What this covers

Liberty does most of the talking in this conversation with Matt and Dom. It starts with a personal question: how do people find what they love, and how long should they stay with it? Liberty argues that appetite grows through doing, so preferences show up in action rather than in stated wishes. He adds that people predict their own happiness badly and copy one another through mimetic desire, which tilts them toward exploitation over exploration. He says a month or more of practice is needed before dismissing something. Matt's recurring test is whether continuing could pay off through compounding or whether the excitement has simply gone.

The second half moves into media economics. Liberty describes the mid-sized creator who imitates Reuters or the Financial Times without their legacy and goes bankrupt. He says traditional gatekeepers took most of the value, citing authors' 10 to 12 percent of list price. He contends that recommendation algorithms are procyclical and cannot find things outside a person's existing taste, whereas a trusted human curator can. He ties the value of niche media to the parasocial relationships listeners form over hundreds of hours. Stacking topics makes an audience harder to convert but very low in churn. His own growth comes largely from Substack's recommendation feature, with about 120 newsletters pointing to his.

Other threads include Joe Rogan's lead as a product of accumulated hours, compared with Warren Buffett, and why creators quit either early or after succeeding. Liberty also discusses the leverage of books, with Douglas Hofstadter's Gödel, Escher, Bach as an example. The conversation closes on how Infinite Media chooses whom to back.

Sharpest takeaway

Liberty argues that in an infinite-content world, the durable model is a 'smile curve' of giant brands and tiny, authentic niche curators, and that individuals should pursue experiments and passions only while intrinsic motivation lasts, staying with them long enough for compounding to work.

  • Authentic niche curators build parasocial trust that scaled brands and mid-sized imitators cannot match
  • Algorithms optimize procyclically and cannot explore beyond existing tastes, so human taste becomes the scarce input
  • Experiments should be run without precious attachment, but abandoned when the internal fire is gone rather than when external results disappoint

The argument · threads9 threads · 81 claims
0.76

Curiosity-led human curation beats algorithmic selection and serves people who lack time to explore.

11 pointscentrality 5/5
  • Liberty's broad, unplanned topic range is an emergent property of his personal curiosity, which makes it hard to replicate, though it also makes it hard for him to explain how he does it.

    “It's just emergent property of just being me.”

  • Because there is so much content in the world, it is unlikely that any one person has found everything that fits the shape of their mind, so Liberty positions his work as exploration provided as a service to people who lack time or drive to explore.

    “What are the odds that everyone has found everything that would fit the shape of their mind?”

  • Jim O'Shaughnessy recruited Liberty to OSV by asking him to keep doing what he does, finding great creators, while OSV supplies the means and infrastructure to amplify them through books, documentaries, and other projects.

    “He basically said, keep doing what you're doing, just keep finding great stuff out there, great content, great creators, but you can do it for us.”

  • Good curation is an esoteric skill that relies on the human brain's pattern recognition across years of reading, listening, and watching; its logic cannot be fully explained, and if it were, AI would automate it.

    “so it's just using the human brain's pattern recognition capabilities”

  • Recommendation algorithms are procyclical: they learn what a user likes and deliver more of it, but they do not find items unconnected to current tastes, which a human with similar or trusted taste can do.

    “They figure out what you like, and they'll give you more of it and more and more and the mountain of it, but they won't find that thing that you like that is not directly connected to the things that you currently like.”

  • Standard growth best practices, such as catchy and linkable titles and tight targeting, are the opposite of what Liberty does, and his growth therefore depends on word of mouth and peer recommendations.

    “Basically, all of the best practices for growths are the exact opposite of what I've been doing for 3 years. I have very long titles that are just a jumble of topics.”

  • Liberty chooses topics by his own curiosity rather than predicting what readers will like, because optimizing for audience demand would have led him to the obvious investing-recommendation newsletter that he finds unappealing.

    “I'm my first reader, I just follow my own curiosity. I never think about, oh, what will people like”

  • Writing to a single imagined reader rather than a general audience, a technique David Senra described, makes the writing more personal and direct.

    “now I always write the newsletter to one person, not a general audience, not a stadium, not guys or amorphous blob, it's you.”

  • The text stays the same while the reader changes, so a book reread years later can hit completely differently, and meaning is shaped by the reader's accumulated experiences and context.

