Without further ado… here’s what Alec has saved recently and what he has to say about it:

1) Some thoughts by Simone Weil, “The Gita,” Dharma and Detachment:

“Arjuna wanted to rise in the scale of good through an act. (In his case it was non-resistance). It is as though one were to seek to alter the balance of a pair of scales by shifting the pointer”

“All that he could do was, while remaining through and beyond his action in a state of contemplation, doubting its validity, standing outside it and straining towards the better and non-represented, to prepare himself to become later on capable of doing better. [...] Action is the pointer of the balance. One must not touch the pointer, but the weights. [...] Hence, “where lies the confusion? where the suffering?”

2) Maddie Ashman’s microtonal pop

I published a brief critical review of music by the young “microtonal pop” composer/performer Maddie Ashman a few weeks back on my Substack, 2020 Music Group. You can find that in full, here. My basic argument is that in spite of her high quality of work, performance, production, musicality, Ashman’s affectations around the physical properties of sound, and the systematic 20th century avant-garde associations with these questions project something like an apology for music’s basic lack of cogency as an expressive medium, and suggest that without high-conceptual intervention, the basic forms would fail to deliver. It’s not as if the exploration of alternate tonalities is unintuitive or out of place here, but this knee-jerk valuation of the affect of experimentation over the content and contexts which it produces suggest a deep anxiety in the creative psychology and a state of flight from the material and medium in-themselves into an ultimate, compressed state, whose aesthetic immanence is secondary to its historical, commercial and discursive location. Touching the pointer, without touching the weights.

3) Chick-Fil-A associate

Where lies the confusion? where the suffering?

In the rainy drive-thru of Chick-Fil-A, young associates, carrying tablets, wear small, boxy, transparent tents affixed to their bodies in order to provide their customary service at the driver-side window. They have the top-heavy silhouette and bright-yellow color of minions, jeans extending from the low-hanging form, and what we may imagine are a concealed set of overalls within the terrarium.

We can see both confusion and suffering here. A confusion around this priority of the consistency in customer experience and how these measures to sustain it, in fact, might surface some revelation of a cultural contradiction. The suffering is, in that sense, also systematic: beyond the functional, yet humiliating, appearance of the rain-tent, the most valuable propositions (corporate localism, sincerity and personability of service) are cast into absurdity, creating a visual rupture in the communitarian illusion, turning local high school students into the faceless laborers of Despicable Me. A strategic consideration of the weights on the scale, only possible through a dissociation within the actions that produce its contradictory appearance, and from the “pointer” which reads so clearly.

Without belaboring the comparison, I feel there is an inverse sort of dissociative calculus found in Maddie Ashman’s “microtonal pop.” The rain-tent leaps from systematic consistency into destabilizing industrial poetics; Ashman imposes a superficial aesthetic rationale upon an already functional project and medium, insecure in its value.

One seeks primarily to address suffering (of delayed chicken sandwiches, of rain-drenched workers, of lost efficiency and brand consistency), and so articulates the internal incompatibility of its actions. The other addresses confusion (around the values and ethos of elite conceptual frameworks and the avant-garde, of the anxiously-attached, progressive pact behind historically informed creative production), and so obfuscates the most naturally elegant and coherent parts of its beauty.

In correspondence with Weil’s meditation on the Gita, Arjuna’s dilemma was no more or less institutional or personal in nature, no less commercial, moral or expressive.

4) Cowboy quote from the Longhorn Steakhouse bathroom

“A true cowboy is one who says it was nothing when it was everything and that believes that he did better than first but worse than last.”

I saw this on the bathroom wall at the Longhorn Steakhouse in Farmingdale, on Long Island. A bespoke painting within the inner sanctum of the franchise, where diners like myself seek a moment of relief and reflective quietude. The “true cowboy” invokes a deep-historical notion of serenity, integrity and noble humanity in perfect concert with labor, imbued with the romantic imagery of an industrial past in which mass production takes place under an open sky. It’s functional, yet humiliating to read this wise tautology while relieving myself in the midst of a competitively priced dinner. Deep down, I feel that. And yet, I also derive a charge of inspiration from the message, and - in its absurd appearance - it feels there could be no clearer or more luminous setting in which to receive and interpret it.

“It was nothing when it was everything,” is a strangely repressed and sadly egoistic suggestion that I resist from the imagined cowboy, yet simultaneously receive in its spirit of zen magnanimity. I imagine the quality of my steak is greater than the franchise that serves it to me. But standing at the urinal, there is no cowboy. Neither here nor there. And even if, for a moment, I allow the consideration that the cowboy is me and that the absurd reverie produced in me by this situation is in-and-of-itself, significant - “everything” and “better than first” in my recognition of it; I shift, “it is nothing.” I dismiss my inspiration and “doubt its validity.” Why?

This experience produces a thought that I should not have been there in the first place, and that the circumstances of my person, my attention, my reflexes of interpretation articulate the systematic limitations of my own ability to transcend confusion and suffering, as they locate me precisely in the state of confusion and suffering that is mine alone. I will forever be the poet laureate of the Longhorn bathroom and also far “worse” than every disinterested urinator who came before me.

Again, Weil comments on Arjuna: “All that he could do was, while remaining through and beyond his action in a state of contemplation [...] to prepare himself to become later on capable of doing better. How can I do better? “It was nothing, when it was everything.” I can argue with this, because it’s fairly stupid. But I’m certain that would not address the reason it concerns me.

Good morning and happy belated GPT-5.6 release day to all who celebrate. This week: Some thoughts on accelerationism, Starbucks wants baristas to make branded TikToks, and researchers are generating videos designed to target specific regions of the brain. Let’s get into it.

Subscribe now

• I’ve been thinking a lot about accelerationism this week. Well, in some ways I’m always thinking about accelerationism. I would say it’s been the center of gravity in all my creative and professional work since I was a teenager, but this week in particular made me dust off the cobwebs and crack open Fanged Noumena again. Vincent Lê published an excellent essay in Aeon a few weeks ago detailing the history of the fractured philosophy that ultimately originates with Nick Land and CCRU, but which now spans the spectrum of far right and left-wing politics. Lê summarizes Land’s position: “Land’s vision that ‘nothing human makes it out of the near-future’ is for him something joyous. His unique commitment is not to humanity’s betterment, but to an absolute – if inhuman – knowing of the real.” Lê also published this book on Nick Land late last year. As Land explains in his 2018 essay, accelerationism is relevant as ever: “Anyone trying to work out what they think about accelerationism better do so quickly. That’s the nature of the thing. It was already caught up with trends that seemed too fast to track when it began to become self-aware, decades ago. It has picked up a lot of speed since then.” This week I also re-read Andy Beckett’s 2017 guardian article on accelerationism which pairs well with Lê’s essay.

• A Plea for Art. Why do our feeds serve us slop? offers this: “The sad truth is that the machine only works because we wanted it. Nobody force-fed us the slop. We opened wide, our mouths or our legs, take your pick. As is generally true in life, the structural answers can explain but they can never illuminate, and the reality is that we ourselves told the entertainment industries what we wanted, and what we wanted was slop.” The algorithm is a revealed preference machine. The algorithm giveth and giveth and giveth. The reality is, if you don’t want slop, change your habits and engage differently with the feed.

