The Abundance Agenda
Emotions, not insight, rule the algorithmic era.
POD: The Abundance Agenda, Episode 126
The media industry, like politics, has been stuck in a scarcity mindset—managing decline instead of building for the future. In this episode, we dig into The Abundance Agenda, the new book by Ezra Klein and Derek Thompson, and explore what a pro-growth strategy could look like for media. Plus, TheSkimm exits to Ziff Davis, the rise of AI-driven advertising, and Anonymous Banker joins to explain why second-tier comedians might be the next big media arbitrage opportunity. Apple | Spotify | other podcast platforms. Watch PvA on YouTube. Out every Friday AM.
Too long… Read this in the browser.
PVA CONVERSATION
The self help frontier
TROY: Ok... Last night I was thinking about you guys and what we should discuss on the podcast this week. Wednesday is for processing the fresh stuff. As always, I fretted about having nothing fresh to contribute, only reassured by knowing good ole’ “safe hands” Brian Morrissey will have pulled a string through something. "I think in narratives," he likes to say. Thank god for that.
I did have a couple of things I wanted to bring up. The first was about this newsletter. I know… it is so boring to talk about the stuff one makes, but bear with me for a sec. I have been wondering if this newsletter format is working for readers. Last week felt dense and long and hard — def not quite a breezy, scannable conversation we had intended. Perhaps our content demands too much effort. It's a trap I often fall into.
This got me thinking more broadly about media, connecting it to our niche-hobbyist media indulgence with something Visa CMO and ex-Def Jam exec Frank Cooper said on a recent Semafor podcast about rap’s rise and cultural transcendence:
The thing that I knew it was serving was a deep cultural need. It's kind of going back to what LL said — the best artists, the best rappers, find something already within people. It's discovering or uncovering that and giving it an outlet. I felt hip-hop was going to do that. When you could see someone in Scarsdale, New York, respond the same way as someone in Brownsville, New York, you realized this wasn't isolated to one group. It spoke to something deep inside people. I knew we would cross the threshold. The leap for me wasn’t hard because what I believed has now become obvious.
Empirically, the newsletter resonates most NOT when we spit out some big, “important” idea, but when something we say makes a reader feel something, helps them see themselves, maybe encourages them to look at something through a slightly shifted lens. Most of what we consume is forgettable. Feelings endure. To wit… a recent newsletter on productivity hacks, one I resisted doing but Brian insisted otherwise, had our most reads. What do I know?
BRIAN: Thanks for the recognition. I hope you’ll come to my Cuban cafe this weekend when you’re in Miami. It has a great guava pastelito.
What you’re getting at is that functional media is very challenged. The most enduring media is built around emotions. I remember watching Scott Galloway do goofy videos for L2, and now he’s a life coach. The iron rule is that if you write online for long enough, you live to become a self-help guru.
What we’re seeing right now in politics with spectacle and provocation is part of this. Beyond the Abundance challenge the Democrats face, and which we discuss on the podcast, is a stylistic issue. The current algorithmically driven Information Space rewards street fighters and provocateurs, not policy wonks. I like Ezra Klein and Derek Thompson, but they are not reaching low-information voters. Those voters are swayed by emotion.
The DOGE stunts are part of the theater. It’s all a show. It doesn’t even matter if Social Security is paying 150-year-olds. That’s a great narrative that evokes emotions. And great shows are about emotions.
The algorithms like emotion because it keeps people around.
TROY: Something else was bugging me. I wanted to take stock of where we are in the next grand media transition that we are now smack in the middle of. I long for a clearer mental model.
Let’s back up for a sec.
The short lived détente of the open web is now completely busted. The Google / Social universe was imperfect in its power imbalance between the aggregators and media, but isn’t that always the way in big distribution critical systems? The aggregators took most of the cash, publishers got a taste, consumers had to live with a shitty digital media experience but got lots of free content in exchange. Creators flourished, content became abundant. Digital won. Print and cable lost. A demand-led market transplanted media’s glorious supply-led past.
