What this covers

Tyler Cowen interviews Patrick Collison, the CEO of Stripe, in a conversation that reverses their usual roles—Collison asks the questions while Cowen ranges across macroeconomics, creativity, cultural change, and the future of stability in Western institutions. The discussion centers on why creative excellence requires a paradoxical structure: Diverse Monoculture that combines shared language and frameworks with selective barriers to outside influence. Cowen argues that global connectivity, despite its welfare gains, is rapidly depleting the small, insulated cultures from which major innovations have historically emerged, leaving a future dominated by a homogenized sameness that merely performs diversity while becoming less creatively fertile.

Cowen moves through several interconnected claims: that macroeconomics excels at retrodiction but cannot predict, so policy should favor minimal stable rules over fine control; that Creative Clusters like Renaissance Florence and Silicon Valley succeed through insulation rather than maximal openness; and that religion and durable cultural capital are underrated precisely because they sustain societal reproduction and resist monoculture. He contends that all systems can be gamed, meaning no policy is eternal, and that stability itself is dangerously underrated in the Western world—a complacency that echoes the intellectual complacency of the period leading to 1910. The conversation also addresses universal basic income (which Cowen has reconsidered), the fragility of tech clusters, why Mood Affiliation distorts reasoning, and his growing optimism about artificial intelligence, which has consistently surprised positively as it has been applied to new domains.

Sharpest takeaway

Cowen argues that creative excellence depends on a 'diverse monoculture'—clusters that combine common language, insulation from outside influence, and selective barriers—and that global connectivity, while welfare-enhancing, is rapidly 'cashing in' on irreplaceable small cultures, leaving a less creative future homogenized into a sameness that merely performs diversity.

  • Macroeconomics is good at retrodiction but cannot predict, so policy should rely on minimal stable rules rather than fine control
  • Successful creative clusters (Renaissance Florence, Silicon Valley, George Mason) require insulation, common framework, and barriers—not maximal openness
  • Religion and other durable cultural capital are underrated because they sustain societal self-replication and resist homogenization

The argument · threads13 threads · 50 claims
0.85

Societal stability is underrated and belief in liberty, rule of law are eroding dangerously.

20 pointscentrality 5/5
  • Since all systems can be gamed and are not indefinitely sustainable, there is an optimal time to crack into the wall—when other parts of society and institutions are still healthy enough to respond positively; on this view Brexit and Trump might be a needed course correction delivered at a relatively healthy moment, which is distinct from supporting the events themselves.

    “when you crack into the wall, there is an optimal time for that, and you want to do it when other parts of your society and institutions are still healthy enough.”

  • Nuclear weapons will always be the most important issue in the world regardless of cyber or pathogenic threats, because many actors have them, they have been used, they may become easier to acquire, they can be launched by mistake, and the surrounding institutions are not always well designed—making Asia's underpriced risk premium Cowen's single biggest worry.

    “One of his big influences on me was simply to get me to see nuclear weapons will always be the most important issue in the world, no matter what else you might think.”

  • Stability is underrated, but the Western world takes it for granted because it has had so much for so long and overweights its last 20-30 years; this complacency is exactly what makes the present a dangerous time, as a creeping deterioration of belief in individual liberty, rule of law, and cosmopolitanism could acquire self-fulfilling momentum—a smaller move than worriers claim but more dangerous because of that momentum, risking a repeat of the lead-up to 1910.

    “I think the Western world takes stability for granted because we've had so much of it for so long. And our basic model, as in macroeconomics, you overweight your last 20 or 30 years.”

  • Unlike Schelling, who believed a strong and enduring norm and taboo against nuclear weapons use has independent force sustaining their non-use, Cowen is not convinced the taboo is that strong—the absence of use may reflect mere self-interest rather than a normative taboo—which makes him more pessimistic.

    “whether it's mere self-interest or whether there's actually a norm and taboo with independent force, I'm not convinced by his arguments. And that's made me a little more pessimistic.”

