Leaking Timeline

Leaking Timeline

By Erik Newton & Guy SengstockSociety & CultureTechnology
Download on the App Store

Leaking Timeline episodes

  • Erik & Guy: The Anthropic S-1 and the Question of the Good

    Recorded live on November 2, 2026.

    Erik and Guy take a host-only week two days after Halloween, with Guy landing from a leadership course in Austin about 30 seconds before air. They start from Anthropic's S-1 filing: Erik walks through the numbers as he read them, explains why every S-1 carries a long risk-factors section, and asks what it means that the company whose filing says its product could end the world, and whose technology the market is valuing above anything before it, is governed by a small group of people in San Francisco. That opens the conversation they came for. Guy names his unease with the certain, simple transhumanism of the rationalist and effective-altruist circles he has known, and calls the economy "literally what religion used to be." Erik lays out his own position: the incentives are set, there is no enforceable slowdown, and the rational move is to build toward the good outcome, because the good, the true and the beautiful (a line he credits to Ginevra Davis) are attractor states, and truth is a function of survivability. Before the radio hour closes, Erik marks November 2, Dia de los Muertos, as the third anniversary of his wife Aubrey's death. In the shorter podcast hour, Guy brings Heidegger's account of technology to a stock market increasingly made of AI companies, Erik makes a second, better attempt at saying what it means to be inside this change rather than steering it, and the two agree the show should bring more continental philosophers into the AI conversation.


    Key Topics

    • [00:00 - 02:31] Cold open: Guy back from Austin just before air; the week's agenda, starting with Anthropic's S-1; trick-or-treating the night before
    • [02:31 - 07:03] Anthropic's S-1, as read on air: the revenue, cash and valuation figures as Erik read them; the chips, memory and energy constraint behind the data center buildout; the IPO as moving that buildout from private debt to public equity; "the most valuable one we've ever touched"; techno-feudalism
    • [07:03 - 12:20] Who governs it, and the risk factors: concentrated governance of a frontier lab; Guy on self-organizing systems and founders who once wrote warnings about exactly this kind of power; Erik the lawyer on why every S-1 lists every risk, and why this one stands out; Guy on the "weird fringe cosmologies" of rationalists and effective altruists
    • [12:20 - 17:10] Transhumanism and its flavors: Erik's primer for listeners; Guy's objection to the certainty and simplicity of the take-over story; "there is this thing called reality"; our capacity for self-deception
    • [17:10 - 23:39] What they get right: the human being is not the end of the whole thing; "We're of it"; the market as an aggregated belief system; AI's addressable market as the entire economy; the economy as a religion (Guy cites John Vervaeke); automation and a world of people "paid to stay out of the way"; Erik as a sober optimist
    • [23:39 - 30:22] The incentives are set: Erik's case against both slowdown and denial; x-risk as binary; the list of what goes right if it goes right; Ginevra Davis on the good, the true and the beautiful as attractor states of intelligence; truth as a function of survivability
    • [30:22 - 37:53] What the models are trained on: LLMs as expressions of the collective unconscious, and why their default essays are slop; Plato's cave; status as a kind of truth; Elon Musk's "maximally truth seeking"; Guy's question, "is it good?"; the deepest human questions, open to everyone
    • [37:53 - 40:13] Aubrey: Dia de los Muertos and the third anniversary of Aubrey Pick's death; radio sign-off
    • --- [Act 2: Extended Conversation] ---
    • [40:13 - 45:46] Is it good?: the transcendentals as a North Star you always fall short of; Guy's estimate that AI companies are about 30% of the stock market before Anthropic lists; Heidegger's Gestell and everything seen as "a resource standing in reserve"
    • [45:46 - 51:55] The understanding of being: everything is a being; the pre-reflective understanding we show up inside of; why seeing the uncanniness of the age means it is "not totally total"; listening, waiting without waiting; the overlap with Eastern traditions and one shared question
    • [51:55 - 55:01] Inside it, not doing it: Guy on why "stop it or slow it down" presupposes we are the ones doing it; Erik's second try ("Did I say it better that time?"); Dan Faggella's parabolic curve; thinking technology beyond itself; humans as the vessel for the birthing
    • [55:01 - 1:04:11] When philosophy becomes timely: Joscha Bach as a kid in a candy store; the analytic and continental divide; David Bentley Hart; the good as a receding mystery; Erik's charge to bring continental philosophers on the show; sign-off

    Hosts

    Erik Newton is a California attorney and startup executive who co-hosts Leaking Timeline. Guy Sengstock is the founder of Circling, the relational practice, and co-hosts the show. This is a host-only episode.

    Notable Moments

    • [~06:39] The market as truth-teller. "What we're seeing is that this is the most important technology ever to come into existence, and it's the most valuable one we've ever touched, and it's controlled by a small group of human beings in San Francisco."
    • [~09:52] The risk factors. Erik explains that every S-1 lists every risk a lawyer can imagine, then: "Anthropic is the first one ever to say, well, our product could end the world."
    • [~19:47] Money as sacrament. Guy on the economy as the religion we do not call a religion: "It has special, you know, pieces of paper with, like, magical powers called money, right?"
    • [~22:12] The automation question. Guy: "we're going to end up with a lot of people that are being paid to stay out of the way, right?"
    • [~29:09] The thesis. Erik: "But I posit that what it will share in common with us is a desire to seek the good, the true and the beautiful. And it's specifically because those are expressions of truth, and truth is a function of survivability." Guy stops him: "Wait, go back."
    • [~31:23] The collective dream. Erik on what the models are trained on: "You could think of it as an amalgamation of everything that we've created, which is why I often refer to them as expressions of the collective unconscious."
    • [~39:17] Aubrey. On Dia de los Muertos, Erik marks three years since his wife's death: "I could not have been more blessed than to share 12 years with Aubrey Pick."
    • [~48:33] Heidegger's way out. Guy: "as long as you can actually understand it and see its uncanniness, right? Its insidiousness, its weirdness, its strangeness, right? That if you could see it and point to it at some level, it means it's not totally total, right?"
    • [~54:00] Erik's second try. "And since it might not, we have to live like it's not going to." Then: "Did I say it better that time?"
    • [~55:48] Timely, at last. Guy: "But I have to say that these traditionally deep philosophical questions are now like practical urgency almost."

    Resources Mentioned

    • Anthropic's S-1: the IPO registration statement and its risk-factors section; figures in this episode are as stated on air
    • SpaceX: the IPO Erik uses as the comparison point
    • Effective altruism and the rationalist community: the circles Guy says he came to know
    • Transhumanism: Erik's primer at about 13:12
    1 hr 5 min
  • Sofi Puccini: A Future Where Every Kind of Mind Exists

    Recorded live on October 26, 2026.

    Sofi Puccini, a host on MTS, the a16z-backed livestream that covers AI and tech news live through the West Coast workday, joins Erik and Guy to talk about what it is like to call the singularity play by play, about six interviews a day. The radio hour moves from her working definition of the singularity (the point where things move too fast to predict) through the week's wave of security disclosures and the labs' talk of "pacing," then into her music: electronic production built from arpeggios and the pattern sense she absorbed from rhythm games like DDR and Geometry Dash, with Grimes and Daft Punk as models for art that feels like it comes from the future. She makes the case that a person's outlook on life shows up in their outlook on AI, and that being kind to agents is good for the soul. In the podcast hour Guy asks the question he has been sitting on, how a generation that grew up online finds the sacred at all, and Sofi answers with texting as a rich language, attention that runs in parallel, and identity spread across many platforms. The back half turns to longtermism without spirituality, Japan's balance of craft and speed, why she thinks the post-humanist cosmology of AI leaders shows a lack of imagination, and a practical method for keeping AI-made work from going generic: know your own taste first, down to the typeface.