That is not to say we don’t miss those days of scarcity, and of… well crafted magazines. There is much nostalgia right now given the dastardly state of things in the legacy business. The forthcoming Grayton Carter memoir, When the Going Was Good, will provide more fodder. As did this weeks admission from 90’s Vanity Fair stringer, Bryan Burrough:
As I look back today, Graydon’s Vanity Fair does feel like some lost world, a gold-encrusted Atlantis ultimately inundated by economic and technological tsunamis, its glories only now being picked over by media anthropologists. I’ve never talked much about what it was like to write there. Because I have always worried about how I’d come off. I mean, the money alone. I’m probably breaking some unwritten law of publishing, but here it is: For twenty-five years, I was contracted to produce three articles a year, long ones, typically ten thousand words. For this, my peak salary was $498,141. That’s not a misprint—$498,141, or more than $166,000 per story. Then, as now, $166,000 was a good advance for an entire book. Yes, I realized it was obscene. I took it with a grin.
Franchise Media had its privileges. $500k in the 90’s is about $1 million today. Three articles! Nice gig.
The next web will not be “surfed.” It will come to you in effortless, personalized and automated bundles.
AI’s warpath is increasingly clear. Change starts with absorbing all evergreen / durable knowledge, streamlining the web’s laborious interface, stripping out a broken monetization model. Pesky hallucination errors don’t matter much. The change is enough to start a behavioral revolution. Consumers will flock to new, low-friction content vending machines (see ChatGPT surpassing 400m weekly users, Google’s share erosion). Naturally, media output will shift to audio and video where more of the value can be retained. Reasoning, deep research, real-time information come next. And on…
AI thrives in knowledge synthesis but struggles with data limitations and is mostly incapable of insight. Over time, video ingestion and sensor proliferation increase AI’s fidelity, expanding AI’s reach across media. This will take years. Human judgment, taste, humor, intuition, and perspective retain value for a long time.
Turns out… AI just can’t ask good questions!… Please see this video with Meta’s Yann LeCun for good perspective on this:
AI assisted production efficiency follows closely behind consumption changes. We all see this now, an inevitable response to new economic realities and the need to compete in a world of abundance. Winners use tools.
Pull back, and you see four forces shaping the next model:
Abundance: As abundant as human conversation. Mediated by algos. Forces complete niche-ification on one side, mass spectacle on the other.
Deep, real-time personalization: But…this will also create a counter-force… deeply human, curated, farm-to-table media.
Production automation: Monolithic media brands beget smaller teams and creators enhanced by AI. Also, AI’s create more and more synthetic content.
Conversation: Massive growth in conversational audio and video of all kinds.
People worry a lot about how the AI's keep getting fed if we stop sending traffic to the web and incentives to create content fall away. Don't worry about this. A sustainable ecosystem will emerge. More of the experience will be toll driven. AI Aggregators will drive subsistence-level compensation to the creation layer. New forms of advertising / transactional payments always emerge. People will keep creating content for fun and money. Because we are human and that's what we need to do. AIs will slurp and slurp and slurp.
After all, what will there be left to do after AI automates everything? We create fucking content! And that’s the good news part of the story. For better or worse you, my friends, will live like content kings!
Will report back as we get clearer on this…
ANONYMOUS BANKER
Where’s the Wiz of Media?
An investor asked me this week where he could find the next “Wiz” in media. Founded in 2020, Wiz reached $100 million in ARR within 18 months and just sold to Google for $32 billion (all cash, no earnouts). It’s a wild trajectory by any measure, naturally sparking envy in sectors like media, where unicorn-level achievements are tough, rare, and almost impossible.
But right now, the media industry is in an entirely different phase: a “clean-up” of companies launched during the digital surge of the 2010s. Advertising and audience headwinds, AI disruption, restless investors looking for exits, and fewer strategic buyers all point to a final wave of acquisitions driven by the hunt for stable (and profitable) properties. There’s a parade of familiar media brands on the block—officially or informally—and it’s no secret that valuations are not what they once were.