  • There is a sharp split between economists who do macro and those who don't: non-macro economists are too cynical (macro is the favored whipping boy and slightly underrated), while too many macro practitioners are too arrogant, believing they work on God's chosen method, when in reality the field is much more pluralistic and everyone is blindly grabbing parts of the elephant.

    “in macro, you have too many people who are simply too arrogant. They think they're working on God's chosen method in different ways. And that's wrong also; it's much more pluralistic than that.”

  • Cowen leans more toward favoring monotheism than religion per se, agreeing with Hegel that polytheism may not be conducive to political order, while noting Hume's point that monotheism itself breaks down into partial polytheism because people want multiple things to worship—so it is not a stable balance.

    “Possibly, I'm more of a fan of monotheism than religion per se. I agree with [Georg] Hegel's remarks on polytheism that may not be conducive to political order.”

  • Twitter would be better if it kept tight character limits and imposed something like a Pigouvian tax via micropayments on photos, subtweets, and threads, allowing only a nonzero, limited amount—because tweets are now too long and rules matter for their own sake.

    “There ought to be like a Pigouvian tax on putting photos and subtweets in your tweets and threads. I favor a nonzero amount, not like the old rules, but it definitely should be limited and taxed with micropayments.”

  • Effective altruism is overrated by the people who know what it is and underrated by the entire rest of the world; much giving is not rational, and forcing it into a culturally specific rationality will produce less giving overall, while EA's certainty that targets like Harvard are bad ideas reflects too much pretense of knowledge—yet relative to ordinary giving mistakes, EA is way underrated.

    “It's overrated by the people who know what it is. It's underrated by the entire rest of the world.”

  • Facebook may be net good if you are a happiness-utilitarian but net bad if you value the aesthetic worth of significant cultural contributions, because by subsidizing sociability too much it strips away hidden, implied subsidies to culture—functions once served by, for example, declaring musical taste in seventh grade—leaving culture weaker even as people become happier and more sociable.

    “Facebook is subsidizing sociability too much. A function, say, that used to be served by music when you're in the seventh grade — "This is the band I like" — it's now served by social media.”

  • America will keep its two-party system for a very long time because the barriers to a third party are very high and the Trump experience shows how flexible existing parties are; the Democrats will do their own version of internal transformation, but no major new or third party will emerge for the foreseeable future—the last new party dates to the 19th century.

    “I think America will have its current two-party system for a very long time to come. The barriers to a third party are very high.”

  • Whenever an objectionable party is elected, the people in opposition become stupider as a near-rule—more emotional, attacking the worst versions of opposing ideas rather than picking and choosing or moving toward the center—and the intellectual Left has done this under Trump rather than giving Trump supporters an open path to becoming something else.

    “Whenever the party elected is objectionable, the people in opposition become stupider, as pretty much a rule. They become more emotional, and they attack the worst versions of the ideas they're against”

  • Starting a new religion is very difficult to pull off in the contemporary world because the entry barriers are high, existing religions are rich and well developed, and success requires a severity that is incompatible with the kind of person who would attend a Silicon Valley talk—so the advice is not to do it.

    “The entry barriers are high. And the religions we have are quite rich and well developed. And what you would have to be to succeed, I suspect, is someone who wouldn't be at this talk.”

  • It is possible the Trump/Bannon appointment strategy is to install many talented, non-political, genuinely different people and give them free rein knowing 80 percent will fail but a few might do good things—a deliberate departure from the Obama model of loyal, smart academics in a semi-well-functioning White House that didn't change many things.

    “"We're going to appoint a bunch of people who are really just different and talented but not political, and give them all free rein, and we know 80 percent of them will fail, but a bunch might do good things."”

  • Cowen now holds a 30 percent chance that we are on the verge of climbing out of the Great Stagnation, based on four or five encouraging signs in the last year: strong wage growth, better integration of information manipulation with real-world processes, and roughly sane healthcare cost behavior in four of the last five years—where the key test is whether technology eases how molecules actually impact your body, as driverless cars would.

    “I now hold the view, there's a 30 percent chance we're on the verge, right now, of climbing out of the Great Stagnation.”