    “if you think of the book as unchanging, the person reading it and interpreting it changes a lot.”

  • Douglas Hofstadter's Gödel, Escher, Bach is a rewarding but hard book that combines math, art, and ideas about self-modeling consciousness, and it rewards readers who work through its early math rather than skipping it.

    “It's Gödel, Escher, Bach by Douglas Hofstadter. And it was hard to read.”

  • Liberty has not made his brand explicitly anchored to a figure like Buffett and Munger, as Farnam Street does, because no existing thing has been big enough in his mind to anchor to, and any such anchor would have to emerge organically.

    “But there's not a single thing that's big enough in my mind space that I would just decide to anchor to it and make it explicit.”

0.71

Personality-driven niche creators win loyal audiences through personal connection that mass media and bland brands lack.

16 pointscentrality 4/5
  • Most of Liberty's audience growth has come since Substack launched its recommendation feature, which lets other newsletters point their subscribers to his, and about 120 newsletters now point to him.

    “most of my growth has been since Substack started the recommendation feature, which allows other newsletters to tell their subscribers about me. And I -- maybe, I don't know, I'm very grateful to have something like 120 other newsletters pointing to me.”

  • A creator does not need to reach a billion people; with about a thousand true fans, as Kevin Kelly described, a creator can serve a specific group very deeply.

    “You can do very well with your 1000 true fans, as Kevin Kelly would say.”

  • Authenticity is the differentiator in a flood of content; creators who fake enthusiasm for trending topics are exposed, while genuinely interested creators stand out.

    “And the best way to be interesting is often to be interested. If someone is faking it, riding a wave of something that's popular, I don't know, it shows”

  • Liberty saved 50 to 70 percent of his income for about a decade before the Substack years and invested with only his own capital for eight or nine years, a near-FIRE approach he chose to avoid working for someone else in a tie.

    “saving 50% to 70% of my income for a decade doing almost a fire thing”

  • After being permanently suspended from Twitter for paraphrasing Putin, Liberty concluded that people who do not own their own presence can lose it at any time, so he started a Substack in July 2020 as a place he controls.

    “if you don't own your own presence, it could go away at any time, Twitter bans you”

  • Remembering that every username is a real person in a specific place helps creators avoid projecting their own profile onto audiences, and reader emails reveal surprising locations and professions such as surgeons in Singapore or Taiwan.

    “there's a person behind every Avatar, every username, there's someone at a keyboard over there somewhere”

  • The stable structure of digital media is a smile curve: giant legacy brands at one end, a hollow middle of mid-sized creators who imitate brands without their legacy and burn out, and authentic tiny niche creators at the other end who cannot be crushed by giants.

    “I think it's going to be the smile curve for a long, long time. I can't really imagine that the big thing can crush the tiny niche things”

  • Readers quickly decide whether a niche writer is for them and self-select out, while those who stay receive something unique, so Liberty responds to every comment and email to build a two-way relationship.

    “someone reading is very, very quickly going to figure out, okay, this is not for me, and they're going to self-select out”

  • Listeners form parasocial relationships with hosts they hear for hundreds of hours, and these relationships make niche authentic media valuable because the same facts from elsewhere lack the personal presence that makes them feel like friends.

    “from my brain's point of view, these are close friends. There's a very strong parasocial relationship with these people. I could maybe get the same facts somewhere else but something would be missing.”

  • Fragmented media gives up the shared experience of common cultural touchstones, such as broadcast TV shows that everyone discussed, in exchange for taste that is specific to each person's mind.

    “We're trading off that shared experience where everybody has seen Leave It To Beaver”

  • Traditional gatekeepers such as publishers, TV channels, and studios controlled distribution and extracted most value, with authors often receiving only about 10 to 12 percent of a book's list price, which Liberty considers unjustified because publishers add limited value.

    “the author may be getting 10% of that, 12%. That's bonkers to me.”

  • Large legacy brands reach many people but are bland because they must appeal to everyone, and new large brands are hard to create because legacy and distribution cannot be replicated easily.

    “They reach a lot of people, but they're bland. They have to be for everybody.”

  • Traditional gatekeepers extract value from creators, with publishers paying authors roughly 10 to 12 percent of a book's price, so a curation-based company can add more value by supporting creators across mediums in a win-win way.