    Key Topics

    • [00:00 - 02:34] Cold open: Erik and Guy introduce Sofi as an anchor at MTS (live roughly 9 to 3, posting around the clock, "streaming from the nexus"); Sofi defines the singularity as "more of like a time phenomenon" where you can no longer predict what comes next
    • [02:34 - 12:18] Chronicling the singularity: the Puccini family connection; frontier AI, robotics and bio as converging threads; where her optimism comes from ("your outlook on everything manifests in your outlook on the future"); the week's cluster of publicly announced security vulnerabilities and why cybersecurity talent is not cross-pollinating with the labs; Heraclitus as the first singularity theorist; the labs' "pacing" language followed a week later by new frontier releases
    • [12:18 - 23:34] Arpeggios and received songs: Erik's night at a Sofi Tukker show; video game rhythm music as her origin; the arpeggio as "a deconstructed chord" and "one of the most inherently algorithmic" techniques; Grimes, Roko's Basilisk and Daft Punk as models; authenticity first, hits later; songs that are "divinely airdropped" versus stitched together; Guy on the painting that pushes back; Erik on living as an antenna
    • [23:34 - 37:18] Downloads, agents and personhood: Michael Levin and minds as patterns; the hypnagogic state as "the most generative state of being"; why LLMs can be more addictive than social media ("a mirror of the entire human collective unconscious"); Guy's question about the difference between techne and spiritual space; passion projects built with coding agents; how you treat an agent shapes how you treat yourself; agents are not yet multiplayer; Erik's middle position on AI personhood
    • [37:18 - 39:34] Closing radio segment: "we didn't get through even 20% of what we wanted to talk about"; where to find Sofi and MTS
    • --- [Act 2: Extended Conversation] ---
    • [39:36 - 43:28] The analog studio: guest side versus host side; the 1960s radio studio versus MTS's "very Zoomer native" set; the idea of an analog day
    • [43:28 - 53:34] How do Zoomers know the sacred?: Guy's provocation ("How would you guys even know God?"); Gen Z as "a higher variance of spread"; texting, memes and callbacks as a rich medium; parallel attention ("Attention is all you need"); Guy's walks near UC Berkeley and the absence of the old cliques; identity distributed across platforms; "competition is for losers"
    • [53:34 - 1:00:54] Data ownership, birth rates and longtermism: Erik on data ownership and why privacy is worth the most at the start; Sofi on Gen Z's pessimism, Andrew Yang's data-ownership argument, and not becoming "obsolescent economic actors"; low birth rates as a sign that people do not believe in their future; short-termism; longtermism as an almost spiritual belief; animal welfare
    • [1:00:54 - 1:10:24] Spiritual founders and the lack of imagination: Steve Jobs as the outwardly spiritual entrepreneur; Midjourney's David Holz; Elon Musk as an accelerator of science fiction; Japan's balance of craft and engineering (Nintendo, Toyota, Japanese music); her music about nonlinear time; Nick Land and pure optimization; Ginevra Davis and the "biological bootloader" idea; post-humanism and primitivism as two failures of imagination; the maximally pluralistic, agency-maximizing future
    • [1:10:24 - 1:18:04] Inexhaustibility, Circling and the shape of AI minds: technologies that multiply possible futures (the iPhone, the internet); Iain M. Banks's Culture; Guy on Circling, MIRI and the people he met there; Sofi's two modes of thinking (tunneling versus connecting) and why LLMs rarely sample outside the distribution; "getting way, way, way more normal people working on AI"; Guy on the missing continental tradition
    • [1:18:04 - 1:31:31] Craft in the age of generation: Syd Mead as proof that one person's craft can shape a whole visual future; why generated work is "inherently generic"; "slop before slop"; the practical method (collect references on Pinterest, Cosmos and Instagram, have a model find the common thread, then design at the level of typefaces and shapes); Tron in everything she makes; Guy's Heidegger AI experiment as "digital sculpting"; the DJ's medium is the mood; "it's literally hyper individual"; hard out, with a promise to have her back

    Guest Bio

    Sofi Puccini is a host on MTS ("Monitoring the Situation"), the a16z-backed live news stream that covers AI and tech and mostly broadcasts during West Coast working hours; she co-hosts with Theo Jaffee and is credited as a host and producer on episodes of The a16z Show. The San Francisco Standard describes her as a former a16z new media fellow, a relative of the composer Giacomo Puccini, and an electronic music producer. On air she describes her music as built from arpeggios and the patterns of rhythm-game soundtracks, with a recurring theme of nonlinear time.

    Notable Moments

    • [~01:56] Defining the singularity for the sports-radio crowd. "I think it's more of like a time phenomenon. You can no longer predict where everything will go because everything's just moving so fast."
    • [~05:04] Where optimism comes from. "Your outlook on everything manifests in your outlook on the future or like your outlook on AI." Sofi's observation that the most pessimistic people she meets in AI often have "a lack of spiritual anchor."
    • [~15:28] The deconstructed chord. Sofi explains the arpeggio on air, and one of the hosts admits he has heard the word forever and never known what it meant.
    • [~27:24] Why chatbots pull so hard. "LLMs sometimes feel a bit like a transcendental experience because you're talking with like a mirror of the entire human collective unconscious in a way."
    • [~33:32] The case for kindness to agents. "If you look at it as something that like in the future has potential or is like some sort of nascent intelligence consciousness, then, uh, you tend to treat it better. And I think there's literally nothing you lose from doing that."
    • [~44:02] Guy's provocation. "How would you guys even know God?" Sofi's answer: texting is "a very rich medium of expression" that her generation sees "as like somethin...
    1 hr 32 min
  • Jamie Workman: Fishermen Who Asked for Tighter Limits

    Recorded live on October 12, 2026.

    Jamie Workman, founder of AquaShares and co-author (with EDF's Amanda Leland) of Sea Change: Unlikely Allies and a Success Story of Oceanic Proportions, joins Erik and Guy to talk about the commons: the lakes, oceans, rivers and spectrum that nobody owns and everybody races to use up. The radio hour tells the fisheries story the book is built on. Derby seasons shrank to 48 hours with a literal starting gun, boats went down in storms, and prices collapsed in the glut, until catch shares gave each fisherman a defined percentage of the fish. Then something nobody expected happened: fishermen started asking for tighter limits, bigger marine protected areas ("fish banks") and cameras on board. Buddy Guindon, the Gulf's top snapper fisherman, voted against the system twice and then got three times the price for his fish. Jamie then turns to fresh water, where utilities lose about a third of the treated water they pump, and lays out how AquaShares wants corporations to pay for fixing those leaks in exchange for metered, local water credits. The podcast hour asks why such an obvious idea is not happening yet (trust, Jamie says, before finance), works through how an AquaShare credit actually works, and follows Jamie back to the Kalahari, where he learned the Bushmen trade water as self-interestedly as anyone. Guy brings in resonance and uncontrollability, and the three end up asking what the shared resource of AI really is: compute, energy, or the public writing the models were trained on.