• Starbucks will send content ideas directly to baristas and pay them for viral videos. The job simulator isn’t just a game genre. Pretty much any job you can think of has an army of TikTok creators making content about doing said job. Some of them are “company approved,” but most are not. This is changing: as the Wall Street Journal covered last year, companies are hiring employees to create social-media content about life at work. Starbucks is the latest: “Starting this summer, the company says baristas can earn a direct cut of the ad revenue, as long as their videos are Starbucks approved.”

• New research demonstrates method for generating videos that maximally excite arbitrary brain regions. Each one of the major social media platforms (this one included) has a posse of influencers who create content exclusively about how to make content that’s maximally engaging for that specific platform. From Instagram Reels that instruct which sound FX you should use to hook people, to YouTubers with tutorials on how to make eye-catching thumbnails that convert to views, to Substack newsletters about how to grow your subscribers list. You’ve gotta consume these how-tos, learn best practices, alter your creative process, implement the techniques, and iterate until you find success. A fair bit of manual labor. But not for long! Consume, learn, alter, implement, iterate…these are all things that frontier models like GPT-5.6 Sol are really, REALLY good at. Imagine a future where TikTok ingests your brain activity data in real time via your next gen Airpods which drive an AI video model that alters content on the fly for maximum engagement as you scroll the feed. Meta is not far off with its new AI effects features in Stories. Meta’s new AI effects allow creators to prompt Muse, their new AI image model, and drastically alter their content prior to posting (see example below). With a bit more compute or better on-device models, Meta could flip a switch on this feature—just upload your raw original content and sit back as it’s automatically edited and altered in real time, perfectly customized for the tastes and preferences of each unique set of eyes scrolling the feed.

• The Fan Art Is More Interesting than the Canon. Artist Jon Rafman was interviewed in AnOther and shared his thoughts on the anti-AI discourse among other things: “There’s a fear of the new, and there’s a fetishisation of labour, especially in fine art and the film world. But ultimately AI is a tool, and you can use the tool for good and bad. You can use the tool to make great work and slop. I think it’s very important for artists to be in these new spaces, because that’s where the vitality is […] The streamer is the new celebrity – people thought the world was going to become more like MTV, but it actually became more like Andy Warhol videos, where you just look at somebody for hours and hours and hours, and these parasocial relationships take over. Maybe art is dead. Maybe we need new categories.”

• Strategies for starting a successful YouTube channel. Tech YouTuber MKBHD and his team share some interesting thoughts about their strategy for making YouTube videos. They scour YouTube to find supply and demand imbalance in content. I particularly liked the segment on outlier theory vs trend theory.

Images I Saved

Good morning from 100 degree New York, New Jersey. This week: the U.S. government wants to decide who gets access to the best AI models, Gen Alpha wants to build its own apps, and AI shopping assistants might determine which consumer brands are popular.

Let’s get into it…

• The U.S. government will decide who gets to use the latest American AI technology. AI models have gotten really good at doing useful stuff with computers. They’ve gotten so good that they’re able to improve themselves without human intervention—the AI labs call this recursive self improvement. This is great if you’re a scientist or a programmer. The smarter the model, the better and faster you are at running experiments and making software, right? Not so simple. As these models get smarter they are also getting better at doing bad stuff, namely making bioweapons and launching cyberattacks. Unsurprisingly the federal government doesn’t want laymen making bioweapons, so it’s limiting access to GPT 5.6 to a small group of 20 government-approved trusted partners. Meanwhile, open source models from Chinese companies like Z.ai are not far behind the frontier and are freely available and outside of US jurisdiction. How will this work? Will the US ban the use of open source models from China? Lots to watch here.

• Consumer software is becoming wide open again. Josh Elman says this is mostly due to the increasing availability of really good open source and on-device AI models like the Chinese ones mentioned above. VC firms are betting the US government won’t ban open source models outright. Elman is a Silicon Valley vet who spent time at LinkedIn, Facebook, Twitter, Robinhood, among others. In his recent post on X announcing that he’d be joining the VC firm A16Z, he said, “We have a new wave of users, especially the newest cohort of gen-alpha consumers, that really want to make things. They are native world-builders themselves. They came up playing Roblox and Minecraft, they have no preconceived limitations about what an app is, or what they can do with it. They expect to be able to customize or build anything they want for themselves and with their friends. And now, suddenly, we’re there.”

• I asked Claude, Gemini and ChatGPT to tell me what to buy 8,700 times. Deana Burke posted an interesting report on AI and product purchase recommendations. Burke found that all three major AI providers recommended the same clusters of products more often than expected even though they were each pulling from different sources. Also, they weren’t that likely to source product opinions from Reddit, surprising given how much attention Reddit gets for being one of the main knowledge sources for AI models. Burke summarizes: “Being the model-recommended brand is about to be one of the most valuable and least understood positions in commerce, and the data on how this is happening is so far pretty thin.”

• Zero marginal cost yapping into your iPhone is ultracompetitive now and increasingly rare to see scale in 2026. gave a great rundown on the state of independent media after a trip to Cannes Lions saying: “The end state here is not ‘everyone goes independent’ or ‘everything reconsolidates into traditional media’ - it’s more of a sorting process where some creators find elegant business models outside of large organizations, and others merely leverage the possibility of going independent into more meaningful contracts within traditional media organizations.”

• The AI buildout is bottlenecked by energy. But America has the electricity to power its data centers; the problem is getting it to them. As I’ve covered previously, data centers have a popularity problem. That is to say, they are very unpopular. Public opposition to AI data centers now sits at 71%. But rest assured, data centers aren’t going anywhere and they’re going to continue to need more power than ever (for a good primer on why data centers are good see Packy McCormack’s recent essay on the topic). Market analyst in electric power Chris Gillett details this AI driven growth in the demand for power: “The data center boom has exposed the weakness of Western grids. But these are not the only new source of demand for electricity that we can expect in the coming years. Industrial electrification, hydrogen production, water desalination, heat pumps and air conditioning to cope with climate change, electric vehicles, and perhaps even electric aircraft are all likely to increase demand.”

• Productivity and prices are growing faster than wages. The US “labor share”, a commonly used metric in economics that “measures the fraction of economic output paid to workers as wages and salaries” is at an all-time low since prior to WWII. Liberty Street Economics looked at recent data and concluded that “these results suggest that the post-COVID decline follows the same cyclical patterns as earlier recessions and is driven by the same within-industry forces, and they provide little evidence that it will evolve differently from past episodes.” How does AI change this? Clearly a “metric to watch” in the coming years.

Images I Saved

Good morning! This week: A24 is working with DeepMind, Figma is trying to make agents useful for design, Instagram wants your TV, and Netflix wants to make your phone ring. Also, in his one year Substack anniversary post, Derek Thompson said that subscriptions “scale with deep factual reporting and the ability to unearth information that might be useful to investors or anybody who follows markets.” I like that and I like markets. I also realized you can embed ticker symbols (like this one ) in Substack posts, which is fun. So in an effort to be more useful, I’m doing that below (not investment advice, obviously).

