through the voice of one Ukrainian, Kateryna Kibarova. Kateryna’s accounts of the ongoing horrors—from the first days, to her return home to Bucha, to surviving the most recent winter amid constant bombardment—capture the seemingly never-ending nightmare facing civilians far from the frontlines.

In February, members of the Persuasion team launched a fundraiser which allowed Kateryna to replace her generator after repeated power surges destroyed it. She was deeply moved—as were we—by how quickly our readers responded, as you will see from the opening of her piece below.

Thank you for being part of our community, and for continuing to follow Kateryna’s story.

—David, Senior Advisor.

I would like to begin by expressing my deep gratitude to every reader, to those people who showed such heartfelt compassion regarding the tragedy I went through this past winter, and to those who helped and spread the word. It is thanks to all these people that I am still alive today.

This winter, there was no electricity, no heat. Everything was broken. My refrigerator burned out. People got food poisoning because everything in the supermarkets was going bad. When we finally ordered the power unit—which was fully installed and connected in April—it provided us with uninterrupted electricity throughout the entire day. It charges itself and handles everything automatically; it has truly been a lifeline for us.

Last month, Russia launched a massive wave of attacks: one Oreshnik, two Zircon, three Kinzhal missiles, 30 ballistic missiles, 54 Kh-type cruise missiles, and 600 UAVs. Of these, eleven ballistic missiles and 44 Kh-101 cruise missiles were shot down. The Oreshnik was not intercepted; neither the Kinzhal nor the Zircon missiles were shot down. There is likely not a single district in Kyiv that was left unscathed.

People who lived near a metro station took shelter in the subway, while those with access to underground parking garages headed down there. I was jolted awake by the distinct sensation of a vacuum—the specific feeling you get when a missile is flying overhead. I sprang out of bed and ran to the basement. Shortly after that, a massive wave of Shahed drones flew right over us. They sounded like mopeds. The missiles came one after another. Missiles, Shaheds; missiles, Shaheds. We sat there, terrified, for three or four hours.

Every district of Kyiv sustained strikes. Lukyanivka was destroyed. There used to be some kind of factory there, but they’ve struck it so many times that it hasn’t been operational for ages. A shopping mall burned down; the market burned down. All the residential buildings were heavily damaged. People in Lukyanivka have learned to immediately run to the metro, which is 70 meters deep. The escalator takes about five minutes to reach the bottom. The explosions and impacts reverberated all the way to that depth, and were so intense that plaster was raining down from the ceiling where people had taken shelter.

What does a typical day look like? An air raid alert goes off while we’re at work. Just like in any office, we have one designated “alarmist” who monitors all the Telegram channels to figure out exactly which direction the attack is coming from and when it’s expected to reach us. The moment we know a Shahed drone is approaching our office, everyone scrambles toward the exit. We head down to the shelter, wait for the drones to pass, and then we go right back to our workstations. In other words, we’ve adapted to living in absolute chaos.

When a specific situation arises—say a café gets bombed or something similar—people help out. They band together and offer assistance. But you understand that you need to keep your documents close at hand. You need to be able to grab them and any cash you have on hand and run. You have to figure out how to survive on your own.

One thermal power plant in Kyiv was hit six times, and is now completely beyond repair. To repair a thermal power plant, you have to order spare parts, which is a very lengthy process. And while they were (and still are) working to get things running again, many people—specifically from DTEK, the company responsible for restoring these power plants—have been killed. The Russians have developed a specific strategy: they strike a target. Then they hit it again just as ambulances or repair crews arrive. Striking twice kills both the staff and the people who might otherwise have survived.

Meanwhile, back in Zaporizhzhia, my mom’s washing machine won’t run. There isn’t enough electrical current to power it. That’s the constant struggle with electricity all across Ukraine right now. If someone had told me just five years ago that I’d be figuring out the ins and outs of generators, or walking to the gas station with jerrycans to fill up on petrol and haul it all the way home, I wouldn’t have believed them.

Ukraine is winning the drone war. Drones are assembled not only in large industrial facilities but also in ordinary garages by anyone with a basic grasp of electronics and a soldering iron. Regardless of how much Donald Trump might deny it—or claim “We don’t need your training, and we don’t need Ukraine”—we have truly surged ahead.

I know a guy—an ordinary guy, but a skilled IT specialist applying his knowhow—who developed a specialized targeting technology, something that increased the speed of execution. It is being used today in combat operations.