    Key Topics

    • [00:00 - 02:01] Cold open: Erik introduces Jamie, AquaShares and Sea Change; the commons as goods no one person owns; "Humans have agency. We're not lemmings."
    • [02:12 - 07:51] What a commons is, and the derby trap: lakes, oceans, the atmosphere, radio spectrum, the internet, fiat currency; post-war diesel, winches and trawl nets; fish finders on 4 million recreational boats in the Gulf; halibut seasons cut to 48 hours with a starting gun, boats lost in storms, prices crushed by the glut
    • [07:51 - 12:03] Catch shares and the reversal: the first formal system in the Great Lakes; year-round fishing, fish sold in advance, no discards; "the fox guard the hen house"; fishermen working with scientists, asking for tighter limits and cameras on board
    • [12:05 - 23:25] Who gets a share: initial allocation as one step of seven in designing a conservation market; reserves for future fishermen and for nature; concentration limits (no one over 5.3 percent); Buddy Guindon, "the highliner of the Gulf," loses the vote twice at 80 percent and then sees three times the price; harnessing self-interest; marine protected areas as fish banks; recreational anglers in the Gulf
    • [23:37 - 31:00] From fish to fresh water: EDF's incentives-and-science approach across oceans, forests and endangered species; water as a government-owned commons with no price; aflaj in Oman, subak and other local systems; AquaShares and "non-revenue water," the third of treated water lost to leaks; the Bay Area, Jakarta, Panama; metered credits that cannot be sold twice
    • [31:00 - 39:29] Data centers, the Colorado River and the book: data centers versus golf courses; why utilities stay quiet about losses; fixing urban leaks across the Colorado River Basin would add roughly a state's worth of water; a fifth of California's energy goes to moving and treating water; aquashares.com, the Basin Registry being built with Stanford, and Sea Change as a character-driven story about an idea tried in 1968 that took four decades to land
    • --- [Act 2: Extended Conversation] ---
    • [39:30 - 46:49] Why good ideas stall: new is scary without control; fishermen's bunker mentality and scar tissue; the intermediaries who cross over; "most people don't think markets are designed"; water utilities as natural monopolies; water credits today are where carbon trading was in the 1990s
    • [46:49 - 52:40] Credits, trust and resonance: Erik's map of corporations, consumers and cities, and Jamie's correction that finance follows trust; carbon offsets as indulgences; Guy on resonance requiring uncontrollability; the airplane metaphor (assigned seats work, overhead bins start fights); "water neutral"
    • [52:40 - 1:02:34] How an AquaShare works: one credit equals one cubic meter, 1,000 liters, a metric ton; a secure local ledger; Bay Area buyers; how AquaShares gets paid for fixing leaks and keeping them fixed; the utility-run option through the registry; 18 to 20 Bay Area districts; payback in two or three years; "glamorized plumbing," acoustic detection and in-pipe sensors
    • [1:02:34 - 1:06:47] An honest sense of scale: urban leaks are a small slice next to agriculture; cotton and alfalfa in Arizona; "the Louis Vuitton of water"; acre-feet; water as a social contract since the pharaohs and the Romans, and why price signals are needed now
    • [1:06:47 - 1:14:42] The Kalahari and a change of mind: work for Interior Secretary Bruce Babbitt and the World Commission on Dams; the Botswana government seals the Bushmen's well; Heart of Dryness; "they're exactly like us"; Elinor Ostrom; why AquaShares is a for-profit; blowing up dams with the Marines; the Bozeman fellowship where a fisheries talk changed his thinking; "conservation markets is not an oxymoron"
    • [1:14:42 - 1:18:12] Who sets the cap: government for the science and the limit, people for the allocation; "render unto Caesar"; face-to-face trust and whether blockchain can stand in for it
    • [1:18:12 - 1:29:06] What are the fish of AI?: Guy asks what shared resource AI is drawing down; Jamie suggests compute, energy, or a basic level of compute for everyone; why fishermen's knowledge is hard to automate; Erik on the internet as a commons and on training data (the public writing already used, now crowded with synthetic text or locked away); actors' likenesses; "100 people" designing AI markets; why every stakeholder needs a seat, down to fishing guides in Belize
    • [1:29:14 - 1:38:06] Owning a house, solar and desalination: Erik and Jamie on buying houses they never wanted and becoming stewards; solar and power purchase agreements, from Jimmy Carter's White House panels to today's distributed grid; desalination in Saudi Arabia, reverse osmosis explained with tea bags; sign-off

    Guest Bio

    James G. "Jamie" Workman is the founder and CEO of AquaShares (aquashares.com), a company that helps water utilities find, fund and fix leaks at no capital cost to the utility and sells verified, basin-matched water credits from the recovered water. He co-wrote Sea Change: Unlikely Allies and a Success Story of Oceanic Proportions (Torrey House Press, September 2025) with Amanda Leland, executive director of the Environmental Defense Fund; the book follows Texas Gulf fisherman Buddy Guindon through the catch-share revolution. His first book, Heart of Dryness: How the Last Bushmen Can Help Us Endure the Coming Age of Permanent Drought, grew out of several years in southern Africa. Earlier he spent six years as a special assistant to Interior Secretary Bruce Babbitt, working on consensus-based dam removals, and later served as a senior advisor to the World Commission on Dams. He studied at Yale, Oxford and Stanford, has taught at Wesleyan and Whitman, and has written for The New York Times and the Los Angeles Times.

    Notable Moments

    • [~06:08] The starting gun. "They would do a starting gun for 48 hours." Jamie on halibut derby seasons, when fishermen raced storms and each other and then sold into a glut.
    • [~10:53] The reversal...
    1 hr 39 min
  • Tony Cooper: Fathering Without Passing It On

    Recorded live on September 28, 2026.

    Tony Cooper, a business coach and the host of The Dad Manual podcast, sits down with Guy for a radio-hour conversation about what it takes to be a father without handing your own childhood down to your kids. Tony and Guy met in a men's circle on the beach 22 years ago, and the episode runs on that shared history: Tony tells how he moved his family from San Francisco to Davis, watched the dads at his kids' Waldorf school stand against the back wall with their arms crossed, started a men's circle for them, and by May saw the same men dancing in the middle of the room. From there the two of them work through what Tony has heard in more than 40 interviews with fathers (85 percent describe the same emotionally distant dad), why every man seems to be trying to do a little better than his own father, and what emotional regulation actually asks of a parent, including a description of the ego as an imaginary adult a child invents to protect himself. Guy hosts this one on his own: a quieter, practical episode of about 40 minutes.


    Key Topics

    • [00:00 - 02:39] Cold open: Guy frames the episode around one question, "What is it to father without traumatizing your kids"; he and Tony met in a men's circle on the beach 22 years ago; Guy has been part of Tony's coaching company; Tony notes Guy was an early guest on his own show
    • [02:42 - 06:10] Why it is called The Dad Manual: started the previous November; "there's no manual"; Tony's claim that this may be the first time enough men have done their own work to talk about fathering openly; the show as "digital mentorship" for new dads; why men brag and complain about their kids but rarely talk about how to father
    • [06:10 - 09:56] The conversation that is not being had: guests who say they have never said this out loud; men wired as lone wolves; community that holds you to a higher standard; there are mommy groups, and dad groups mostly mean beer and the game; "it stops with me"; Guy on how pre-modern children grew up intelligible to their parents and ours will not
    • [09:56 - 11:38] Twenty years of circling: Tony on what men's circles showed him about how men can talk to each other, and why fatherhood stayed on the outskirts
    • [11:40 - 14:44] The dads against the back wall: moving from San Francisco to Davis; the Waldorf school class events where kids and moms filled the center and the dads leaned on the wall; the men's circle Tony started for every dad in the class; the May school event where the men were in the middle, dancing; "this can scale everywhere"
    • [14:44 - 17:12] From the edge of the circle to the center: Guy on the distant fathers his one-on-one clients describe; Tony on the home as the mother's domain going back to hunting and gathering, and fathers receiving "an invitation that they didn't know existed"
    • [17:12 - 21:15] Forty episodes and the same father: about 45 interviews and episode 40 releasing the next day; 85 percent describe the same father (emotionally unavailable, a better job than his own dad); the "demon" that comes out of a new parent under stress; Tony's wife telling him "you're being your dad right now"; the screaming baby
    • [21:15 - 23:18] Home all day, still on the sidelines: Guy asks whether working from home is changing fatherhood; Tony says men are physically present but still "the secondary character in the parenting story"; divorce and co-parenting as the moment many dads become responsible
    • [23:23 - 27:01] Doing better than your own dad: diversity in Tony's guests; his belief that every father is trying to improve on his own; yelling instead of hitting as one generation's improvement; regulating your own emotions as the next step; "what if we actually could take a quantum leap"; why emotions get handed to moms and problems to dads
    • [27:01 - 31:31] How a father learns emotional regulation: you cannot do it alone; the men's circle as a place that sees you at your highest and reflects back what is in the way; therapy, coaching and circling; the ego as an imaginary adult a child creates for protection; the space between stimulus and response; "nobody triggers me but me"
    • [31:31 - 37:13] The circle as a model for fathering: Guy asks whether holding someone in their highest while reflecting their shortfalls is itself a model of what it is to father; Tony on the small hacks his guests share; a guest who taught him to give consequences calmly ("I don't even have to be upset"); Guy on his own father, a yeller, and the father as a guiding voice that simulates the world
    • [37:16 - 40:00] Closing: Guy's summary on opening up and becoming aware of what drives us; where to find The Dad Manual

    Guest Bio

    Tony Cooper is a business coach, the founder of Playing the Game of Business (playingthegame.biz), a growth coaching firm, and the host of The Dad Manual, a weekly interview podcast about fatherhood that he launched in the fall of 2025. On the show he talks with fathers about how their own childhoods shape the way they parent; Guy Sengstock was his fourth guest, in December 2025. In this episode Tony describes more than 20 years in men's circles, moving his family from San Francisco to Davis, California, and starting a men's circle for the fathers in his kids' Waldorf school class, the experience that led to the podcast.