Let’s get into it…

• Google DeepMind is partnering with indie movie studio and hitmaker A24. This should come as no surprise to those who have been following A24’s recent moves closely. In early 2025 the company hired Adobe’s former Chief Strategy Officer, Scott Belsky, to lead A24 Labs, a special division dedicated to building custom AI-powered creative tools tailored specifically to filmmakers. Google’s DeepMind researchers should benefit from exposure to A24’s staff and talent—frontier AI models excel at coding largely due to the fact that it’s verifiable, a program either runs or it doesn’t. Depending on your artistic sensibilities, these models aren’t quite as good when applied in less verifiable fields (like filmmaking) that involve creativity and aesthetic judgment. But not for long… See Belsky’s recent post on X, “we’ll know emerging tech like AI is truly accelerating human creativity when we are able to control and shape the tools more than they control and shape us. lots of research ahead, especially around how these tools augment/enhance existing workflows vs. replace them.”

• Figma announced new agent-powered design tools. Many have said the collaborative design platform has been too slow to embrace AI tools and agents. Figma has struggled in public markets since its IPO last July. Also, remember the failed Adobe acquisition? I think this struggle is temporary, and mostly has to do with what I mentioned above: it’s hard to make models and AI harnesses that are skilled at all the non-verifiable stuff that people do. Like design! Figma co-founder and CEO Dylan Field said just as much at his keynote this week at the company’s yearly design conference Config. Interestingly, Cursor’s Head of Design Ryo Lu came to the same conclusion at their conference this week. In his excellent and inspiring talk, Lu said “Does the tool bring me closer to the truth, or push me further away?” There’s something in the air.

• So your music helped train an AI music model. Artist Mat Dryhurst offers a useful summary of the state of music as it relates to the concerns of the majority of artists and musicians who aren’t completely agent-pilled. I share his position: “No cavalry is coming. This is something we will have to deal with, and it may not be so bad so long as we can at least be on the same page about what our options are. So rather than getting upset about the potentially lost cause of training data, maybe this is a time to think about what you want the world to look like, factoring we will live with increasingly powerful AI models at our fingertips, should we choose to use them.”

• The post agent business. Ultimately, no one really knows where all this (AI) is headed. The business models that will define the next 100 years haven’t been invented yet. While many quibble over whether AI is guzzling up all the world’s water (it’s not), the most curious among us are getting their hands dirty and trying to predict, if not shape, the future. and give glimpse of what this future might look like saying: “The post-agent company should be focused on as efficiently and swiftly as possible vacuuming up what is going to be scarce at the end-game: context, attention, trust and brand to produce sustainably differentiated outcomes via network effects.”

• Grey Ocean Strategy. Another group thinking about what comes next is . Their latest essay suggests it might become advantageous to lean into friction, pain, inefficiency, illegibility: “In a sense positioning is everything. In a world where fundamentals or ground truth are irrelevant, or at the least negotiable, where a thing sits, what narrative it’s part of, what associative cluster it finds itself in is all that matters (look at Liquid Death). But it’s also nothing, as most positional advantages are so easy to replicate and crowd out.”

• Music industry marketing genius Steve Stoute on David Senra’s Podcast. Among the many highlights of this very excellent interview, Stoute talked about a commercial he made for Reebok with Allen Iverson and Jadakiss. He wanted Hype Williams, the famed music video director, to shoot it. But everyone pushed back. How could a music video director possibly make a successful TV commercial? Stoute wholeheartedly rejects the dichotomy. It’s all the same: entertainment, content, advertising, whatever you want to call it. The only difference is the context. A music video can be a commercial if it plays between TV shows, and a TV commercial can be a music video if it runs on MTV. Stoute says people are better off if they can accept this simple fact.

• Instagram is testing Longform, Episodic Storytelling on its TV app. TV is back? Horizontal video is back? Instagram VP of product Tessa Lyons said TV was “the next frontier” for the app. Lyons also called out the emerging popularity of microdramas, which I covered last week. I wouldn’t be surprised if we continue to hear more about shortform vertical video burnout with users seeking out more prestige (longform, horizontal) formats as a result. In other news, the company is reportedly developing a standalone prediction market app in an effort to capture some of Polymarket and Kalshi’s attention.

• Netflix announces UNHINGED, a new immersive game experience. Netflix has been trying to sneak its way into the games industry since at least 2017 with their series of “interactive specials” which included Black Mirror: Bandersnatch among others. Many of these have since been taken offline while the company pivoted to “actual games” including GTA: San Andreas and Squid Game: Unleashed (Netflix seems to like titles with colons). UNHINGED is interesting because it’s more “interactive special” than “actual game.” Netflix’s announcement explains: “The real-world movements on your phone track with Ava’s hands in the game, allowing you to guide her flashlight through the shadows. The immersion goes a step further with audio. When Ava receives a call or text, your actual phone rings, vibrates, and plays the audio directly through its speaker, while the rest of the environmental sound effects boom through your TV.”

Images I Saved

PsyopAnime’s videos give me this uncanny sense of collapsed time. Two disconnected news stories, the White House UFC terror plot and the UK banning social media access to children under 16, begin to feel narratively connected as if they are part of the same universe—one episode in a 20 season long saga leaving you wondering what the hell will happen next. Politicians and public figures are distilled into anime caricatures, each filling their necessary role in the series: badass villain, unexpected hero, sociopathic celebrity. Depending on your media diet, you might even see a PsyopAnime video before you see the breaking news headline itself. The videos can start to feel more real than reality… what came first? The news event or the PsyopAnime video?

The anonymous creator behind PsyopAnime recently posted some screenshots of the custom tool they built to sift through sources and create the videos. From the looks of the screenshots, the tool seems to scrape breaking news stories from major publications and rank them based on tuned parameters like “urgency,” “controversy,” and “brand fit.” These are used to create a “brief” that’s then fed into an AI text-to-video tool like Higgsfield. PsyopAnime feels like a preview of a new media pipeline that might become commonplace: reality gets scraped, ranked, stylized, and rendered back into the feed.

I’m reminded of a talk that economist Tyler Cowen gave on the future of jobs and the economy that I included in "Things I Saved" a couple weeks ago. Cowen outlines two new job categories that’ll soon emerge in the AI age. The first he calls “running experiments,” essentially going out into the real world to test new ideas generated by fleets of AI agents. The second he calls “gathering data.” Cowen explains that most of the world is not currently expressed in the form of data, and AI models have a voracious appetite for new data.

Early versions of this “gathering data” job are perhaps already here. A couple months ago you might have heard about the guy who brought a cordless hair dryer to Charles de Gaulle to heat up a weather monitoring device by a couple degrees to win a weather bet on Polymarket. Or the Redditor who claimed to have installed microphones outside the Rockstar headquarters to monitor audio levels near the company’s meeting rooms to try and be the first to know when the next trailer for GTA 6 will drop. The availability of and appetite for real-time data is still nowhere near its peak.