One young woman, an ordinary mother of two, went off to fight, leaving her children at home in the care of their grandmother. While she was serving in combat, she fell in love. This is how we got to know her—my company has a project where we donate wedding rings to military personnel currently in active service. Other companies have joined in to help: some provide wedding dresses, others catering.

Subscribe now

Early in the war, our guys were unprepared. Their only knowledge of war came from the Counter-Strike video game. Now there are more instructors, training camps, and specialized programs. Additionally, we’ve appointed a truly excellent Minister of Defense, Mykhailo Fedorov. He previously developed a mobile app called Diia, which serves as a centralized digital repository for personal documents.

Whenever our forces launch drone strikes, the Russian authorities cut off internet access because our drones navigate using mobile network signals. People in Russia are left for weeks on end without any internet connection. Now, the government has even started releasing propaganda videos claiming that the internet isn’t necessary—that life is better without it. Russians are asking: “What is this? We’re living in the 21st century. Are you trying to drag us back to the era when all we had was Tetris?”

On our side there is an intense focus on preparation, training, and self-assessment. That is why we are achieving results and that is why territories are gradually being liberated. Still, I can’t say that we currently feel like we are definitely going to win. People are experiencing a profound sense of fatigue. Men in particular are under an immense psychological burden; I’m referring to those who are not currently serving and who are, quite simply, afraid to step out onto the street because military recruitment officers might pick them up.

For years before the war, our army wasn’t properly built up. Russia wanted Ukraine to be an easy target, unable to offer resistance, and so during the periods Russia was influential in Ukrainian politics, military training was neglected. Now, naturally, men are afraid to fight, because they don’t know how. It’s like putting someone behind the wheel who doesn’t know how to drive and telling them to floor it up to 100 kilometers an hour.

I believe that, following our victory, mandatory military training should be instituted for both men and women. Here in Kyiv, the daughter of a well-known millionaire goes out at night and shoots down drones. In schools, students are now being taught how to administer first aid, how to apply a tourniquet, how to handle a firearm. I want to sign up for that kind of training. We have to be realistic and prepared: we can never be fully assured that Russia won’t come marching in again or launch another attack.

That’s why I can’t really say right now that there’s a prevailing sense of victory here. People understand that we are making headway, that we are advancing—but we also understand the immense human cost. We know how many young men have died and how many children will never see their parents again. It is truly heartbreaking. But that is simply our reality right now.

Kateryna Kibarova is a Ukrainian economist and resident of Bucha.

Translated from the Russian by Julia Sushytska and Alisa Slaughter. This transcript has been lightly edited for concision and clarity.

About the Translators: Julia Sushytska was born in L’viv and is a Resident Associate Professor in Comparative Studies in Literature and Culture at Occidental College. Alisa Slaughter is Professor of Creative Writing at the University of Redlands. They co-edited and translated a selection of lectures by Merab Mamardashvili, A Spy for an Unknown Country (ibidem-Verlag, Stuttgart, 2020) and Both Sides Face East: Durable Words, a collection of poems, essays, and stories in nine languages that began as a response to the invasion of Ukraine, but stakes a wider claim on behalf of human value and integrity (Academic Studies Press, May 2025).

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now

This article is brought to you by American Purpose, the magazine and community founded by Francis Fukuyama in 2020, which is proudly part of the Persuasion family.

Have we been here before? Scholars and pundits alike have identified striking parallels between today’s political developments and those of the interwar years: democracy in retreat, radicalized parties going from strength to strength, ethnic minorities scapegoated, dictators and would-be dictators advancing, economic crisis, the breakdown of a liberal international order. We even find a deadly pandemic in both periods, the Spanish flu of 1918-20 and Covid-19 in 2020-23.

The parallels send shivers down the spine. Tellingly, the economic crisis that began in 2008 was branded the “Great Recession”—a deliberate echo of the “Great Depression” that began in 1929. If that era is where we are headed, it is time to buckle up. Two dark premonitions are often combined: that democracy has experienced a decades-long rollback, and that we are now on the eve of something like the early 1930s, with democratic setbacks poised to turn into freefall.

For nearly two decades, the American organization Freedom House and the British magazine The Economist have published annual assessments of democracy with strikingly bleak titles, such as “Discarding Democracy: Return to the Iron Fist,” “The Global Expansion of Authoritarian Rule,” and “Global Democracy in Retreat.” These reports have been accompanied by a surge of books from scholars and public intellectuals, many of whom draw comparisons with the interwar period during which democratic systems collapsed across much of Europe and Latin America.