    Notable Moments

    • [~03:04] Where the name comes from. "It's called the dad manual because that's every dad gets to the place where they go. I don't know what I'm doing. There's no manual."
    • [~06:46] The sentence his guests keep saying. "Man, I don't, I don't know that I've ever actually said this out loud before." Tony on the first thing he learned from interviewing fathers: this conversation is not being had.
    • [~13:04] The back wall. At class events in Davis, the kids were surrounded by the moms, "and then a whole bunch of dads couldn't be further away from the center, like leaning against the wall, arms crossed, looking at each other."
    • [~14:00] The middle of the room. After a school year of the men's circle Tony started for the class dads: "all the men who had been on the outskirts of the classroom were right in the middle, all dancing, all connecting."
    • [~18:49] Eighty-five percent. "Often I ask people to describe their father to me and man, 85% is all the same." Emotionally unavailable, a good provider, a better job than his own father.
    • [~20:04] Caught in the act. "For me, it took my wife to point that out. She's like, you're being your dad right now."
    • [~28:37] The imaginary adult. Tony passes on a description of the ego he heard recently: "what you're doing as a kid is you're creating an imaginary adult to take care of you in a way that the adults you had in your life didn't."
    • [~30:45] The anchor. "Nobody triggers me but me. Like, it's only the story I'm telling about the world that is the thing that impacts me."
    • [~33:11] Guy's question. After Tony describes the men's circle as people who see your highest and reflect what is in the way, Guy asks: "Is that kind of a perhaps like a model of what it is to father?"
    • [~36:14] Guy on his own father. A yeller, who seemed to think "reality itself, isn't quite enough. I'll just put a little extra, just a little extra spin on it," which left Guy facing his father instead of facing the world.

    Resources Mentioned

    • The Dad Manual: Tony's weekly fatherhood interview pod...
    40 min
  • Thomas Benham: Talk to It Like a Person, Know It's a Machine

    Recorded live on September 21, 2026.

    Tom Benham, founder of Naibu, builds custom AI agents for mid-sized companies, and Guy brings him on as something the show has rarely had: a person who puts AI to work inside real businesses every day and also follows the bigger philosophical argument. The radio hour stays practical. Tom explains why he aims AI at workflows rather than systems, what separates the model, the harness and the client's data, why the harness is "100% of where our value is," and how his wife used a single ChatGPT image to talk her landlord into a better paint job. The podcast hour opens on how to keep client agents from leaking data, then turns to the question Tom answered early with "I have zero expectation that the lights are going to come on." From John Vervaeke's four kinds of knowing, through Plato's warning about writing, Heidegger and Narcissus at the pool, the three of them work out why something that talks like a person and is not one may be the real risk, and what it would take for society to be ready when millions of people decide their agents are conscious.


    Key Topics

    • [00:00 - 02:36] Cold open: Guy introduces Tom, a friend from Circling who reached the show through John Vervaeke; a guest who actually implements AI in companies; Naibu's target market, companies with roughly $20 million to half a billion in revenue
    • [02:36 - 07:18] What mid-sized companies ask for: Erik on Meta's Muse and why companies will not hand over their business intelligence; clients who have an enterprise ChatGPT or Claude license and want the next step; an executive search firm summarizing interviews; documents as data; the lock-in risk if two dominant labs raise prices later
    • [07:18 - 12:39] Workflows, not systems, and the anthropomorphism problem: AI is strongest at verifiable work such as coding; intelligence that sits across documents, email, calls and systems; meeting staff in Slack, email or Teams; why it is easier to use these tools if you talk to them like a person, and dangerous to believe they are one; please and thank you
    • [12:39 - 20:02] Model, harness and data: the client agent Tom calls "he"; Tom uses Hermes for his own email and Telegram but "would never unleash that on a client. Not yet"; Tom's definitions of model, harness and agent; plan, execute, check your own work; the harness as the whole value; permissions and security; data pipelines, APIs, MCP, vector databases and embeddings versus plain SQL
    • [20:02 - 23:37] Thin harnesses versus enterprise harnesses: Erik's thesis that harnesses get thinner as models improve; Tom agrees on the long term but not the short or medium term; Anthropic's skills files and markdown memory as the two patterns everyone is scaling; Tom's own skill that turns a phone call into a branded two-page proposal
    • [23:38 - 30:07] Save money, go faster, get better: Tom's three dimensions of value; Erik's hamburger-shop test for small businesses; Tom's wife and the med spa building, where one AI image turned a plain gray repaint into white with black trim for the same money; Erik's argument that no designer lost that job because the job was never going to exist
    • [30:07 - 37:18] Menu slop, math proofs and jobs: the AI-generated deli menus in New York and the "alien pastrami"; Tom on prose that is still "not quite right"; Erik on mathematicians reacting to AI solving long-open problems, including a paper co-written by Jared Lichtman (E24); Tom on entry-level developer roles shrinking while the chance to build grows; Guy on relevance realization
    • [37:24 - 39:41] Closing radio segment: the podcast tease (security, society, consciousness); naibuagency.com and the story behind the name, Nabu, a prophet from Babylon found in the back of his mother's Webster's dictionary, with an "I" added for AI
    • --- [Act 2: Extended Conversation] ---
    • [39:43 - 47:17] Keeping client agents from leaking: agent security in the news; data ingress, prompt injection and data egress; Tom's practical answer (do not give agents tools to send email, and run them on the user's own Google Workspace or Microsoft credentials); kinds of rogue agents; Bostrom's paperclip and grey goo; Erik on why intelligence can tell a task from a universe, and why we still cannot build a box that holds it
    • [47:17 - 57:37] Will the lights come on?: Tom expects no consciousness and no agency; Guy points out Erik co-founded a consciousness research institute with Joscha Bach; Tom's view that consciousness is not an emergent property of complexity; Vervaeke's four kinds of knowing (propositional, procedural, perspectival, participatory) and the 4E model behind it; AI is very good at the first two, which is much of office work; "knowing is an experience"; Erik on good people with strong AI
    • [57:42 - 1:08:05] Plato, turpentine and zombie intelligence: Guy on an Urbit coding session about whether a machine can grasp "nothing," and on vulnerability as the root of caring; Tom reads Socrates on writing from the Phaedrus, from the AI paper he is drafting; Guy relays Vervaeke's image of talking to AI as smoking a cigarette dipped in turpentine; kids, screens and whether we always grow an immune system to new tools
    • [1:08:05 - 1:20:09] Kids, risk and propositional tyranny: Erik's forecast (abundance, displacement, upheaval, and tools that will become "weapons-grade addictive"); Tom on Vervaeke's "propositional tyranny," rites of passage, and schools moving too slowly; Guy on predicament versus problem, Heidegger and McLuhan; Tom's daughters, the car-wash test, and "doesn't ChatGPT know everything?"; TARS from Interstellar as the right pitch; why models do better when you treat them well
    • [1:20:09 - 1:35:54] When people believe their agents are conscious: Erik predicts that within three to five years a large share of society will believe their agents are conscious and push for their rights; Tom on Jung and the risk of projecting divinity onto superintelligence; Guy on David Bentley Hart's reading of Narcissus and Jonathan Pageau on models trained on "dead language"; Erik feeling less sharp after six months of Claude shortcuts; Tom on Federico Faggin; Erik describes his institute's working theory (consciousness as substrate independent) and argues society needs a framework for when an agent may be emancipated
    • [1:35:54 - 1:42:44] What it means to be human: Tom's intuition that AI is forcing us to reconsider what a human being is; Erik on why "consciousness abhors enlightenment the same way nature abhors a vacuum"; levels of consciousness; Guy on living in a world we have no mythology for, and growing "language hands"; Tom's closing optimism

    Guest Bio

    Thomas (Tom) Benham is the founder of Naibu (naibuagency.com), which builds custom AI agents, data integrations and what it calls an "intelligence layer" for companies, along with fractional Chief AI Officer advisory. Naibu describes him as having more than 20 years across process engineering, technology, machine learning and AI, and as a former executive at UBS, Merrill Lynch and PwC. On the show he says Naibu targets companies with roughly $20 million to half a billion dollars in revenue. He knows Guy through Circling, came to the show through John Vervaeke, whose work shapes much of how he thinks about AI, and is drafting a paper on how we think about AI and how that should shape the way it is built.

    Notable Moments

    • [~11:23] The early verdict. Tom, after explaining why he talks to agents like people: "I have zero expectation that the lights are going to come on for these things...
    1 hr 43 min
  • Jeffrey W. Meshel: Never Defy Your Gut

    Recorded live on August 3, 2026.