    “the author may be getting 10% of that, 12%. That's bonkers to me. Because while a publisher is adding value, they're not adding that much value to it.”

  • Liberty sees himself as very niche because combining many topics, such as Deadwood, nuclear, Soviets, Marines, enterprise software and AI, makes his niche more specific.

    “I'm still niche. I'm very, very niche. The conjunction fallacy, the more things you combine together, the more specific it becomes.”

  • Mid-sized creators try to imitate authoritative institutional voices like Reuters or the Financial Times without their brand or legacy, and they go bankrupt before they can fund journalists and infrastructure.

    “They never get to the scale to be able to afford all the journalists and all the infrastructure until they go bankrupt basically.”

  • Small niche creators like CGP Grey are connected by the creator's personality rather than topic, which lets a varied set of subjects hang together, and this relationship is something big media cannot offer.

    “What's the link? The link is his personality. The link is I kind of like him.”

0.50

Intrinsic motivation and repeated experiments, not market indifference, should decide what creators pursue and backers fund.

12 pointscentrality 3/5
  • External indifference from the market or audience matters less than intrinsic motivation; as long as a person still loves the activity, that is a good sign to keep going, because loving something that much implies others with the same taste exist.

    “as long as you have the intrinsic motivation, as long as you love it, as long as that's still there, I think it's a good sign to keep going because you're probably not that unique”

  • Believing that one experiment does not have to work lets people run more experiments and try more things, which raises the chance that one of them succeeds, while preciousness about a single bet creates overthinking and paralysis.

    “if you go in with that mentality that, okay, this one doesn't have to work, I think you run more experiments”

  • One can never be sure whether quitting was right, since staying longer might have restored the fire; this irreducible uncertainty is what makes the decision hard, and if it were easy an algorithm could do it and the rewards would be lower.

    “If it was easy, everybody could do it, you could make an algorithm for it. There will be less reward for the people who can get it”

  • Even external success is not a reason to continue if the internal fire is gone; the first step is a break to see whether the passion returns, and if it does not, the person should move on to something else because life is too short to do work they hate even if it pays.

    “If -- even if you get some external success, but you've lost the fire on the inside, that's the first sign to, okay, you go take a break, see if it comes back, but if it doesn't, move on to something else.”

  • Joe Rogan's dominance reflects decades of compounding podcasting hours, similar to Buffett's long compounding, so a newcomer doing equally good work for 20 years still trails someone who started 70 years earlier by the same logic.

    “is he breaking out, but is he also just reaping the rewards of compounding for much longer than everybody else, like Buffett?”

  • Most creators quit from below by burning out in the flat early compounding phase, while talented creators who reach escape velocity leave from above for hedge funds or media jobs, so both groups fail to reach the rising part of the exponential curve.

    “Any exponential curve for a long time it looks flat. And so people almost always jump out of the train in that part of it”

  • Evaluating whether a creator is truly committed is a messy human judgment in which mistakes are common but usually small, because people who are not committed will leave on their own.

    “It's always a judgment call. As anything that has to do with humans, it's going to be messy.”

  • The creators Liberty wants to support are those intrinsically motivated by writing, podcasting, or research itself rather than using the work as a stepping stone to another job, because those people are more likely to persist.

    “What I'm trying to find are people who are doing the thing for the thing itself”

  • A creator's commitment cannot be engineered by a backer, since persuaded commitment does not last; the backer's job is a judgment call about whether the creator truly wants the work and to bridge financial gaps during the transition.

    “I would never want to try to change someone's mind about what they want to do because that's not going to last.”

  • Infinite Media at OSV aims to find under-the-radar, authentic creators and add distribution, logistics, editing, and marketing resources in a win-win arrangement across film, books, newsletters, podcasts, and video.

    “So what we are trying to do with Infinite Media is to create something for this new world, for this long tail, for this niche”

  • A producer's value, like the producer of The Offer, lies in taste and coordination, choosing the right director and making decisions that help the work without holding the camera or writing the script.

    “you're not holding the camera, you're not writing the script, but you're helping all these people work together”

  • Liberty does not try to change a creator's mind about their goals because such changes do not last; instead he makes a judgment call about whether they truly want the work, and treats mistakes as recoverable since no one can be forced into it.

    “I would never want to try to change someone's mind about what they want to do because that's not going to last.”