  • Michel Houellebecq's Submission is a Straussian work with secret meanings: rather than being anti-Islam, it actually envies Muslims in France for believing in something, functioning like Jonathan Swift's satire as a criticism of a contemporary French intellectual class that no longer believes in things—a class to which Houellebecq himself belongs.

    “I think actually he's envying Islam and the Muslims in France for the fact that they believe in something.”

  • Cowen's 'business model' of trying to respond to every email he receives is effective because showing the right temperament in responses leads people to tell him everything he needs to know, which is why he considers email underrated.

    “I feel if my responses show the right temperament, that people will tell me everything I need to know. And I think that's worked pretty well. So email's underrated, in my worldview.”

  • A good rule for friendship is to really be yourself and be willing to be weird, letting people self-select around that rather than trying to make friends; friends provide companionship, common sense of purpose, and—crucially—tell you when you are completely off the mark, which is why working with people you have friendly ties with is very important.

    “a good rule of friends is to really try to be myself and be willing to be weird, if it's even weird, and let people self-select around that, to some extent. And that's worked well for me, more than trying to make friends.”

  • The world needs more single-issue blogs on important, underrated causes like penal reform and animal welfare that persistently deliver the message that something is wrong; though they may not be profitable, they can have a big positive social impact, whereas philosophy is very hard to blog because its arguments take too long.

    “Single-issue blogs on issues of importance. If you take something like penal reform, an underrated cause... The idea that there should be more blogs that track that and persistently deliver the message, "Something is wrong here."”

  • Quoting Frank Knight, economics' main function is to offset the stupid theorizing of other people—it is valuable as a form of discipline and a way of thinking that inoculates you against other kinds of mistakes, even though in some ways it may be a mistake itself.

    “Frank Knight once said something wise, maybe overstated... he said, "The main function of economics was to offset the stupid theorizing of other people."”

  • Once you have read a certain number of key, worldview-shattering books, books become an overrated and relatively weak, impotent way of learning new things; for simply learning something, a blog post is often better than a much longer book, though books still matter for embodying, storing, and certifying knowledge.

    “once you've read a certain number of key, earthquake, worldview-shattering books, books are way overrated. They're actually a pretty weak, impotent way of learning new things.”

0.61

Culture is the most neglected determinant of economic growth and innovation.

4 pointscentrality 5/5
  • The intersection of culture and economics is by far the most underinvested area of economics; understanding the determinants of growth is largely cultural, and bridging economic and cultural ideas is something economists are very bad at, making it macro's centrally neglected topic.

    “Culture and economics, for me, is by far the most underinvested.”

  • Renaissance Florence, with a population of only 60,000-80,000, produced enduring achievements of very high lasting market value, which naive economic theories cannot explain because small populations should not be able to do this—revealing there are missing inputs to certain kinds of production we do not understand, and studying them is key to understanding clusters like Silicon Valley.

    “in very naive theories of economics, that shouldn't be possible. People in Renaissance Florence, they didn't produce a refrigerator that we're still using or a tech company that we still consult.”

  • Successful clusters share the right kind of diverse monoculture with implicit barriers to too much outside influence, yet are stimulated by some major outside prods—as Renaissance Florence was prodded by rediscovering Chinese, Arabic, and antique technologies, and Silicon Valley is prodded by Moore's Law driving the urgent question of what is now possible.

    “they also tend to be stimulated by some major outside influences, such as in the Renaissance, rediscovering technologies from China and the Arabic world, rediscovering antiquity — a huge prod.”

  • Highly creative groups require a paradoxical blend: a common language and shared framework, a semi-diverse monoculture that focuses attention, optimal insulation from too many outside frameworks, and implicit barriers to outside influence—while still being stimulated by some major outside prods (like rediscovering antiquity for Florence or Moore's Law for Silicon Valley).

    “they also tend to be stimulated by some major outside influences, such as in the Renaissance, rediscovering technologies from China and the Arabic world, rediscovering antiquity — a huge prod.”

0.60

Economics cannot predict or control recessions despite improved diagnostic tools.