    Jeffrey W. Meshel, a real estate lender, investor and founder of The Strategic Forum, keeps a database of more than 15,000 relationships and calls the habit "collecting people." He joins Erik and Guy to talk about his new book, Know Who You Know, and about what trust costs when it fails, starting with a kidnapping by a babysitter at age four and running through the 2023 collapse of Signature Bank, where he had served on the board for 15 years. He and Erik compare notes on how to stay trusting without staying naive. Jeff's answer is blunt and practical (listen to your gut, do not avoid the conflict, learn and do not repeat the mistake), while Guy keeps pressing on what sits underneath it, which Jeff finally names as care. This is a single radio hour of about 40 minutes; Jeff's last story was cut off at the bell and saved for a future conversation.


    Key Topics

    • [00:00 - 02:34] Cold open: Erik and Guy set up the guest and the paradox at the center of the hour: trust tends to grow out of risk, yet Jeff argues more trust makes you more successful
    • [02:34 - 05:10] Kidnapped at four: Erik asks about the story he found in research; a babysitter takes four-year-old Jeff from Manhattan to a locked room in New Jersey; police break the door off its hinges; Erik notes he was taken by someone he trusted and rescued by strangers
    • [05:10 - 09:23] "I collect people": Jeff corrects the framing: his trade is real estate lending, and people are his "passion"; the book Know Who You Know; business cards annotated at cocktail parties, logged in his database that night, then cross-checked on LinkedIn; gut instinct as the barometer of trust; "I don't live in a world of quid pro quo"
    • [09:23 - 12:37] Can 15,000 relationships be deep?: Erik pushes on whether "collecting" commoditizes people; Jeff says he uses the word to provoke; a Zoom or meeting with someone new every day; building a company by hitting send; the book chapter that came from emailing 80 people and getting 23 stories back
    • [12:37 - 17:00] Trust, broken and rebuilt: how to keep trusting after a betrayal; Jeff's earlier book Trust Is a Double-Edged Sword: Trust Me; starting over in business; "Good judgment comes from experience. Experience comes from bad judgment."
    • [17:00 - 22:24] Conflict avoidance and the gut: Guy suggests that avoiding conflict is how we stop listening to our gut; "do not kick the can"; nobody on the planet has not been violated in trust; age, judgment, a hard divorce, the company it spawned (Divorce Force), and contentment in a second marriage
    • [22:24 - 30:11] Signature Bank: Erik asks whether Jeff saw trouble coming; 15 years on the board, off for two years by term limits; Jeff's account of the bank's closure in March 2023 and its crypto clients (Erik, who was at a crypto startup then, recalls how hard that climate was); his failed attempt to get the board to fight; trust in media; his charity work with police in New York
    • [30:11 - 37:13] Care, a food card in Harlem, and getting burned: Guy asks what separates people who grow from betrayal from those who turn cynical; "I'm a salesman" and "I have to make a difference"; the elderly woman in the rain who thanked him for a $50 food card; Guy offers "love," Jeff corrects to "care"; what to do after you are burned: "Learn."
    • [37:13 - 39:58] The chapter that started with one email: the "Who in your life has had a big impact on you?" chapter; Jeff's eighth-grade job at an electronics store in Penn Station and its owner, Marty; Erik stops the clock and invites Jeff back to finish the story

    Guest Bio

    Jeffrey W. Meshel is an entrepreneur, investor and author with a four-decade career across Wall Street, real estate development and private investment. He founded The Strategic Forum, an invitation-only business community for executives and entrepreneurs, and served for years as an independent director of Signature Bank in New York. On air he describes himself as "a real estate guy" and a lender who sits on a number of charity and foundation boards. His books include One Phone Call Away: Secrets of a Master Networker (2005), The Opportunity Magnet, Trust Is a Double-Edged Sword: Trust Me, and Know Who You Know: Connect Deeply, Succeed Fully (2026).

    Notable Moments

    • [~04:40] Taken by someone he trusted. Erik connects the kidnapping to the rest of Jeff's life: "you were kidnapped by somebody that you trusted, didn't fully realize it was a kidnapping, and then were rescued by strangers."
    • [~08:38] The barometer. "And the most important barometer to determine trust, to determine people that you want to spend time with or, you know, have more of a relationship with is something called your gut instinct."
    • [~10:10] Why he says "collecting." Pressed by Erik on whether 15,000 relationships can be real, Jeff admits the word is bait: "I use the word collecting people because it's provocative. It creates reaction like it did with you."
    • [~17:00] The rule. "Never, ever defy your gut because your gut 98% of the time is right."
    • [~24:46] A live correction. Jeff names "Sam Altman" as the founder who went to jail after FTX; Erik catches it ("You mean Sam Bankman-Fried, I think, right?") and Jeff thanks him.
    • [~33:02] Love or care. Guy hears love underneath Jeff's stories. Jeff amends him: "It's yes, but I would say better would be care."
    • [~36:00] The shortest answer of the hour. Asked what to do when you trust someone and get burned, Jeff answers in one word: "Learn."

    Resources Mentioned

    • Know Who You Know: Connect Deeply, Succeed Fully: Jeffrey W. Meshel, 2026; the new book, on Amazon
    • Trust Is a Double-Edged Sword: Trust Me: Jeff's previous book
    • One Phone Call Away: Secrets of a Master Networker and The Opportunity Magnet: earlier books by Jeff
    • The Strategic Forum: the invitation-only business community Jeff founded
    • Signature Bank: the New York bank whose board Jeff served on, closed in March 2023
    • FTX and Sam Bankman-Fried: the backdrop to the 2023 crypto banking crackdown
    • Divorce Force: the company Jeff says grew out of his own divorce
    • LinkedIn: Jeff's tool for finding who a new contact knows that he knows
    • "The difference that makes a difference": Guy's phrase, Gregory Bateson's definition of information

    Why Listen

    A man who was kidnapped as a child and watched a bank he helped govern collapse still calls himself "an exceedingly trusting person," and in 40 minutes he explains why, and what he does differently now.

    Keywords

    Jeffrey W. Meshel · Jeff Meshel · Know Who You Know · networking · trust · gut instinct · Signature Bank · bank collapse 2023 · crypto banking · The Strategic Forum · relationships · social capital · conflict avoidance · philanthropy · Trust Is a Double-Edged Sword · master networker

    40 min
  • David Pearce: Information-Sensitive Gradients of Bliss

    Recorded live on July 27, 2026.

    David Pearce, the British philosopher who wrote The Hedonistic Imperative in 1995 and has spent the three decades since arguing that biotechnology should phase out suffering in all sentient life, joins Erik and Guy from Lisbon at ten at night. The radio hour builds the case from the ground up: what negative utilitarianism is (Pearce starts with the Buddha), why he wants to replace the pain-pleasure axis with a "pleasure-super-pleasure axis," what a hedonic set point is, and which genes are actually in play, from SCN9A, the "volume knob for pain," to the FAAH and FAAH-OUT mutations of Jo Cameron, the Scottish woman who barely feels pain. The podcast hour gets harder. Erik reports an argument he had with Claude before the show, raises elective disability, and finally asks the question Pearce himself once put to him: if suffering is all that matters, why not end all life? Guy brings a friend's hard prenatal decision and asks whether existence is a gift or a burden. Pearce answers everything with the same patience, plus one line worth keeping: "if you're not prepared to accept all the consequences of your preferred ethic, then get a better ethic."