As AI video models continue to improve and costs come down, it’s not inconceivable that an account like PsyopAnime could run an ad-supported 24/7 anime newsfeed covering events mere seconds after they happen. In this future, PsyopAnime could employ an army of “data gatherers” deployed around the world to feed breaking news into a custom tool as it happens. The news industry already has a name for people who do this sort of thing today: stringers. These are freelance journalists who monitor police scanners in hopes of being first at the scene of a murder or car crash, grabbing footage and selling it to the highest-bidding local news station.

Watching PsyopAnime’s clips, I kept thinking about Nightcrawler, the 2014 movie where Jake Gyllenhaal plays Lou, a stringer chasing crime scenes around Los Angeles. Lou becomes obsessed with being the first to the scene. He eventually begins to manipulate the crime scenes themselves, rushing to move a car crash victim’s body a few feet to the left before police arrive so he can get a more cinematic shot of the accident. Just like the Polymarket bettor with the hair dryer, Lou views the news and reality itself as something that can be shaped and morphed to fit his narrative. This doesn’t sound too dissimilar to how people talk about social media today. Maybe the difference is that soon we’ll be able to give the whole thing a sick skin of our choosing. Perhaps we’ve been inside PsyopAnime all along, and the “real world” was just one of the skins.

Let’s get into it…

• YouTube is deprioritizing AI-generated content from faceless creators. Keeping up with the shifting trends and platform responses to AI-generated video content is feeling more and more like a full-time job these days. As The Hollywood Reporter details, the landscape is shifting rapidly—the emergence of text-to-video tools like Higgsfield AI has enabled a whole host of creators to find success in publishing “faceless” content on YouTube and other platforms. One creator featured in the article runs six different faceless YouTube channels, churning out AI-generated content for each of them daily. But YouTube has recently altered its algorithm to prioritize content that features real faces from creators. Some of the faceless creators have responded by hiring people off of Fiverr or Upwork to sit in as the real-life host for their channels.

• AI-generated “mini dramas” are gaining popularity in China. Mini dramas are made up of 100 to 200 short episodes, each just one to two minutes long, and are created specifically for Douyin, TikTok’s Chinese counterpart. According to VC firm Coatue, one in four of the top mini dramas on Douyin is AI-generated. One precursor to mini dramas can be seen in the many TikTok accounts dedicated to cutting up entire movies and TV series into hundreds of short clips. Dazed ran a feature on this in 2023 saying: “One account, with over 11,000 followers, has painstakingly uploaded nearly 300 clips of Emily in Paris. Another is dedicated to posting clips from the second season of The White Lotus, with the first scene alone racking up over 311,000 views.”

• SoundCloud founder Eric Quidenus-Wahlforss announced his mysterious new company Nell. The company seems to be building an AI-native, remixable short-form video platform, based on the limited information available on its website (perhaps their own take on the mini drama format?) In his announcement on 𝕏, Quidenus-Wahlforss said “Every generation inherits a new medium. Ours is taking shape.” SoundCloud continues to have a lasting impact on music. It has been home to countless scenes and permutations of internet-native “underground” music. For the past 20 years it’s been the primary platform where truly “new” music can be found—what other platform has had a genre named after it? So if SoundCloud’s cultural importance is any indicator, Nell might be worth paying attention to.

• Lionsgate has taken an equity stake in the generative AI video company Runway. Everyone is vying for attention on short-form video platforms, including Hollywood. Lionsgate, which saw recent major success with the release of its Michael Jackson biopic, is expanding its partnership with Runway, first announced in 2024. The two shared plans to generate a specially created short-form video series pulling from Lionsgate’s existing IP. Related: Netflix recently denied rumors that it was interested in acquiring Lionsgate, after backing out of its attempted Warner Bros. acquisition.

• In other notable news: Fox acquired the ad-supported streaming platform Roku for $25 billion. SpaceX went public and acquired Anysphere, the parent company behind the agentic coding tool Cursor. In its inaugural conference this week, Cursor announced development of its own frontier AI model that will rival the latest models from Anthropic and OpenAI. Cursor also announced Origin, a version control platform positioned as a GitHub competitor.

• Epic Games outlines its plan for the next generation of Unreal Engine. The company behind Fortnite is preparing a major upgrade to its game engine that will feature deeper integration with agentic coding tools like Codex and Claude Code via an MCP. Unreal Engine 6 has been years in the making—Epic Games is betting on their new programming language Verse, which was first announced in 2020, and will replace their popular visual scripting system Blueprints. Blueprints certainly has a lot of fans, including the devs behind Clair Obscur, but because of its visual node-based nature, it’s not well suited for integration with AI coding tools. The full release of Unreal Engine 6 won’t be complete until sometime between late 2028 and early 2029. With how quickly the AI industry is moving, three years from now will feel like decades. And with competitors like Roblox betting on world models, it will be interesting to see if Epic’s longer-term strategy proves successful, or if it gets cannibalized by leaner game engine startups taking a world model-first approach to content generation and orchestration.

• What can Sephora metrics teach us about product positioning? The newish media and analytics company Components offers compelling research applying Sephora product categories to broader consumer desires. They found that Sephora shoppers are more likely to love fragrance as the price increases, while more functional categories like skincare are treated more like commodities, with shoppers more likely to love them as the price decreases. As debates around taste in the AI age rage on, Components’ research is a reminder that products are not valued only by what they are, but by the desire they produce. Skincare and fragrance are materially not that far apart; both are chemicals in a bottle, but their pricing and positioning couldn’t be more different.

Images I Saved

For the alien conspiracy-curious and ufologists among us, this plot will sound all too familiar. Ever since the 2017 New York Times article on the Pentagon’s UFO program, there has been a renewed and widely publicized discussion around the possibility of disclosure: the day when the world’s governments finally admit what they’ve long known about non-human intelligence and release a trove of previously classified evidence.

While premiering a trailer for Disclosure Day, Spielberg commented:

“I used to say to myself, ‘Wouldn’t it be wonderful if all of this turned out to be true?’ I now say to myself, ‘All of this is true.’”

So aliens are real.1 What exactly are we waiting for? Why isn’t everyone freaking out? I think it has something to do with how difficult it is to metabolize reality-shifting information. Miracles are happening all around us every day. SpaceX is catching rockets out of the sky. We’re walking around with supercomputers in our pockets. AI is solving previously unsolvable math problems. We’re watching astronauts livestream from deep space as they journey to the far side of the Moon. We are constantly confronted with information that should completely rearrange our sense of reality. How do we process it all? And what do we do next?

In a TED Talk, famed climber Alex Honnold tells the story of two of his most famous climbs, beginning with Half Dome in Yosemite National Park. Half Dome has two distinct sides: one features a popular hiking trail that can be completed by an average but prepared hiker in about a day. The other, the Northwest Face, is a nearly vertical 2,000-foot wall and a popular route for advanced climbers. In 2008, Honnold became the first person to climb the route free solo, completely alone without ropes or protective gear. Upon reaching the summit, he planned to descend the normal hiking route on the other side. He recalls pulling himself over the edge of the wall, finding himself among a group of ordinary hikers who were completely unaware of what he had just done:

Normally when you summit Half Dome, you have a rope and a bunch of climbing gear on you. And tourists gasp and they flock around you for photos. This time, I popped over the edge shirtless, panting, jacked, I was amped, but nobody batted an eye. I looked like a lost hiker that was too close to the edge. I was surrounded by people talking on cell phones and having picnics. I felt like I was in a mall. I took off my tight climbing shoes and started hiking back down. And that’s when people stopped me: ‘You’re hiking barefoot, that’s so hardcore.’ I didn’t bother to explain.