Take the following three books: How Democracies Die, by Harvard political scientists Steven Levitsky and Daniel Ziblatt, with the telling subtitle What History Tells Us About Our Future, the American historian Timothy Snyder’s On Tyranny: Twenty Lessons from the Twentieth Century, and former U.S. Secretary of State Madeleine Albright’s Fascism: A Warning. These books point to what the authors see as striking parallels between developments in Europe during the 1920s and 1930s and those unfolding today, particularly in the United States.

Yet the claim that today’s troubling developments mirror those of the interwar period is misleading—rooted in a superficial reading of that era and a flawed historical analogy. As the interwar U.S. Supreme Court justice Benjamin N. Cardozo once warned, metaphors and analogies intended to free our thinking can just as easily end up enslaving it.

Persuasion events

To begin with, the extent of our current predicament has been exaggerated. We have certainly seen instances of democratic backsliding and assertive authoritarian behavior since the turn of the millennium. But many studies—including some mentioned above—have overstated both the severity and the persistence of the negative democratic trend. In fact, global democracy measures and long-term historical analyses show that democracy grew substantially in the late 20th century, reached a high point around 2015, and has only recently shown a downturn, which is so far—with the very recent exception of the United States—confined to weaker democracies or reflects authoritarian regimes shedding superficial democratic features; it does not reflect a broad retreat within well-established democratic systems.

The more specific warnings about a repeat of the interwar developments are also problematic, for two reasons. First, they ignore crucial differences in the historical context, which we outline below. Second, the resilience of interwar democracies has been overlooked. Observers have mainly focused on the democratic failures of the 1920s and 1930s, and especially those of Italy and Germany. But this perspective ignores the fact that many long-established, well-institutionalized democracies successfully weathered the crises of the 1920s and 1930s, avoiding both widespread political radicalization and a breakdown at the ballot box or in the streets.

It is important to note just how different the interwar period is from what we find today. The very term conjures up an image of crisis. It denotes the 21-year interval between the two most destructive conflicts in modern history, from the end of World War I in 1918 to the outbreak of World War II in 1939.

The war between 1914 and 1918 profoundly destabilized European societies. Around ten million men died in the trenches under conditions that are difficult to comprehend today, while millions more returned home physically injured or psychologically scarred. Many veterans found it hard to reintegrate into civilian life, and large numbers gravitated toward extremist paramilitary movements in countries such as Germany, Austria, and Italy. In the 1920s and 1930s, these groups contributed to widespread violence and instability. The brutal clashes between police and protesters on May 1, 1929, in Berlin—“Bloody May” (Blutmai)—left 33 civilians dead by police gunfire. A staggering 89 demonstrators were killed by the police in Austria on July 15, 1927, after the Ministry of Justice was set on fire. In Germany, right-wing militias bearing menacing names such as Stahlhelm and Sturmabteilung fought with left-wing militias like the Reichsbanner Schwarz-Rot-Gold. Similar events took place in Italy in the early 1920s, where fascist blackshirts (Squadristi) fought communists and socialists.

Meanwhile, the economy was in tatters. In the years after the 1918 armistice, Europe experienced a severe inflationary crisis that far exceeded the levels seen during the high-inflation episodes of the 1970s and 1980s. In 1923, the Weimar Republic experienced hyperinflation on a nearly unimaginable scale. Postage stamps were issued with face values in the billions of marks, and wages were sometimes paid multiple times per day because money lost value so quickly that it had to be spent immediately. This hyperinflation wiped out savings while also creating opportunities for speculation and enrichment. Inflationary pressures also spread to Italy and the newly established states of Poland, Czechoslovakia, and Hungary, where they caused significant economic and social strain.

There was a brief respite in the second half of the 1920s. The Locarno Treaties of 1925 normalized relations between France and Germany, and Germany was admitted to the League of Nations, the precursor to today’s United Nations. The years from 1924 to 1928 are often referred to as the “Golden Years” of the Weimar Republic, a period during which Germany’s young democracy appeared to be gaining stability and confidence.