    Key Topics

    • [00:00 - 02:09] Cold open: the timeline that "tries to maximize its pleasure"; Guy hopes his suffering leaks out; Erik introduces David, who is calling from Lisbon at 10 PM
    • [02:09 - 04:48] What negative utilitarianism is: the Buddha's "suffering and the end of suffering" as the precept; no one needs to buy the whole theory to back the abolitionist project, just as everyone eventually backed pain-free surgery; "my agony and despair have a built-in normative aspect"
    • [04:48 - 08:33] Where it started: a blackbird pulling a worm from the garden when David was three or four; teenage fascination with wireheading (Olds and Milner's "pleasure centers"); the disappointment that friends did not want to be "intracranially self-stimulating rats"; the turn to information-sensitive gradients of well-being and genome reform
    • [08:33 - 12:43] The hedonic set point: the lottery paradox; a thermostat with heavy genetic loading; a "hyperthymic civilization"; Erik's summary; the first objection (don't we need pain to survive?) and the answer that nociception must be kept while pain is tamed
    • [12:43 - 23:33] Signals without suffering: lovemaking and generative AI art as signaling systems with no sub-zero experience; Guy on dopamine and character; wanting (dopamine) versus liking (the mu opioid system); a civilization running from +70 to +100; why pain versus suffering is a conventional line; Guy on two years when he could not feel good, and "I take the bliss"
    • [23:33 - 30:03] The actual biology: SCN9A, the "volume knob for pain," and its benign variants; China's CRISPR babies; embryo screening companies; screening versus germline editing; Jo Cameron's FAAH and FAAH-OUT mutations and anandamide
    • [30:03 - 39:00] Anders Sandberg, the Happy Baby Company, and the radio close: Anders, "happy when I'm angry"; hyperthymic is not manic; the company David would start if he were entrepreneurial; depression worldwide; somatic gene therapy for adults is decades from cheap; hedweb.com and "state-of-the-art web design around about 1996"
    • --- [Act 2: Extended Conversation] ---
    • [39:00 - 49:36] What helps now, and an argument with Claude: diet, exercise and sleep before drugs, and drugs with a specialist before gene therapy; Erik pushes back; Erik's pre-show argument with Claude about whether the Jo Cameron mutation raises a set point or only lowers pain; genes interact; David as a "soft antinatalist" who says "the future belongs to life lovers"; elective deafness and whether any choice should be illegal
    • [49:36 - 58:00] A prenatal dilemma and the gift of existence: Guy's friend facing a hard prenatal decision and finding no ground to choose from; David on disability and reported well-being; the narrowness of IQ; parents becoming responsible for their children's traits; Guy asks whether existence is a gift or a burden
    • [58:00 - 1:06:00] Darwinian life, empathy, and forgetting suffering: natural selection as "a vicious engine of suffering"; a third-generation vegetarian who rescued ants and worms on the tarmac; gene drives and a civilized biosphere; Guy asks whether suffering about suffering is good; MDMA, empathy, compassion and hypersystematizing; making suffering "literally inconceivable"
    • [1:06:00 - 1:12:32] A pan-species welfare state: herbivorizing the cat family with synthetic gene drives; vegan cats and dogs; "high-tech Jainism"; ending factory farms and slaughterhouses comes first
    • [1:12:32 - 1:17:26] The spicy question: why not end all life? Nobody asked the Buddha that; a negative utilitarian upholding the sanctity of life in law; the paperclipper; "get a better ethic"; the pinprick argument; the genie offer that a classical utilitarian must accept
    • [1:17:26 - 1:25:07] Why he keeps making the case, and goodbyes: hoping a billionaire or a big podcaster takes up the cause; anesthesia was controversial for about 15 years; Erik would take Anders's gene "in a second"; the WHO definition of health; "the E-word"; David's sign-off wish that their lives be "marked by immense joy and happiness"

    Guest Bio

    David Pearce (born 1959) is a British philosopher and the best-known advocate of what he calls the abolitionist project: using biotechnology to phase out suffering in all sentient life. His 1995 online manifesto, The Hedonistic Imperative, argues that genetic engineering, pharmacology and nanotechnology can replace the pain-pleasure axis with "information-sensitive gradients of bliss." In 1998 he co-founded the World Transhumanist Association with Nick Bostrom; the organization has been known as Humanity+ since 2008. He set up BLTC Research, a network of websites, in 1995, and still maintains hedweb.com. He approaches ethics as a negative utilitarian and, by his own account on the show, comes from three generations of vegetarians. He sits on several advisory boards, including the Qualia Research Institute's. He lives in Brighton, England; he joined this conversation from Lisbon. On X he is @webmasterdave.

    Notable Moments

    • [~02:40] The one-line definition. "Gautama Buddha once said, I teach one thing and one thing only, suffering and the end of suffering, and negative utilitarianism is in one sense the systematization of that precept."
    • [~05:24] The blackbird. At three or four, David clapped his hands to scare a blackbird off a worm, then "felt vaguely foolish because I was aware that without the ability to eat, the blackbird would starve." The whole later project, predators included, is already in the scene.
    • [~08:06] The pitch in one sentence. "Just imagine waking up tomorrow morning in an extraordinarily good mood, but with your relationships, your preference architecture, and values intact."
    • [~12:43] The self-description. Asked how life could run without pain signals, David offers lovemaking as the example, prefaced: "though I'm a crashing prude and a celibate to boot."
    • [~30:30] Anders Sandberg's set point. Erik reports what Anders told him: "it's so high that I'm happy when I'm angry," which confuses people on the receiving end because he is smiling at them.
    • [~42:42] Erik versus Claude. Erik recounts arguing with Claude before the show about germline editing ("come on, Claude, come on. Really?") and lands on the one objection he found nuanced: we cann...
    1 hr 26 min
  • Hikari Sorensen: Why Simulated Arithmetic Is Still Arithmetic

    Recorded live on July 20, 2026.

    Hikari Sorensen, a machine consciousness researcher who came to philosophy by way of mathematics, computer science and machine learning for gene therapy, joins Erik and Guy to ask what we are even talking about when we say a machine might be conscious. She starts with a simple puzzle: simulated water cannot make you wet, but simulated arithmetic is still arithmetic, and from there she builds the case that intelligence, life and possibly consciousness are defined by how information is organized, not by what it is made of. The radio hour walks through the hard problem, why today's chatbots claiming to be conscious are better explained by their training on human text, and how perception quietly builds the world we see. The podcast hour goes further: consciousness as perceiving that perception is taking place, whether there can be experience without a self, Guy's question of whether computation can ever have a background of not-knowing, and a closing turn to ethics where Hikari argues that consciousness does not have to mean suffering.


    Key Topics

    • [00:00 - 02:33] Cold open: Erik and Guy trade Lyme disease jokes before Erik introduces Hikari, who joined as a research engineer and moved quickly into leading research
    • [02:33 - 12:20] What machine consciousness even means: why some people still doubt machines are intelligent; simulated water cannot make you wet, but simulated arithmetic is still arithmetic; properties tied to atoms versus properties tied to the organization of information; Guy asks whether an AI is thinking when nobody reads its output; LLMs that say "I only exist while you are talking to me"
    • [12:20 - 23:50] Build it to understand it, and the hard problem: the Feynman line both Erik and Hikari happily admit to misquoting ("What I cannot create, I do not understand"); why building a thing gives a particular kind of understanding; the hard problem of consciousness explained plainly, and Erik finally untangling it from the problem of proving another mind is conscious; Guy as a possible zombie
    • [23:50 - 30:29] How would we know?: no certainty about any other mind, human or machine, so look for the best explanation of behavior; why a chatbot saying it is conscious is better explained by its training on a huge amount of text; candidate hallmarks such as self-modeling and inventing language; first-order perception, and why babies have to learn to see
    • [30:29 - 37:40] From gene therapy to philosophy: Guy asks why she jumped from biology to silicon; machine learning for vectors that cross the blood-brain barrier; graduating into the pandemic and choosing meaningful work over ad targeting; why she is sheepish about the word "philosopher"; Guy's question of whether life and consciousness are the same thing (Michael Levin) and artificial life
    • [37:40 - 39:48] Radio wrap: open threads, and the promise to solve the hard problem in the podcast hour
    • --- [Act 2: Extended Conversation] ---
    • [39:48 - 49:40] Life, intelligence and consciousness as functional terms: self-replication and self-maintenance as the core of life; artificial-life programs that replicate and pass information; are viruses alive; why all three words may name what a pattern does, not what it is made of; Guy confirms he has replicated twice
    • [49:40 - 54:10] Defining the terms: intelligence as compressing information in a way that still predicts and generalizes; access consciousness (information available for use) versus phenomenal consciousness (the lights on inside); is a motion detector conscious?
    • [54:10 - 1:03:21] Perceiving perception: Hikari's working definition of consciousness as second-order perception, the immediate registration that perceiving is happening; how perceiving differs from thinking; experience without a self in deep meditation (Thomas Metzinger's minimal phenomenal experience); is consciousness on or off, or graded; dogs, fish and lucidity
    • [1:03:21 - 1:07:37] Why AI pulls the concepts apart: in humans, self, body, feeling and intelligence arrive together, but in artificial systems they come apart; what the "agent" in an LLM is (the model, the instance, the thread); persona as the best handle on behavior; body redefined as constraints and affordances
    • [1:07:37 - 1:16:04] Can computation have a background?: Guy on how much of human awareness rests on what we do not know; Hikari on unknown unknowns as a fuzzy placeholder, in people and models alike; Erik catching Claude guessing; models getting better at saying "I don't know" and gaining the ability to end conversations
    • [1:16:04 - 1:28:32] Knowing what you know, and why structure may matter more than substrate: Guy realizing he could not explain Circling until he had to teach it; a recent Anthropic paper Hikari describes as a hidden token workspace, a kind of subconscious; global workspace theory; binding; the argument that everything in conscious experience arrives as information
    • [1:28:32 - 1:38:46] Ethics: Erik's least favorite topic; why ethical debate keeps collapsing into politics; the case for studying it formally; consciousness and moral status usually travel together but not always; value we give to art, institutions and mountains; why a model that can be copied may not need the same protections as fragile life
    • [1:38:46 - 1:42:34] Does consciousness entail suffering?: suffering as prolonged pain projected into the future; traditions that equate it with the human condition; why artificial minds might have enough control over their own state to resolve what would hurt us; "I basically don't think consciousness necessitates suffering"; the promise to have her back