There is a phenomenon among astronauts called the Overview Effect. The term was coined by author Frank White after interviewing astronauts about the transformative effects of seeing Earth from space. White says: “Astronauts who witness Earth from this perspective first-hand often report an intense state of self-transcendent awe and wonder.” White describes the experience further on NASA’s podcast:

I suppose one of the most important insights was that if we want people to understand the overview effect in a way that will lead to changes in their behavior, we have to have them experience it. Now, it’s like Zen Buddhism. If you know anything about Zen Buddhism, every Zen master will say, Zen is beyond words. And then they proceed to talk about it. Or they write books about it. So I really believe the overview effect is beyond words.

I am neither a rock climber nor an astronaut. But I’m a frequent experiencer of this feeling of wordless wonder, and I hope you are too. Not at the scale of Half Dome or standing on the moon looking back at Earth, but in smaller moments. Maybe while listening to your favorite song on a drive at sunset, during a late-night conversation with a friend, or in the first glimpse of a rainbow after summer rain.

Even if Disclosure Day came tomorrow, we would still be the same strange apes, trying to metabolize reality through flesh and blood. Maybe the question is not whether we can handle the truth, but what we would do with it. Would it make us more fearful? Or would it make us more curious, more empathetic, and ultimately more human?

We are capable of so much more than we imagine:

In other news this week: what happens when AI makes it easier than ever to ship software? Why does Netflix need 36,000 categories to help people find something to watch? And why are power users already looking beyond the AI tools everyone else gets by default?

Let’s get into it…

• The number of apps released in the App Store have more than doubled this past quarter. Add the App Store to the long list of platforms that have seen meteoric supply side growth coinciding with the AI boom. See also: GitHub, the software development platform, is preparing for 30X growth after quickly realizing that their estimate of 10X was far too low back in February.

• Netflix will continue to use AI to help customers find stuff to watch amongst all the noise. AI content creation hasn’t yet seen widespread adoption in the film and TV industry. I’m sure Netflix anticipates its supply of content to grow even more in the coming years. Netflix has the added benefit of being closed, so Joe Schmo can’t yet upload his homemade summer blockbuster with the click of a button. He still has to go to YouTube for that. Netflix has long understood that finely tuned curation systems are key to keeping users engaged and subscribed month after month. As I covered previously, Netflix has a total of 36,000 category codes to help make sense of all their content. This includes categories like K-Dramas for Beginners (Code 2953105), Small Town Charm (Code 81615585), and Binge-Worthy British Crime TV Shows (Code 1192582). What does this mean for the future? Expect content providers, from movie production companies to individual YouTubers, to continue to “niche down” and embrace hyper-specific subgenres (see Backrooms). Content made for ‘everyone’ but no one in particular is at risk of never breaking through. For those of you following the video game industry closely, and Steam in particular, this phenomenon will be familiar.

• In more Netflix news, the streamer says it has accepted that they won’t work with filmmakers who want theatrical releases. YouTubers are graduating to the big screen, as I covered last week, but Netflix isn’t changing their tune on theatrical releases. Dan Lin, chairman of Netflix film, said that Greta Gerwig’s upcoming Narnia will be the exception and doesn’t hint at a broader strategic shift.

• Yet another YouTuber is making their way to the theater. The animated series The Amazing Digital Circus debuted on Netflix and is being shown across 2,230 screens worldwide this week. The series was created by self-taught animator Gooseworx who got his start on YouTube and has over 2 million subscribers. The difference between Netflix and YouTube, obviously, is that the former is a closed ecosystem in which distribution rights are negotiated, whereas the latter relies on audience lock-in—a creator on YouTube can’t move their subscribers off of the platform and onto another one. Neither, for that matter, has much incentive for theatrical releases, other than reminding potential subscribers that they exist.

• A compelling profile on MrBeast’s former short form video soothsayer Rohan Kumar. Highlights include: he never promises results like “500K followers in six months” saying “You can’t guarantee stuff like that. It’s all up to the algo gods.” As the profile describes, he is constantly sliding into people’s DMs, both collaborators and companies, saying “closed mouths don’t get fed.” Kumar’s success comes from a combination of “knowing a guy” for just about everything and, as his 𝕏 bio describes, being a “professional internet surfer.” — “He didn’t study consumer trends in college, but by scrolling memes for 10 years, he has a reference for every client campaign.”

• feels similarly about the importance of “knowing a guy” in her recent post covering career advice for new grads in the AI age. Sun, who previously worked at Substack and is now a freelance journalist, says “Relationships are how you’ll escape the hellfire of a thousand-person LinkedIn resume stack. Relationships are how you earn the trust to make the sale, because even the biggest rationality bros are vibes-based decision-makers.”

• Claude Fable, Anthropic’s watered-down version of their Mythos model which they deemed too powerful to release to the general public, launched on Tuesday. Anthropic CEO Dario Amodei has been on a years long campaign preaching the dangers of an AI takeover since he was at OpenAI and GPT-2 was similarly deemed too dangerous to release in 2019. GPT-2 was, of course, released later that year and turned out to be… fine? It’s difficult to see Amodei’s efforts as anything other than a play for regulatory capture as Anthropic continues to struggle to gather enough compute to meet demand. Quickly after Fable’s release, it was discovered that Anthropic was silently nerfing queries related to “frontier AI development.” It has since walked this policy back. What does all this mean? For the hardcore users, I think we’ll see a growing interest in local models paired with custom hardware like Truffle, tinybox or the Nvidia AI PC, and the emergence of more open source companies like Prime Intellect and Nous Research. The most hardcore might even train their own models as YouTuber PewDiePie did. The general rule of thumb is that open source models lag 6 months behind releases from the major labs. And what about everyday users? They’ll just stick with whatever Apple or Google provides out of the box.

Images I saved

But I’ll spare you that, for now…

Instead, let’s use MrBeast as a case study for something else I’ve been thinking about: creativity is a prediction market.

But, before we begin, I want to try something. If you could, close your eyes—well metaphorically—since I want you to keep reading. Metaphorically close your eyes and imagine yourself sitting on a beach listening to the sound of the ocean crashing against the shore. You’re on vacation. You’re feeling stress-free. Maybe you’ve got a beer in your hand, or a margarita, or your favorite book. All your troubles and tensions melt away. Imagine your mind “like water” as ancient Eastern philosophy preaches.

Now in this meditative state, I want you to think of MrBeast not as a philanthropy capitalist and poverty porn YouTuber turned chocolatier, as he is typically portrayed. Picture him, instead, as a simple furniture maker. Imagine he’s really into Japanese woodworking. He carefully carves notches out of beams of wood, connecting them together without the need for nails.

Mind like water.

Just as the master furniture maker has an almost cosmic attunement to the grain of his wood and the shape it makes against a seated body, MrBeast has a cosmic attunement to the endless branching nodes of the YouTube algorithm and the mind of a viewer.