3 pointscentrality 5/5
  • Economics has advanced in retrodiction—we can now trace where bad mortgages were made and which banks held them in the Great Recession with detail unavailable for the 1979-82 recession—but it has made no progress in controlling or predicting business cycles, crashes, or recessions, and the best theories imply such events are intrinsically difficult to predict.

    “in macro, we're very good at retrodiction. If you look at the Great Recession, and you try to trace when and where were the bad mortgages made, and which banks held them, and when did they go bad, we can trace that with a level of detail and care, that, say, we couldn't have done for the 1979-1982 recession.”

  • On important macroeconomic issues, economics will not become predictive any time soon—not in our lifetimes—because we cannot rerun alternate histories and better statistics alone will never settle key disputes; there is plenty of micro (e.g., Uber's datasets) where prediction works well.

    “I don't think it will be predictive any time soon, not in our lifetimes. Not on a lot of important issues. There's plenty of micro where it's predictive.”

  • A carbon tax would be a fine idea but is in many ways overrated; the major factor in climate change may simply be luck—whether we invent something cheap enough that the whole world, including Africa and Vietnam, wants to adopt it—so getting science and the tech world to work better is probably higher impact than the political battles that feel so important.

    “a carbon tax is, in many ways, an overrated idea. Getting science and the tech world to work better is probably higher impact.”

0.46

Global homogenization exhausts small diverse cultures that produce creativity.

3 pointscentrality 4/5
  • Globalization triggers an intense, fruitful flowering of small cultures when they are first reached by richer cultures to trade with, but eventually those smaller communities are absorbed by larger ones; over the last century we have done an unprecedented amount of this 'cashing in,' so along some critical dimensions the next century will be less creative because we have depleted so many small groups—an optimal stock-depletion problem we may have pressed too hard.

    “until they're in some way reached by larger, richer cultures they can trade with, in many instances, they're not that creative... But eventually, that peters out as the smaller communities are absorbed by larger ones.”

  • Modern globalized cities exhibit a 'monoculture of extreme diversity'—every city has the same array of ethnic restaurants and we call it diversity, but we are all diverse in the same way, fooling ourselves; Taleb's point that free trade creates dangerous monoculture is intellectually serious even if reversing it would require infeasibly returning to 1500.

    “in some ways, we're investing in literally a monoculture of diversity. And that's a little dangerous.”

  • Caring about the distant future does not steer you away from exploiting small diverse cultures, because if you don't mine those cultural stocks their outputs deteriorate and decay anyway—most past culture is a closed book to us because it was never exploited—so the right approach is exploitation plus preservation, and we may simply have done too little preservation.

    “if you don't mine the stocks of these smaller diverse cultures, their outputs deteriorate and decay. So there's so much from the past we'll never have a clue about because it's gone, and we never "exploited" it.”

0.46

Immigration should select for global public-goods creators, signaling toughness to filter talent.

3 pointscentrality 4/5
  • Inequality and the absence of UBI function as signals to prospective immigrants about what kind of country this is—'life here is tough'—and selecting for productive immigrants who generate public goods is more valuable than optimizing for domestic happiness, because the US, like Stripe and Atlas, generates public goods for the whole world; thus immigration ethics should be viewed as an advertising campaign for a creativity-maximizing country.

    “By having a lot of inequality, you're putting up another kind of sign for immigrants: "Life here is tough. And if you live here, well, tough is tough."”

  • While open borders makes economic sense and the US is not yet at its sustainability limit, a United States with open borders would be politically unworkable in less than 10 years because political culture requires a common language and shared set of myths and delusions; open borders would kill the goose laying the golden egg, so Cowen favors much more immigration but not unrestricted immigration.

    “A United States with open borders, I think, politically, would be unworkable in less than 10 years, even though it makes economic sense.”

  • Australian and Canadian points-based immigration systems work great for those countries but would be wrong for the US because they bring in too many high-paid service professionals (like dentists) who don't generate large productivity gains; the US should instead select for engineers, creators, dreamers, artists, and eccentrics—public-goods producers—and its policy should be more random or appear more nonsensical.