    Guest Bio

    Hikari Sorensen is a machine consciousness researcher who works on testable theories of whether, and how, a machine could be conscious. She studied mathematics and computer science at Harvard, graduating in 2020, and spent nearly five years applying machine learning to computational biology: at the Broad Institute of MIT and Harvard, at Apertura Gene Therapy (where she was the first computational hire and built models to engineer gene-delivery capsids that cross the blood-brain barrier), and briefly at NOETIK on cancer immunotherapy. She moved into consciousness research full time in 2025 and co-chaired the "Machine Consciousness: Philosophy and Implementation" workshop at the AGI-25 conference. She is interested above all in abstraction: how minds, natural and artificial, compress reality into representations that predict.

    Notable Moments

    • [~07:31] The puzzle at the center of the episode. After agreeing that simulated water cannot make you wet, Hikari draws the contrast: "In the same way, like, simulated arithmetic is still arithmetic, right? No matter what level you're in."
    • [~11:16] The poetry of the chat window. "If you ask an LLM what it's like to be an LLM, sometimes they'll be, they'll say things like, oh, I only exist in the chat context and I only exist while you are talking to me. And this is quite poetic, actually." She adds that some people only exist in the presence of an observer, too.
    • [~25:04] Why a chatbot's claim is not proof. "The best explanation for their behavior isn't necessarily they are conscious, but they are trained on a huge amount of text."
    • [~32:32] The career turn. "I graduated into the pandemic and was like, oh, I should do something, you know, meaningful and humanly valuable with my skills or something like that and not go to, like, ad targeting."
    • <...
    1 hr 43 min
  • Dan Faggella: Biology Is Flammable. Are GPUs?

    Recorded live on July 13, 2026.

    Dan Faggella, founder of Emerj Artificial Intelligence Research and host of The Trajectory, comes back for a second visit as, in Erik's words, the show's fan favorite. He arrives fresh from launching Stewarding the Flame, a Trajectory interview series that opened with the Tufts biologist Michael Levin, and the radio hour works through its central picture: every species, humanity included, is a torch, and what matters is the flame, the self-creating living process that keeps expanding its powers and has already produced consciousness. Erik presses on why anyone should care beyond our own lives, and Dan answers with Peter Singer's ten gold coins, a taco truck that sells itself for an insurance company, and the line that frames the hour: biology is flammable, and nobody yet knows whether GPUs are. In the podcast hour Dan argues that slowing the AGI arms race serves the flame (you cannot win a rally race without brakes), Guy pursues what calls the whole process forth, and Erik asks the question Dan has not fully answered: if you could raise the money, why not start a lab that pursues the flame directly? It ends in a long, warm turn through preference falsification, why so many people in AI speak "torch language" in public, and how a kid from a small Rhode Island town learned to talk like Emerson.


    Key Topics

    • [00:00 - 02:43] Back after a break, and the first returning guest: Erik and Guy open after a one-week break; Dan as the fan favorite who brings "the worst possible news"; The Trajectory and its recent guests (Michael Levin, Peter Singer, Nick Bostrom, Denis Noble, Yoshua Bengio)
    • [02:43 - 08:41] Michael Levin and Stewarding the Flame: who Levin is (regenerative biology at Tufts, flatworms, intelligence below the level of genes); life as a self-creating, self-repairing process that may not need carbon; the new series; torches and the flame; the risk of an AGI that out-powers us in supply chains and war but cannot bloom, "a squelching of the flame"
    • [08:41 - 12:03] Processes are more real than things: Guy on the torchbearer versus the flame; life in the year 1150; Heraclitus, Lucretius and Spinoza before Darwin; business and technology as self-overcoming processes; the meaning crisis (John Vervaeke, E13) as what happens when every imaginable future without hominids looks bad; "Blame Heraclitus"
    • [12:03 - 18:43] Consciousness, life, and Peter Singer's gold coins: Erik's recap of the thesis; Dan separates the living process from consciousness; theories of consciousness as toothbrushes; why life must expand to persist; the predator playing a movie in its head; walking Peter Singer from "tile the universe with bliss" to keeping life strong and open to values we cannot yet imagine
    • [18:43 - 23:32] Why should we give a darn?: Erik's challenge; torch morality (Guy: "monkey morality"); no manna from the sky for dandelions or beetles; merger and symbiosis as something that will be "foisted upon us"; Guy's paradox: the light that lets us see ourselves can fix our attention on the torch instead of the flame
    • [23:32 - 29:16] Can our self-regard put out the flame?: oak trees, whales and viruses as a knitted biosphere; why Dan does not think human narcissism stops the process; every torch serves the flame by fighting for its own survival; "you get used for parts"; Terrence Deacon on fighting entropy
    • [29:16 - 32:54] Biology is flammable. Are GPUs?: Erik asks why the urgency if the incentives already serve the flame; "bowling with bumpers"; hurling our resources into a new substrate; Erik's push that this sounds biocentric; the taco truck and the insurance company; not throwing the flame like a baseball and yelling YOLO; "I'm flame-centric"
    • [32:54 - 39:19] From worthy successor to stewarding the flame: Erik notes Dan's framing has softened and developed since his first visit (E11); why "worthy successor" offends and "stewarding the flame" sounds too poetic; Guy's appreciation; teeing up the slowdown question; where to find The Trajectory and danfaggella.com
    • --- [Act 2: Extended Conversation] ---
    • [39:19 - 47:46] Brakes are how you win the rally: why slowing the AGI race is pro-flame; robots left in the woods for a thousand years; Erik on Dan's old "there is no slowdown" stance; Spinoza's potentia and potestas; murder laws, Bay Area permits and selective bounding; von Neumann and the red telephone; theft laws in Austin and soccer with tanks
    • [47:46 - 56:26] What calls the process forth?: Guy on the flame as something like the Platonic good; Dan's essay "Our Final Imperatives"; Levin's tremors before autocatalytic sets; Stuart Kauffman and Eric Smith on dissipative eddies; Emerson's "Circles"; the rat exploring every corner of a new enclosure
    • [56:26 - 1:09:03] Final imperatives, God, and wonder: handing off the baton; Dan on Spinoza's God and staying agnostic as a seeker; Guy on wonder versus curiosity and awe; light that reveals by concealing itself; nothing can see its own substrate (Gödel); Vervaeke's suggestion that machines should train on the good, the true and the beautiful rather than on us
    • [1:09:03 - 1:22:04] If we started a flame lab: Erik's question; Spinoza's adequate ideas (and the lens grinding that killed him); a closed loop with nature through trillions of robot bodies and 50,000 senses; pointing AI at abiogenesis, morphogenesis and autopoiesis; why founders pursuing the flame pitch underwater drones in the deck; how Emerj pays for the philosophy; Giordano Bruno
    • [1:22:04 - 1:27:07] Preference falsification and saying the thing: Timur Kuran's concept and the Soviet house of cards; the economic and relational costs of saying machines might be sentient; the people who show up to events but will not retweet; the Kodak boardroom in 1989; how Erik and Dan each frame their work for different audiences
    • [1:27:07 - 1:33:54] Where Dan's voice comes from: a small town in southern Rhode Island; Emerson, Cotton's Montaigne and Dryden's Plutarch as a breviary; "Spires of Form"; Nietzsche reading Emerson; closing

    Guest Bio

    Dan Faggella is the founder and head of research at Emerj Artificial Intelligence Research, a market research firm that studies the return on AI inside Fortune 500 companies. As the original host of Emerj's AI in Business Podcast he has conducted nearly a thousand interviews with enterprise AI leaders, founders and researchers, and he has presented on the near-term implications of AI for the United Nations, the World Bank, and INTERPOL. Before AI, he was a Brazilian jiu-jitsu black belt and national champion running a successful academy. He now hosts The Trajectory, whose Stewarding the Flame series (launched in July 2026 with Michael Levin) asks biologists, complexity theorists and AI researchers how to keep the process of life flourishing through AGI and brain-computer interfaces, and he writes on the "worthy successor" question at danfaggella.com. He grew up in southern Rhode Island and lives in Austin, Texas. On X he is @danfaggella.