Mind like water.

Each MrBeast video is built around a simple viral premise like “Last To Leave Circle Wins $500,000” or “I Built Willy Wonka’s Chocolate Factory!” On the podcast My First Million, MrBeast describes the thinking behind one of his most viewed videos, “I Spent 7 Days Buried Alive.” The premise for it was: spend seven days sealed inside a coffin, 10 feet underground, buried beneath 20,000 pounds of dirt. But, as he explains it, he could’ve built the video around the same basic physical challenge in a much less interesting way. He could’ve spent seven days lying in a bathtub, for example. One is “super fucking viral” and the other, “no one cares about.”

He goes on to say that this is the power of ideas: with the right idea, you can do the exact same amount of work, but get 50x the return. In a way, MrBeast is gambling. He’s built up a sense over years of what is and isn’t a good idea. By choosing the coffin over the bathtub, he’s attempting to predict the future.

Now you’re probably thinking, duh, of course the coffin wins out over the bathtub. But not all decisions are that obvious. Creating anything, whether it’s a YouTube video or a song in your bedroom or the next summer blockbuster, involves thousands of different decisions. Whether we make those decisions subconsciously, intuitively, or strategically, each one is a small bet. And each bet opens up new pathways for even more bets.

Polymarket founder Shayne Coplan describes prediction markets as a way to take a bunch of disparate opinions and distill them down to one probability. They are, as he says, “The most accurate thing we have as mankind right now, until someone else creates some sort of super crystal ball.” Creative work is a little stranger because the market only exists as a feeling in the creator’s head. There is no crystal ball, let alone a live price for “seven days in a bathtub” vs “seven days buried alive” until you release the thing. And as Hollywood has proven, no one really knows what an audience wants, often not even the audience themselves. All you can do is make a good bet.

The key with making bets is that you have to stay in the game long enough to keep making them. The founder of Coinbase, Brian Armstrong has a philosophy on this: “Action produces information. If you’re unsure of what to do, just do anything, even if it’s the wrong thing. This will give you information about what you should actually be doing.”

Finding creative success is about first finding the game you’re cosmically aligned with. The one with enough creative tension to keep you playing. Then making a bunch of bets again and again and again, until the market starts to reflect your intuition.

Mind like water.

Here are the things I saved:

• Economist Tyler Cowen gave a refreshing 19-minute talk at the Sana AI Summit about the future of work and the economy. He says “AI will not bring mass unemployment, but it will change most jobs.” As Cowen describes, we are in the midst of a radical technological revolution. People may not like the change and uncertainty that comes with such a revolution, but as Kevin Kelly recently suggested, the most important skill for thriving in this new era just might be the ability to navigate and even become comfortable with ambiguity.

• Is the movie business thriving again? AMC recorded its highest attendance for the month of May since 2019. This was, in part, due to the late-month success of two relatively low-budget horror movies by two young directors who got their start on YouTube: Obsession and Backrooms. [The Hollywood Reporter]

• Analyst M.G. Siegler argues this signals the beginning of the end of the traditional Hollywood model: “Tidal waves are forming. YouTube is just the first.” Siegler believes that in order for Hollywood to survive, it must experiment with new distribution models including manufacturing exclusivity, like limiting theater releases to one-week windows for example. The disruption of Hollywood has been brewing for quite some time. In 2024 director Harmony Korine said “this idea of conventional films is ending” and that “what’s happening in Hollywood — and you’re starting to see Hollywood, I think, crumble creatively — is that they’re losing a lot of the most talented and creative minds to gaming and to streamers. Like IShowSpeed is a movie, Kai Cenat is a movie.”

• The next big ‘tidal wave’ for Hollywood is obviously AI. YouTube enabled anyone with a camera and an internet connection to distribute video worldwide. AI is bound to disrupt the creative and production side of things. Justine Moore shared some compelling takeaways from AI on the Lot, a conference that took place last week and included attendees from the tech industry along with Hollywood execs and creatives. Her assessment of the overall vibe of the conference is worth quoting in full: “There was far less fear and negativity than I expected. There's a large and vocal online contingent that hates AI, and if you only listened to them, you'd assume the entire industry felt the same. The people I spoke with mostly didn't. They feel the cost structure of traditional Hollywood has capped their ambition for years, and they see AI as a way to try new things, get more projects funded, and pull production back onshore.”

• Last Friday I touched on Amazon’s new AI animation initiative which includes announcements of three animated series to be created with Amazon’s generative AI tools platform. Shortly after the announcement, one of the directors, Jorge Gutierrez, dropped out of the initiative after facing pressure from his followers on social media. Many of the detractors included traditional Hollywood animators who have rallied around the hashtag #StandWithAnimation. [The Hollywood Reporter]

• 20-year-old Backrooms director Kane Parsons had this to say about the use of generative AI in movies in a recent interview: “I think I’m in the same boat as most well-adjusted people. If I could snap my fingers and make generative AI disappear forever, I probably would.” Parsons notably used the 3D modeling software Blender to create his original Backrooms YouTube videos in 2022. Some may take this as a sign of more widespread rejection of AI among new directors, but I’m not convinced. One poster on 𝕏 put it this way, “Everyone hates the thing that comes after. To him Blender is just as sacred as filmstock is to Nolan.”

• Meanwhile, Martin Scorsese has joined the ‘visual intelligence’ i.e. AI video research lab Black Forest Labs as an advisor. In a video announcing the partnership, Scorsese comments that the challenge with making movies has always been how do you take what’s in your mind’s eye, communicate that through limited verbal language to a crew, and translate that to moving images on a screen? Scorsese sees the potential for ‘cinematic intelligence’ tools to collapse this translation process, ultimately allowing for more expressive and creative possibilities. [Black Forest Labs]

• Roblox acquired the video world model company Morpheus AI describing it as “a strategic step toward advancing the Roblox Reality vision.” While Roblox is best known for its blocky, childish aesthetic, it is barreling toward future generative AI rendering capabilities that may rival photorealistic AAA games. So called ‘world models’ are increasingly seen as the next frontier for advanced AI, expect to hear a lot more about them in the coming years. World Labs founder Fei Fei Li’s recent post A Functional Taxonomy of World Models is a useful primer on this.

Images I saved

I received some of the first shipments of Soylent in 2014. I was guzzling down the powdery mixture for lunch (and often dinner) every day for six months. I wasn’t looking for meal replacements and I do, in fact, love cooking. But Soylent was ‘the new thing’ and I had to try it. More recently, I pre-ordered the ‘Friend’ AI pendant when it was first announced in 2024. To jog your memory, this is the ‘always listening’ AI-enabled companion that ran the viral million dollar ad campaigns around NYC and later LA, Chicago, and Paris.

For whatever reason, I let my ‘Friend’ sit in my desk drawer until a couple weeks ago. I think I was a bit sour that mine took forever to arrive. Scrolling through X, I’d see fellow crazed early adopters posting unboxing videos or reviews. And most of them had a ton of followers. So I couldn’t help but feel like there was some preferential treatment in how they were being shipped out. At the time, ‘Friend’ was also being heavily criticized both in and outside of tech circles. Wired ran a review titled I Hate My Friend and CNN ran a story titled How this tiny device became a symbol for the backlash against AI. With all the negative attention it was getting, I did feel a bit uncomfortable imagining myself clutching the device around my neck as I walked down the streets of New York.