    “they would bring in too many high-paid professionals in service sectors that don't bring that many productivity gains. Having more advanced-degree professionals isn't what we want.”

0.44

Stable monetary rules are the best macroeconomic policy available to modern economies.

2 pointscentrality 4/5
  • A Federal Reserve that is predictable and keeps the nominal flow of purchasing power growing at a steady rate is close to the best macroeconomic policy we can achieve; people did not know this in 1929, and it represents a huge victory for the contemporary world—favoring minimal stable rules over attempts at fine control.

    “the notion of having a Federal Reserve System that is predictable and tries to keep the nominal flow of purchasing power growing at some steady rate, we know that something like that is close to the best we can do”

  • Any good rule or system—including the US Constitution or a sound macroeconomic rule—can eventually be gamed, often through people using politics to distort or lie about the indicators, so there is no eternal solution, only policy semi-solutions good for 20 or 30 years.

    “all systems in the world eventually can be gamed. It could be your macro rule would be good for a long time, but it could be gamed if only through people figuring out how to use politics to distort or lie about the indicators.”

0.44

Redistribution requires opacity and politics creates resentment asymmetry killing UBI sustainability.

2 pointscentrality 4/5
  • Redistribution programs like the earned income tax credit work today precisely because they are not transparent—almost no one, even PhD economists, can correctly explain how the EITC works—so many sustainable structures of the contemporary world must remain nontransparent, and we will likely evolve disability insurance into an obfuscated partial UBI.

    “It works today precisely because it's not transparent.”

  • Universal basic income makes logical sense as societies grow wealthier, but Cowen changed his mind because of the political dynamics revealed in 2009: people hated mortgage cramdowns less for the bailouts themselves than for the idea of a neighbor getting a break they did not get—so the asymmetric treatment of neighbors that redistribution creates engenders politically unsustainable resentment.

    “they hated the idea of the person next door getting a break that they didn't get, when they had, in their own mind, worked harder and paid all their bills. And that proved impossible.”

0.31

Diversity in correct beliefs is lower than in incorrect ones by pure mathematics.

2 pointscentrality 3/5
  • A correct point of view will be less diverse than incorrect ones because there are many more ways to be wrong than right; thus being completely right about something is itself a form of non-diversity, and the kinds of diversity elevated in coastal American political discourse may not be the kinds that actually matter.

    “a simple principle is, our correct point of view will be less diverse than people who are wrong because there are many more ways to be wrong.”

  • Mood affiliation is judging arguments by the mood of the argument rather than its substance—e.g., treating an optimistic argument as more likely correct because optimism feels comfortable, or rejecting a substantive claim because someone failed to condemn something with the 'right' mood—and the antidote is strict Bayesian unbundling that lets you hold optimistic, pessimistic, approving and condemning moods simultaneously and inter-temporally.

    “Mood affiliation was a concept I first coined to refer to people who judge arguments by the mood of the argument.”

0.31

Randomized trials and micro-level empirics produce knowledge though macro remains unpredictable.

2 pointscentrality 3/5
  • Randomized control trials, though expensive and limited to narrower questions, are remarkable because they produce actual knowledge that is probably true; it is odd to turn down that reward even if you cannot use them to answer big-picture political economy questions that no one can steer.

    “randomized control trials, they're expensive, but you do actually learn things from them, which are probably true. That's remarkable. They contain actual knowledge.”

  • Economics blogs flourish more than other social-science blogs because economics has a common framework (microeconomics) that all economists and informed laypeople can grasp, plus an empirical side whose results can often be communicated in a single sentence; sociology lacks the common framework and philosophy lacks easy empirics and has arguments that take too long.

    “Economics has a common language... We have microeconomics. You can put an argument into micro and all economists can grasp it... And we have an empirical side.”

0.31

AI development continues surprising upward, exceeding past predictions substantially.