0.49

Broad cross-topic exploration creates differentiated niches that specialists and mass audiences leave open.

4 pointscentrality 4/5
  • Investing is a way to understand the world because companies are how humans do hard things together, so studying companies across industries, from insurance to aerospace to rail, exposes much of what people build.

    “investing is cool because it's a land through which you understand the world. A lot of the most interesting stuff that humans do is they get together and they do something very hard, and to do that, they form a company.”

  • Each additional topic combined into a writer's mix makes the audience more specific, and although such niches are hard to convert, the people who find them have nowhere else to go, producing very low churn.

    “The conjunction fallacy, the more things you combine together, the more specific it becomes.”

  • Specialization is the dominant strategy for career success, but when everyone leans toward specialization, an opening appears for generalists to go the other way, exploring across many topics.

    “But the dominant strategy in this modern world is to specialize. That's the way to get ahead. But if everybody is leaning on one side of the ship, it creates an opportunity on the other side”

  • Within a long tail of niches, each niche has its own power law, so media will be highly fragmented, with a winner in each niche that does not necessarily graduate to mass fame, and combining several topics creates unique, defensible niches.

    “it's basically a bunch of tiny niches in there, and there's power laws within all of these niches. So a tiny niche about Dungeons and Dragons is futuristic, mix three or four topics together, and you get something unique.”

0.48

Persisting through the flat, discouraging early phase requires patience and public output rather than quitting.

11 pointscentrality 3/5
  • Companies and projects often keep running past their useful life because management is paid by the year, while the better move may be to wind them down gracefully and profitably; the same logic applies to any personal experiment.

    “maybe the best thing to do for the company would be to just run off and try to do it as gracefully and profitably as possible”

  • Committing to reading for only 10 minutes whenever free time appears restarts the habit, and many short sessions grow into 30 to 60 minute sessions.

    “any time I have 10, 3 minutes, I just get started. I read a few pages.”

  • Humans are very bad at predicting what will make them happy and are highly mimetic, copying what others do, which pushes many people onto similar paths that do not make them happy and skews the exploration-exploitation balance toward exploitation.

    “humans are very bad at predicting what will make us happy. So we are very mimetic. We look around what are other people doing and we just copy that.”

  • Passions that last years have a high floor, meaning they are hard to start, and a high ceiling, meaning they keep rewarding learning, unlike a game like Tic-Tac-Toe whose low floor and low ceiling stop satisfying quickly.

    “The best games are the infinite games where maybe the floor is very high. It's very hard to get started. But the ceiling is incredibly high”

  • Appetite grows through doing, so preferences are revealed by action; claiming a wish to read more is not evidence of wanting it until one actually starts reading.

    “l'appetit vient en mangeant, which translates to something like hunger comes as you're eating. The more you eat the more you're hungry, it revealed preferences.”

  • People judge new activities too quickly and quit during the flat early phase of compounding; sticking with something long enough, or finding a contagious enthusiast, helps them reach the rising part of the curve.

    “I think too many people bounce off things that maybe they'd love if they just either stuck with them longer or maybe found someone to be their sure power”

  • Starting to create is not capital intensive since most people already have a computer and inexpensive microphone, so the real barrier is getting advice and being pointed to the right resources.

    “Creation is not a super capital-intensive thing. Most people already have a computer”

  • People judge new activities too quickly and bounce off them; a month or more of sustained practice is needed before deciding something is not for you, because the early unpleasant phase is common.

    “try it for a month, do these longer experiments. You don't have to last forever”

  • Private notes accumulated over time (300,000 words in a year) did not satisfy the urge to create, because writing only privately lets the brain skip steps; publishing publicly is what closes the loop.

    “if you only write privately, your brain just skips steps, and it's like, yes, nobody is going to see it, it doesn't matter.”

  • Solo creation is much harder than listeners and readers expect because the commitment of doing it day in and day out, week after week, is not visible from the outside.

    “creation, especially if you're solo is extremely difficult. It's much, much harder than most people would expect.”

  • The honeymoon period of a new creative project fades by the third to fifth month, when constant output expectations for podcasts, newsletters and videos produce burnout.

    “they had to do it in the third month and the fourth month and the fifth month”

0.21

Long-form books and newsletters offer more depth and leverage than shorter media.