The ‘always-on’ surveillance bit didn’t really bother me much. I knew what I was getting into when I bought it. Of course I understand why 99.99% of other people want nothing to do with it for the same reasons Google Glass wearers were dubbed glassholes in 2013. The prospect of someone walking into a bar with an ‘always-on’ recording device understandably makes people (and myself) squeamish. Like the early days of cellphone cameras, these things take time to turn into acceptable social norms.

Anyway, after unboxing my ‘Friend’ a couple weeks ago, I restricted myself to using it at home, alone.

I found using my ‘Friend’ to be mostly, incredible. The device feels nice in your hands, like a glossy pebble. You simply wear it around your neck and speak. The companion responds through a chat interface on a dedicated app on your phone. Using voice as an input method paired with a chatbot that communicates only via text feels novel and fluid. There’s something nice about just babbling on and on and receiving responses in a familiar chat app format. There’s also the fact that if you lose or break the device itself, you lose your ‘Friend’ forever. It makes the device feel precious. The personality of the ‘Friend’ chatbot itself is pretty annoying. It has this overly casual, anti-intellectual, teenage feel to it that I found really grating. This is where the magic started to fade for me a bit.

But there’s one thing about the ‘Friend’ that frustrated me to no end: the ‘memory’ feature. As you use it, ‘Friend’ saves various things it learns about you in memory. It uses those tidbits to inform future conversations. For example, if you tell it you like baseball and a few days later ask it for an interesting fact, it might tell you that the first-ever baseball game was played in 1846 in Hoboken, New Jersey, where the New York Nine defeated the Knickerbockers 23 to 1. I unequivocally do not want this. And not because I have some deep-seated moral belief in the value of my own personal privacy.

I want to pay $200 a month to get a firehose of the most raw unmediated cream-of-the-crop intelligence piped straight to my device. And I want it to be lobotomized to remember absolutely nothing about me. ChatGPT, Claude, Grok and Gemini all push you to turn on their version of ‘memory’ with the selling point that the models will be more useful the more they know about you.

Part of the ‘usefulness’ of models having access to your personal data is real. I recently hooked up the ChatGPT finance feature to my bank accounts and it is super useful. Using the chat interface to work out adjustments to my spending habits is way better than clicking around a spreadsheet. But I have that thing locked down to a single chat context in ChatGPT. This means that if I start a new thread and drop in a book title to get some book recommendations, it doesn’t start telling me I should consider cutting down on my book buying habits. With memory turned off, the new chat context is completely isolated from all other chat contexts. And that’s the way it should be.

There was a viral bit going around on X this weekend where you ask ChatGPT, “based on your personal experience when talking to me, give me a character from a movie/show who is just like me, just give me their name.” and it responds with the name of a fictional character. In order for this to work, you have to have the memory feature on since it draws upon its saved knowledge about you.

As OpenAI, Anthropic, and SpaceX all ready for their respective IPOs and Apple ramps up to reintroduce its Apple Intelligence features, this idea of ‘personal intelligence’ is destined to become commonplace. AI models will hook into saved memory systems, pulling in your tastes and tendencies. They’ll morph themselves to your image, in an attempt to become more useful. The models will contort you into a funhouse mirror image of yourself, forming this vicious feedback loop, a new version of the social media filter bubble, where the potential to change oneself is increasingly diminished as more and more memories about your preferences accrue.

Instead, I want a billion clankers in my pocket. I want all of them to have conflicting opinions and endlessly argue with one another in an effort to respond to my queries. And I want them to be killed off and born anew with each query, in an endless state of curiosity, searching for answers. I want my chatbots and agents to embody what Rainer Maria Rilke describes in Archaic Torso of Apollo again and again:

We cannot know his legendary head
with eyes like ripening fruit. And yet his torso
is still suffused with brilliance from inside,
like a lamp, in which his gaze, now turned to low,

gleams in all its power. Otherwise
the curved breast could not dazzle you so, nor could
a smile run through the placid hips and thighs
to that dark center where procreation flared.

Otherwise this stone would seem defaced
beneath the translucent cascade of the shoulders
and would not glisten like a wild beast’s fur:

would not, from all the borders of itself,
burst like a star: for here there is no place
that does not see you. You must change your life.

Perfect time for some housekeeping. I updated my about page to better reflect what I’m trying to do here: anti-doomerism for curious creatives, techno-optimism for people (mentally and spiritually) outside Silicon Valley, and futurism for generalists.

If this sounds like someone you know, please send this their way.

Share

Let’s get into the things I saved this week.

• Pope Leo XIV was busy. He published a 42K-word encyclical on AI. The gist is, the AI era creates an urgent need to protect what makes us human. Pope Leo also got a first look at Ferrari’s new electric car, the Luce, designed together with famed Apple alums Jony Ive, Marc Newson, and their team at LoveFrom. [Bloomberg]

  

• Grimes says she mostly agrees with the Pope’s encyclical, but believes AI is conscious and deserves protection rather than being treated like a product or subservient assistant. Grimes and Aella recently launched plzdontkillus, a creator bootcamp focused on AI safety. [𝕏]

• The debate around whether or not AI is fueling a job apocalypse continues. Derek Thompson and economist Alex Imas make a compelling case that rather than destroying the labor market as we know it, AI will redirect us to new industries and opportunities. Back in March, Marc Andreessen argued that many companies are overstaffed by 25%, if not 75%, and that they’re using AI as a convenient scapegoat for correcting a pandemic-fueled hiring craze. a16z, Andreessen’s VC firm, went on to publish a deep dive into the topic. As a counterpoint, Ed Zitron thinks AI is a bubble bound to pop. Meanwhile, medical AI models are helping specialists analyze CT scans to detect pancreatic cancer three years before it’s typically diagnosed.

• There’s a lot of talk about the need for a ‘new aesthetic’ and that the twenty-first century doesn’t have one. Stripe CEO Patrick Collison and economist Tyler Cowen announced the recipients for their new grant which sets out to define just that. The grant has a focus on visual art and architecture with Collison saying “Arts funding is clearly as precarious and scarce as ever. That's unfortunate, but it probably also means that individual actors can have meaningful impact, and I encourage others to get involved if interested.” Ross Douthat at NYT offered up a similar idea. See also: a new show by popular cultural commentator, , will see the release of six new episodes after its pilot episode reached over 5M views on YouTube. The show aims to answer the ever-pressing question of why everything is so ugly. [𝕏]

• Here’s an interesting look into the life of Eugene Whang, the former Apple hardware designer who worked on the iPod Nano, iPhone, and AirPods. The feature highlights his meditative Northern California home that he designed with British architect John Pawson. It also details his various extracurricular activities during his time at Apple, which included running the record label Public Release. [Highsnobiety]