2 pointscentrality 3/5
  • AI in the broad sense has developed more quickly than people thought and even more quickly than Cowen predicted in his 2011 book Average Is Over, and each time we apply it to new problems we are positively surprised—remaining on a positive-surprise curve—explaining why he is more optimistic now than two or three years ago.

    “AI, in the broad sense of that term, seems to have developed more quickly than people thought. And even in 2011, I wrote a book saying, "AI will develop more quickly than we think."”

  • In chess, much of the creativity already stems from AI and computers playing chess, and Cowen has listened to computer-composed music and art—not yet impressive but not ruled out—suggesting AI-generated culture is a real near-future possibility.

    “in chess, so much of the creativity already stems from AI, computers playing chess. I've listened to computer-composed music, art. I don't think it's impressive yet, but I don't rule it out by any means.”

0.31

Religion sustains population stability and cultural capital that secular societies lose.

2 pointscentrality 3/5
  • Religion is a special form of cultural capital that, even adjusting for income, has predictive power over birth rates; secular societies fall into much lower birth rates and fail to reproduce themselves, which is among the most damning things about a society, and America's persistently strong religiosity for its income level helps keep its population stable or growing.

    “Religion, even adjusting for income, has predictive power over the birth rate. It's one reason why America still has a stable or a typically growing population”

  • Guaranteed annual income could work well in smaller, more cohesive countries like Australia and Denmark where life is already less harsh, but America's unique extreme-Puritan vision of work-based personal projects pursued obsessively is so special and valuable to the world that the US neither should nor will adopt UBI any time soon.

    “I am an American exceptionalist. We have this unique vision, a kind of extreme Puritanism, adopting personal projects that are work based and obsessively seeing them through in a determined way”

0.31

Tech clusters are rare, fragile ecosystems requiring diversity monoculture and external prods.

2 pointscentrality 3/5
  • Although restrictive land use regulations are generally terrible and most of the country should build like Houston, the Bay Area may be the one place where restrictions make sense: most residents want them, so removing them would be a tax, and since these residents produce amazing global public goods and will face rising taxes anyway, loosening building might wreck a uniquely productive diverse monoculture rather than enhance it.

    “I wonder if the Bay Area isn't the one place in the world where building restrictions might make some sense because most of you want the restrictions.”

  • The final number of true tech clusters will be surprisingly small—the Bay Area, several parts of China, and a smaller but potent Israel—because the ecosystem is fragile and depends on project evaluation skill and closeness to the few major financial centers; Rust Belt regions that try to build tech hubs will fail, and may best become empty, though the Trump administration will redistribute income back to them.

    “I think the final number of tech clusters will be quite small: the Bay Area and several parts of China, Israel being a much smaller but still potent one. I think it will shock people how few tech clusters develop.”

0.18

Intellectual power emerges from systems spanning many voices, not from individual thinkers.

3 pointscentrality 2/5
  • No single person, blogger, or tweeter is very good no matter how clever; the real intellectual power comes from an embedded algorithm across blogs, Twitter, and social media, so the skill worth obsessively training is reading the system rather than relying on any one individual's views.

    “I don't think any one person is very good on Twitter, no matter how smart or clever they may be. And I don't think any one blogger is very good at all. So there's a kind of embedded algorithm in blogs, Twitter, other social media.”

  • The fundamental commonality linking the George Mason economists (Kling, Hanson, Caplan, Tabarrok) is how they process information—a willingness to be analytical about everything and to take beliefs from one area and apply them consistently to all other areas of life, with feedback into research—contrasting with academics who think scientifically only about their narrow research but not their politics or lives.

    “if you believe something, at least try to see it through: What else does it imply? What else does it imply for all other things I believe in life? And then allow that to have feedback into your research.”

  • Marginal Revolution succeeded on the principle that if you have something to say, giving it away for free will lead to something, and if you don't, giving it away for free will tell you that you suck—so the right attitude is 'full steam ahead,' combined with persistence and determination for something you enjoy and believe in.

    “If we have something to say, giving it away for free ought to lead to something. And if we don't have something to say, giving it away for free ought to tell us we suck. Full steam ahead.”