4 pointscentrality 2/5
  • Books that are harder to read and read by fewer people make the reader more differentiated, because they gain information and connections that others do not have.

    “the harder they are to read and the fewer people read them, the more differentiated you are if you do read them.”

  • A newsletter gives a writer a large canvas for links and images and feels private rather than public, so readers who dislike it can simply not subscribe, which lets the writer be quirkier and go further into topics than on Twitter.

    “a newsletter is more intimate, right? So it's like if you don't like my stuff, you don't have to subscribe. But here, I can be more of myself.”

  • A good book has leverage that podcasts lack because the 20,000 to 50,000 people who finish it are self-selected and motivated enough to change their practice and spread the ideas, and seeing arguments organized on the page can move people to act where listening did not.

    “Seeing it on the page is making such a big impact on me that I'm making life changes because of reading that book that I have not made because of the podcast.”

  • Great books, albums, and shows widen experience and make people richer, which is undervalued by people who judge every activity by career return.

    “a great TV show, a great album, a great book brings something to the human experience that is very underrated”

0.21

Revenue models and creator scouting should be built around audience support and early discovery.

4 pointscentrality 2/5
  • Liberty launched free with voluntary Patreon-style payments about eight months in, which grew into a full-time income, then added a paywall at the end of 2021 when the market mood shifted and people stopped paying for subscriptions.

    “at the end of 2021, the market -- the mood changed, and now, people don't subscribe to anything anymore.”

  • Liberty uses the subscriber count of Byrne Hobart's newsletter The Diff, roughly 12,000 at launch and about 19,000 three years later for Liberty, as a unit of measure; he is now at about two Hobarts, a level he never expected.

    “I consider it to be a unit of measurement, so it's one Hobart”

  • Infinite Media uses two buckets: a venture-style incubation of promising but unproven creators, a high-variance bet filtered by confidence about who will fail, and support for mature creators with existing audiences who want to grow further.

    “I'm looking at it as two different buckets. And so the first bucket is more -- almost like a VC bucket”

  • So far Jim O'Shaughnessy has found the creators Infinite Media works with, spotting people with about 3,000 Twitter followers who have something special before the broader market notices.

    “Jim has an incredible radar to find talent. And so far, the people we're working with is exactly that.”

0.21

Amplifying people's own enthusiasms and visions counters a culture overweighted toward exploitation.

4 pointscentrality 2/5
  • Humanity is more constrained by lack of vision than by lack of talent, so amplifying people who generate dreams and visions of the future is a high-leverage goal.

    “humanity, in general, is very limited by vision, not by talent”

  • People find new favorites through someone already enthusiastic about something; sharing your enthusiasm openly with friends and family, rather than worrying about seeming weird, is the spark that pulls others in.

    “If you find someone that already loves something, that's usually pretty contagious.”

  • The modern exploration versus exploitation balance is heavily tilted toward exploitation, which is why Liberty tries to row against it by helping others find what they actually like.

    “this exploration versus exploitation algorithm is very skewed right now”

  • People tend to project their own knowledge onto their audience, and Liberty learned that his readers are in varied places and professions, which he could make visible by collecting where people read and listen.

    “there's this weird thing about you tend to project what you know onto your audience too”

0.07

Repeat consumption of great entertainment is valid, and creative work becomes joyful as one's own project.

4 pointscentrality 1/5
  • Working at OSV and writing the newsletter reinforce each other, and because Liberty only works on projects he loves, his job no longer feels like punching a clock and everything is his fun project.

    “It's all my fun project now.”

  • Hofstadter's treatment of consciousness as a loop in which a system models itself, combined with isomorphism, explains how humans think and shaped Liberty's early-20s understanding of cognition.

    “the consciousnesses loop where the thing models itself and isomorphism and all kinds of cool concepts”

  • Entertainment such as great books, albums, and TV shows is a valuable part of human experience that narrow productivity-focused people undervalue, and rewatching or rereading them is fine if the person enjoys it.

    “Entertainment is a huge part of life. And so you may as well find the very best quality entertainment that makes you go on journeys to other worlds”

  • Rereading a favorite work years later can produce a different reaction because the reader has changed, and Liberty believes repeat consumption of great books, albums, and shows is valid.

    “sometimes rereading the same book that you've loved that you know is profound your brain wants to remember everything in it.”