Two recent deals in the past week highlight the clean-up: TheSkimm’s sale to Ziff Davis and Tastemade’s sale to Wonder (Marc Lore’s latest consumer-focused company). Adam Ryan from Workweek wrote a solid rundown on TheSkimm deal (link); my favorite line… “The world and its attention moved on [from TheSkimm]. And when that happens in media, enterprise valuations fall faster than your open rates.”
Like many digital-era darlings, TheSkimm grew quickly on VC capital. While this exit will return some capital to the investors, the founders, and other common shareholders will likely be squeezed and see very little cash proceeds. Similarly, Tastemade also just sold for ~$90mm and had raised more than $130mm, making the common stock completely underwater. I’ve seen deals in the past where the cap table gets adjusted to get alignment with founders and the management team to agree to a deal but in these scenarios, the value to common is a few million dollars at most, when it gets split up; mid-level developers at OpenAI are generating more cash after selling some of their stock in a secondary sale vs. the teams at TheSkimm or Tastemade who have invested years into the companies.
In this market, it may be less about finding the next billion-dollar media breakout and more about surviving the shakeout. The real winners have to create and maintain franchise value while figuring out how to harness the power of AI, adapt quickly to shifting consumption habits, and hold onto audience loyalty when it matters most.
Links from the group chat
1: Google’s NotebookLM can now create mind maps for you. Ironically this link takes you to an AI generated Perplexity explainer on the feature. This says something… It was the best one I found!
2: Shame. How the G.O.P. Went From Championing Campus Free Speech to Fighting It
President Trump and Republican lawmakers say that new laws and policies are necessary to protect students from harmful and objectionable content, to prevent harassment and to discourage conformity.
3: Return of a good Apple ad by Spike Jonze.
You'll never go broke if you sell to:
- Men's lust
- Women's desire for beauty
- Elderly's health
- Children's education
- Rich people's fear of loss
- Poor people desire to get rich quick
Understand human nature, and you'll always find a way to succeed.
Blue, X
4: There’s good reasons for Chinese to flood the market with powerful AI models (FT)
The scale of this approach could fundamentally reshape AI’s economic structure. If open-source AI becomes just as powerful as proprietary US models, the ability to monetise AI as an exclusive product collapses. Why pay for closed models if a free, equally capable alternative exists? Recommended Business InsightJune Yoon The US has spurred the Chinese chip industry.
For Beijing, this strategy could be a powerful weapon in the US-China tech war. US AI companies, built on monetisation through enterprise licensing and premium services, could find themselves in a race to the bottom — where AI is abundant but profits are elusive.
5: Another reason the Chinese are gonna hijack the global car industry (Autoblog)
BYD just unveiled a new EV battery system capable of recharging a vehicle in just five minutes, potentially erasing range anxiety and putting electric cars on equal footing with gas-powered convenience. This breakthrough hinges on advanced lithium iron phosphate chemistry, promising affordability and widespread adoption.
6: But… China’s Government Is Short of Money as Its Leaders Face Trump (NYT)
Tax revenues down, government with less money to help consumers or exporters as Beijing braces for Trump’s tariffs.
7: AI Action plan (Platformer)
Tech platforms got a rare opportunity to present President Trump with a wishlist — and they're using DeepSeek's success to push for controversial policies.
8: Solana dives into the American culture war with an anti-woke ad—then promptly deletes it. (X)
Coinbase also want to hit the heartland.
9: Trump, Republicans at record-high ratings as Democrats falter (Fox News)
Half of voters, 49%, approve of the job Trump is doing as president, matching his high from April 2020. That’s also better than at the same point in his first term (43% approved in March 2017). He is at high marks among key groups, including women, Black voters and voters under age 30. (For reference, in January, a 52% majority of voters approved of the job Trump was doing handling the presidential transition.)
Nine in 10 Republicans approve of Trump, while the same number of Democrats disapprove. Six in 10 Independents disapprove of his job performance.