These golden years came to an abrupt end in the autumn of 1929 with the onset of a new and terrible crisis. The Great Depression was and remains the most severe economic downturn of the modern era. Between 1929 and 1933, global industrial production fell by an estimated 37 percent, world trade declined by 68 percent, and agricultural prices dropped by as much as 75 percent. In response, many governments introduced high tariffs and turned toward economic self-sufficiency—a policy chillingly known as “beggar-thy-neighbor”—which nearly brought international investment to a standstill. In Germany, the economic collapse led to mass unemployment, creating fertile ground for anti-democratic movements. Support for Communists and National Socialists surged as the crisis deepened between 1929 and 1933.

Meanwhile, the international order created by the victors of World War I collapsed. In 1933, Japan disregarded the League of Nations’ ruling that its occupation of Manchuria was illegal, and in 1936, the League futilely condemned Italy’s invasion of Abyssinia (Ethiopia). That same year, Mussolini’s Italy formed an alliance with Hitler’s Germany, while the democratic great powers were largely preoccupied with their own domestic challenges. On the international stage during the 1930s, might did make right.

While the world has experienced many crises in recent decades, they pale in comparison with what we have outlined above. For this reason alone, political developments are unlikely to repeat themselves. But there is another fascinating facet of interwar democratic stability that has too often been ignored.

A large democratic wave engulfed Europe after the victory of Western democracies in 1918: In 1919, only a small number of European countries—including the Soviet Union, Hungary, Spain, Portugal, and Albania—were governed by regimes that had not come to power through free elections. Two decades later, the situation had dramatically reversed as the number of European democracies effectively halved, with the democratic breakdowns equally distributed in the 1920s and the 1930s.

This is what keeps today’s observers awake at night. It seemingly shows that even democracies in Europe are not immune to breakdown during crises. But what is often forgotten is how these breakdowns took place in countries that had only democratized after the armistice in 1918. A large body of research has demonstrated that such new democracies are fragile during crises in general and economic crises in particular. The fact that half of the democratic breakdowns took place before the onset of the Great Depression in the autumn of 1929 illustrates their fragility.

More interesting is the resilience of the old democracies. Faced with the same cocktail of crises—economic, social, political, and international—all European countries that already had been democratic in 1918 survived the crises of the 1920s and 1930s. In these countries, democracy had developed gradually from within through long-standing internal political struggles between established elites and emerging mass movements, including those of peasants and workers. This history gave democracy a strong foundation, as both political parties and the broader population came to cultivate mutual respect and to internalize the basic principle of democracy: “sometimes we win, sometimes we lose.”

In countries such as Denmark, Sweden, the Netherlands, Norway, the United Kingdom, and Switzerland, extremist parties attracted little or no support during the 1920s and 1930s. In countries such as Belgium and France, there was more radicalization, both in the streets and at elections. But throughout the old democracies, we find a pattern where the democratic center was able to come together and hold the forces of radicalization at bay.

As political scientist Kurt Weyland has noted, contemporary debates are often marked by conceptual stretching, with terms such as “coup” and “fascism” being used in loose and expansive ways.

A closer look at the interwar period shows that many observers have made false historical analogies with that period. On the first page of How Democracies Die, Steven Levitsky and Daniel Ziblatt note about their native country, the United States, that “…even though we know democracies are always fragile, the one in which we live has somehow managed to defy gravity.” Upon inspection, it becomes clear that a surprising number of countries have, in fact, managed to defy gravity. The resilience of old and established democracies during crises is perhaps the most important lesson to be derived from the interwar developments.

This is not to make a case for complacency. The challenges facing democracy today are real, and the trendline of the past decade is certainly worrying. But the interwar analogy, however vivid and emotionally compelling, is the wrong lens. It conflates the structural fragility of brand new post-1918 democracies, in a world marked by war and about to face a series of dramatic and mutually reinforcing crises, with the very different situation of long-established systems today. It mistakes turbulence for terminal decline. And, as Cardozo cautioned, it risks enslaving our thinking precisely when we need clear thinking most, by obscuring what today’s democracies actually require to defend themselves, and what history suggests they are capable of withstanding.

Jørgen Møller and Svend-Erik Skaaning are both professors of political science at Aarhus University. This article is based on their new book, Seven Myths about Democracy (open access).

Follow Persuasion on X, Instagram, LinkedIn, and YouTube to keep up with our latest articles, podcasts, and events, as well as updates from excellent writers across our network.

And, to receive pieces like this in your inbox and support our work, subscribe below:

Subscribe now