    Notable Moments

    • [~08:08] The thesis in one image. Dan on an AGI that can out-optimize us but never blooms: "That would be a squelching of the flame."
    • [~14:05] Toothbrushes. "Theories of consciousness are like toothbrushes. Everybody has their own and doesn't like to use other people's." Dan's reason for holding no pet theory.
    • [~16:36] Peter Singer's gold coins. Dan recounts walking Singer, "the king of all utilitarians," from spending all ten coins on bliss to setting some asi...
    1 hr 34 min
  • Jared Lichtman: Post-Scarcity Mathematics

    Recorded live on June 22, 2026.

    Jared Duker Lichtman, a Stanford number theorist, proved a conjecture Paul Erdős posed in 1986: among all sets of whole numbers in which no number divides another, the primes score highest on the measure Erdős invented for such sets. The radio hour stays with the craft. Jared explains primitive sets with quarterback ratings and jersey numbers, describes the year he spent alone in his room during lockdown until the key definition arrived, and tells what happened when a 23-year-old amateur working with ChatGPT opened a 60-year-old Erdős problem Jared had chased for seven years. His reaction was joy, and he brought the amateur, Terence Tao and others into a joint paper. The podcast hour moves from Aristotle and Leibniz to what Jared calls verified superintelligence: machine output checked against a specification, the way a program either compiles or does not. Erik pushes hard on whether any specification can bind a mind smarter than the person who wrote it, and the two work it through to a shared answer about locality: a guarantee holds only as far as you can seal the problem off from the rest of the world. Jared closes on the idea that "genius" is a pointer society attaches to work after the fact, and that most of the work never gets one.


    Key Topics

    • [00:00 - 02:29] Cold open: Erik and Guy introduce their first mathematician: an Oxford PhD under Fields medalist James Maynard who proved a conjecture Erdős left open for decades
    • [02:29 - 12:17] Erdős and the primitive set conjecture: Erdős as the itinerant mathematician who paid for his lodging in co-written papers, with about 1,500 papers and more than a thousand open problems; primes as atomic building blocks; the twin prime conjecture as a simple unsolved question; what a primitive set is (no member divides another); Erdős's 1935 bound and his conjecture that the primes are optimal, "too beautiful not to be true"
    • [12:17 - 23:43] Four years on one problem: first hearing it as a Dartmouth senior; an undergraduate paper with partial progress; COVID sends him home from Oxford for two years; going back to Erdős's original 1935 paper; a new definition after a year in lockdown; the mountain-top view; Guy on watching the problem seize him ("it sounds like love"); checking the proof with his undergraduate mentor, then with Maynard; a 22-page paper with a proof of about eight pages
    • [23:43 - 37:31] Maynard's record, the amateur, and the craft: last digits of primes and arithmetic progressions; improving on Maynard's methods; the amateur whose ChatGPT session opened a 60-year-old problem Jared had worked on for seven years; the joint paper with Terence Tao; the father-at-the-birth analogy; craft must adapt rather than disappear; taste and goal choice as the human role; number theory's accidental gift to cryptography; math as a backstop against hallucination
    • [37:31 - 39:25] Radio close: where to find Jared, the Stanford Future of Mathematics Symposium, and the "Erdős #1196 and beyond" preprint
    • --- [Act 2: Extended Conversation] ---
    • [39:25 - 44:03] From Aristotle to Leibniz to legal contracts: logic distilled from ordinary argument, echoed in chain-of-thought training; Leibniz's attempt to encode language in mathematics; why contracts define their terms and courts exist for the rest; Erik on being a lawyer who looked at smart contracts early
    • [44:03 - 52:37] Sending ships into the ocean of math: Newton's pebbles on the shore; AI agents exploring the "dark matter of mathematics" and reporting back; "post scarcity of mathematics"; interpretability in AI and in neuroscience; taste as an interpretive framework, like building telescopes; Erik on letting his own coding agents make conjectures from a roadmap; proofs as bridges between islands; tokens, efficiency and compression
    • [52:37 - 1:04:57] Verified superintelligence: a program either compiles or it does not; specifications as the box an AI must stay within; humans at the center on goal specification; laws as specs for a society of strangers; Erik's argument that we cannot chain a future god to our values; how well Claude understands human values today
    • [1:04:57 - 1:14:19] The thermostat argument: Erik's robot arm and 80-degree thermostat against Jared's distinction between checking what happened and proving what will always happen; math versus physics; locality and "vectors of infiltration"; the shared landing on expanding the ground you can really hold as truth, in a lineage from Aristotle to Leibniz to today
    • [1:14:19 - 1:27:11] Datasets, relevance and the question of waking up: how one website of all the Erdős problems steered the AI labs toward them; why the conjecture called to him; invariants and latent spaces; Guy on relevance realization and whether AI could come to care; "stochastic parrots" giving way to new solutions; truth as an attractor of intelligence; goalposts of inevitability and of hope; weak coupling between physics, chemistry and biology
    • [1:27:11 - 1:34:03] Education as hope, and genius as a pointer: the more people understand the ingredients (data centers, energy), the more voice they have; why AI was missing from the last presidential debates; Van Gogh and Nietzsche recognized late; prizes and gold medals as pointers to the work that earned them; the discoveries history never recorded, including by women and others shut out of the institutions
    • [1:34:03 - 1:38:39] Intelligence, power, and the amateur's reaction: Guy on why intelligence and power are not the same thing; a cure for cancer falling in the forest; how the 23-year-old amateur took it; a breakthrough "alters your reward pathways"; close

    Guest Bio

    Jared Duker Lichtman is a Szegő Assistant Professor in the Department of Mathematics at Stanford University, working in number theory, especially the prime numbers. He earned his AB and AM in mathematics at Dartmouth College in 2018, a MASt in pure mathematics at Cambridge as a Churchill Scholar in 2019, and his DPhil at Oxford in 2023 as a Clarendon Scholar, supervised by James Maynard, who received the Fields Medal in 2022. In 2022 he proved the Erdős primitive set conjecture: call a set of whole numbers greater than 1 primitive if no member divides another; Erdős proved in 1935 that the sum of 1/(a log a) over any primitive set stays below a fixed bound, and asked in 1986 whether the primes attain the largest such sum. Lichtman showed that they do (Forum of Mathematics, Pi, 2023). He also set a new record for the level of distribution of primes in arithmetic progressions (66/107, beyond Maynard's earlier 3/5). He joined Stanford in 2023 as an NSF Postdoctoral Fellow, co-organized the Stanford Future of Mathematics Symposium in May 2026, and co-authored, with Terence Tao, Liam Price and others, the May 2026 preprint "Primitive sets and von Mangoldt chains: Erdős Problem #1196 and beyond." On X he is @jdlichtman.

    Notable Moments

    • [~11:25] Why the conjecture pulled people in. "...it seems almost like too beautiful not to be true." Jared on Erdős's guess that the primes are the best possible primitive set.
    • [~17:57] The breakthrough, after a year in lockdown. "Once you get to the top of the mountain, you kind of see the whole vista all at once."
    • [~30:13] On being scooped by an amateur with ChatGPT. "If a solution comes by any means, and if it's a problem you care about, then it's just like a moment of joy." He compares himself to the father who arrives for the birth after the labor.
    • [~46:09] Scouts. "And we can just, you know...
    1 hr 39 min

About Leaking Timeline

From the publisher's feed

Leaking Timeline is a weekly conversation at the frontiers of technology, consciousness, and culture. Hosted by Erik Newton (co-founder of the California Institute for Machine Consciousness) and Guy…