"Give me a bum note delivered with a drop of sweat over perfect pitch every time"
-Roger Daltrey
Vis Alex and AGI, in the words of Marge Gunderson in Fargo "I'm not sure I agree with you a hundred percent on your police work, there, Lou." Five years ago the absolute standard response from almost all AI experts was AGI in 2100 - so eighty years, that has dropped precipitously year on year and now you'd be hard pressed to find all but outliers predicting 2-3 years tops. A lot of the dismissal, doubt and disclaimery around AGI seem comparable to a old Buddhist analogy - as the equanimity laden Joseph Goldstein says; " it's not hard to be aware, its just hard to remember to be aware' similarly it's not hard to think exponentially, its' just difficult to remember to think exponentially.
You can't try out latest frontier models like KIMI.AI or get deep in the weeds with Grok3 (obv Alex is gonna have to take my word on this) and not now see far more than just glowing embers of genuine intelligence - and yes it will still flatter and tell you want to hear - but heck for a 1,00 years we've known "a book is not a mirror, if an idiot looks in a sage will not look out".
Anywho, you asked for frameworks I got 'em - roll up roll up! Getcher luberly cogent comprehensible insightful overviews of AI here, - fresh off the noggin and no mistake guvnor.
______________
THE HYPERLUDIC FRAMEWORK
The term "hyperludic accelerant" refers to a specific subset of technologies that exponentially influence the creation, distribution, and manipulation of information. Unlike other General Purpose Technologies (GPTs) such as the steam engine or electricity—which primarily transformed physical processes—hyperludic accelerants: language, writing, The printing Press and now AI operate directly on knowledge itself.
These rare technologies are considered "hyperludic" (from "ludic," meaning play or game) because they fundamentally change the rules of how humans think, communicate, and solve problems across domains. They don't merely speed up existing processes; they create entirely new possibilities for human cognition and collaboration.
What makes these technologies so transformative is their cognitive symbiosis with the human brain—they mirror and amplify our natural abilities to process information, enabling us to tackle increasingly complex challenges.
- THE FOUR HYPERLUDIC ACCELERANTS-
LANGUAGE: THE FOUNDATION OF SHARED UNDERSTANDING
The emergence of complex language—estimated to have developed over 50,000 years ago—was humanity's first hyperludic leap. Language transformed isolated individuals into interconnected societies capable of abstract thought and collective action. It enabled the development of culture, cooperation, and the transmission of knowledge across generations.
Key impact: Language mitigated the dominance of physical strength, empowering those skilled in communication and persuasion. It created the first knowledge hierarchies while simultaneously enabling unprecedented collaboration.
WRITING: THE GENESIS OF ORGANISED CONTROL
Emerging around 3,000 BCE, writing revolutionised how information was stored and utilised. By externalising memory and making knowledge durable, writing enabled the development of complex bureaucracies, legal codes, and historical records that transcended individual lifespans.
Key impact: Writing concentrated power in the hands of literate elites (scribes, priests, administrators), enabling the expansion of empires and religious authorities through centralised control of information.
THE PRINTING PRESS: THE CATALYST FOR DEMOCRATISATION
Gutenberg's invention in the 1440s marked a watershed moment in human history. By enabling mass production and distribution of texts, the printing press democratised access to knowledge, fuelling the Renaissance, the Reformation, and the Scientific Revolution.
Key impact: The printing press challenged established authorities (the Church, monarchies) by empowering individuals with direct access to information, fostering critical thinking and scientific progress while creating new forms of information overload.
ARTIFICIAL INTELLIGENCE: THE DAWN OF COGNITIVE ENHANCEMENT
Today, AI stands as the fourth hyperludic accelerant, poised to transform our relationship with information in unprecedented ways. AI systems can not only process vast amounts of data but can generate new content, personalise experiences, and enable advanced analysis beyond unaided human capacity.
Key impact: AI is simultaneously democratising access to sophisticated cognitive tools while concentrating power in those who control AI infrastructure, creating a complex new landscape of opportunity and risk.
THE 12 UNIQUE QUALITIES OF HYPERLUDIC ACCELERANTS
- THEY EACH FUNDAMENTALLY CHANGE THE "RULES OF THE GAME"
Unlike technologies that merely improve efficiency within existing systems, hyperludic accelerants redefine how entire societies engage with information itself. Writing didn't just enhance oral storytelling; it created entirely new concepts like permanent records and codified knowledge. Similarly, AI isn't just automating existing tasks—it's redefining what's possible in creativity, decision-making, and problem-solving.
- THEY ARE INFORMATION-BASED (RATHER THAN ENERGY-BASED)
While many GPTs like steam engines or electricity transform physical energy, hyperludic accelerants manipulate information itself—how we store it, share it, and act upon it. AI, like writing and printing before it, works directly on knowledge, multiplying our cognitive capabilities rather than just our physical power.
- THEY SUBSUME THEIR PREDECESSORS
Each hyperludic accelerant contains and expands upon previous ones in a cumulative layering that creates exponential power. Writing captured language; printing scaled writing; AI now absorbs, interprets, and amplifies all three previous accelerants. This recursive building upon prior breakthroughs creates compounding effects that transform society.
- THEY COMBINE PREDICTABILITY WITH RADICAL UNPREDICTABILITY
Unlike technologies with predictable outputs, hyperludic accelerants produce emergent, sometimes chaotic results that surprise even their creators. The printing press predictably produced books but unpredictably sparked religious revolutions. Similarly, AI consistently processes data but often generates unexpected insights, solutions, and creative works that extend beyond its explicit programming.
- THEY TRIGGER EXPONENTIAL COMPLEXITY
By expanding the possibilities of human thought, hyperludic accelerants create compounding effects where new insights trigger new questions, and solutions create novel problems. The printing press didn't just produce books—it triggered scientific revolutions, religious reformations, and political transformations that continue to reverberate today.
- THEY CONTAIN THE SEEDS OF THEIR OWN SOLUTIONS
Hyperludic accelerants are both the source of disruption and the solution to the problems they create. The printing press caused information overload but also enabled the distribution of organisational systems like encyclopaedias and indices. Similarly, AI creates challenges around information verification but also offers tools to manage the complexity it generates.
- THEY DEMOCRATISE ACCESS AND POWER
Each accelerant has expanded who can participate in knowledge creation and utilisation. Language allowed all humans to share ideas; writing preserved thoughts beyond a single lifetime; printing brought texts to the masses; and AI is now enabling anyone with a smartphone to leverage sophisticated cognitive tools previously reserved for specialists.
- THEY ENABLE ENTIRELY NEW CULTURAL AND SOCIAL FRAMEWORKS
Beyond improving productivity, hyperludic accelerants fundamentally reshape social structures. The printing press contributed to the rise of nation-states, public education, and democratic discourse. AI is now poised to transform concepts of work, creativity, expertise, and even identity in ways we're just beginning to understand.
- THEY FUNDAMENTALLY ACCELERATE KNOWLEDGE SHARING AND INNOVATION
Each accelerant dramatically speeds up the cycle of knowledge development and application. The printing press accelerated the Renaissance by allowing ideas to spread faster than ever before. AI is now compressing innovation cycles—what once took years (drug discovery, materials testing) can now take weeks or even days.
- THEY TRANSFORM POWER STRUCTURES
Each hyperludic shift has realigned who holds influence in society. Writing empowered literate elites; the printing press challenged religious monopolies and created a merchant publishing class. AI is now creating new power dynamics between those who control computational resources, those who create algorithms, and those who supply or generate data.
- THEY ACT AS A CATALYST FOR NEW ETHICAL FRAMEWORKS
These accelerants destabilise existing moral frameworks, necessitating new ethical systems. Copyright laws emerged from printing press disruptions; now AI is challenging our concepts of authorship, accountability, privacy, and algorithmic fairness, requiring novel approaches to governance and regulation.
- THEY ACCELERATE FEEDBACK LOOPS
Unlike technologies that improve linearly, hyperludic accelerants create self-reinforcing cycles that accelerate change. AI research improves AI systems, which then accelerate research further, creating potential for increasingly rapid transformation that outpaces our traditional adaptive mechanisms.
Blimey - and there's you chaps across the water finkin he was just some lippy lairey Limey with a north & south bigger than his loaf of bread. :-)