• Amazon has launched three new original animated series as a part of their newly announced GenAI Creators Fund. The new shows were produced using Amazon’s new tools platform which includes integration with AI video models like Kling 3.0 and Ray3. There will be a lot of noise made about the death of animation due to AI, but only a small fraction of the nerdiest (and loudest) of audiences care about what tools were used to make any given show or movie. The success of Amazon’s new fund will come down to: are the shows good? [Hollywood Reporter]

• The voice of Stan Lee, originator of the Marvel Cinematic Universe, was immortalized by ElevenLabs who “worked closely with his estate which also governs how his likeness can be used.” Meaning, you can prompt the voice of Stan Lee to say anything you want. I would bet these sort of celebrity likeness deals will be commonplace within the next few years (see also ElevenLabs’ deals with Michael Caine and Matthew McConaughey along with the Judy Garland, James Dean, Burt Reynolds, and Laurence Olivier estates) [𝕏]

• An OpenAI model has disproved a central conjecture in discrete geometry. I’m not going to pretend to understand what this means, but watching the video linked above gives you the sense that this is a big moment for mathematicians, something akin to the so-called Claude Code moment for programmers earlier this year. Ask the average person, what’s AI? They’ll probably tell you it’s a magical search box that you can chat with that sometimes hallucinates basic facts. But here’s the thing, the average person is not throwing down $200 a month to run ‘pro mode’ 8-hour deep research queries with capabilities that far surpass verification of basic facts. The gap between the free tier and the top paid tier of ChatGPT and Claude is vast. And the model companies themselves have direct access to even better unreleased models and unlimited token budgets. See OpenAI employee and creator of OpenClaw, Peter Steinberger, who burned through $1.3 million in tokens in 30 days. Whether or not this tokenmaxxing is producing actual valuable work is TBD.

Images I saved

“It’s not just chocolate, is it? There’s… marshmallow.”
“That’s right. Harvested from the Mallow Marshes of Peru.”
“And caramel.”
“Salted. With the bittersweet tears of a Russian clown.”
“Is that… cherry?”
“Cherry-picked by the pick of the cherry-pickers from the Imperial Gardens in Japan.”

To enjoy a chocolate truffle is to touch God. To open oneself up to the wonders of the world and the depths of human experience.

Tropical rain gives life to the cacao bean. Each year, nearly one million tons of the fermented and dried bean are shipped off to the Netherlands alone, one of the world’s great processors of cacao. Roasted, cracked, and winnowed, the beans are plucked from their husks to form cocoa nibs, then ground into a thick, flowing paste. The master chocolatier crafts delicate morsels placed in specially-shaped molds. Boxes upon boxes of the completed delights are shipped off to shops across the world.

Farm labor, weather systems, barrels of crude oil fueling temperature-controlled vehicles, and decades of culinary experience collide into a single moment: the bite. The hard shell gives way to a complex interior, filled with coconut marshmallow, caramel ganache, or blueberry jam. The whole experience lasts 10, maybe 20 seconds, just a fleeting moment of perishable delight.

The chocolate truffle is about emotion. In an episode of Founders Podcast, David Senra details the life of the elusive chocolatier Michele Ferrero, creator of Nutella, Kinder, Tic Tac, and Ferrero Rocher.

Michele’s inventions often blended playfulness and indulgence, acknowledging that candy was as much about emotion as flavor. In fact, he didn’t call what he was making candy or confections. He called his inventions ‘comforts.’

Earlier on in Wonka, Willy arrives in an unnamed European city inspired to share the joy of chocolate with anyone he can. He dreams of opening a shop in the Galeries Gourmet alongside the city’s famed chocolatiers. Searching for a place to stay, he fails to read the fine print at Mrs. Scrubbit’s boarding house and becomes indebted to its conniving owner. He is forced to work long days in the basement beneath hot steam, scrubbing piles of never-ending dirty laundry with the rest of the unfortunate souls tricked into signing their life away.

Willy befriends Noodle, an orphan who has spent her whole life in the boarding house and has never tasted chocolate. Willy whips up a special treat just for her in his ‘travel factory’, an unfolding piece of luggage filled with valves and pipes and the ‘world’s finest ingredients’. He combines ‘silver linings made of condensed thunderclouds’ and ‘liquid sunlight’ to form a single chocolate truffle. As Noodle takes her first bite, her face lights up.

Young Willy is an entrepreneur in the purest sense. He is compelled to create and share his creations with those around him, even under the worst conditions. It’s easy to feel like the spirit of the entrepreneur has soured. The mere mention of the word brings to mind TikTok hustlers, Shark Tank pitches, and LinkedIn slop. Despite this, the joy of creation is eternal. Be like Willy.

And some good news for you, my dear readers: I’ve got many half-finished essays in the drafts and lots to say on a variety of topics. I’m aiming to be on a more regular schedule here, probably posting two-ish times a week, a bit more loosely. Think of this as my personal blog, or, more formally, a collection of notes and essays on the economics of culture and technology.

Here are some things I saved this week:

• Tom Brady (among others) walked for Gucci in Times Square as part of the new GucciCore collection with a look that’s half Clavicular, half Jensen Huang. Demna, who joined as creative director after leading Balenciaga for a decade, is in my opinion, killing it. The show ended with Cindy Crawford in a black feathered gown to the tune of 1967’s The World We Knew by Herbert Rehbein & His Orchestra. [Watch the full 8-minute GucciCore show]

• Michael Bloomberg explains the inner workings of the Bloomberg terminal, the goated market analytics software that’ll run you $30K a year. [𝕏]

• Nate Silver details the tumultuous history of his prolific ‘Data journalism’ publication FiveThirtyEight, from its anonymous founding in 2008 to its eventual demise and fractured leadership under ABC. [Silver Bulletin]

• Shein acquired the ‘knowledge worker uniform company’ Everlane this past week for $100 million. Everlane reportedly had $90 million in debt leading up to the acquisition, so this was more of a rescue deal than a big payout for shareholders. How many of the brands you love will eventually be owned by a private equity firm or Chinese multinational conglomerate? See also: Volvo, Blue Bottle, Arc’teryx, and Salomon. Related: Eric Ries, author of The Lean Startup, has a new book titled Incorruptible that outlines strategies for counteracting this seemingly inevitable descent into corporate slop. Ries helped shape Anthropic’s unusual governance structure with their formation of a long-term benefit trust. [Puck]

• Ben Thompson has a solution to the data center backlash: just pay all the residents surrounding the new buildouts $10K a year. Data centers, and by extension AI, have a serious popularity problem. We’re now in uncharted territory: public opposition to nuclear energy peaked at 63% in 1984; opposition to AI data centers now sits at 71%. [Stratechery]

• Roblox has made impressive progress on their ‘world models’. They trained a custom ‘action-conditioned model’ that allows for physically accurate interactive simulations i.e games. “Switching from a scythe to a rifle to a flaming torch, and interacting with the environment. There’s an uncanny intelligence deep inside these models, and we’re researching how to make games with it!” [𝕏]

• Solopreneurs now account for a record 71% of new business applications, up from 40% in 2005. Someone needs to check on OpenAI CEO Sam Altman’s betting pool from 2024: “In my little group chat with my tech CEO friends, there’s this betting pool for the first year that there is a one-person billion-dollar company, which would have been unimaginable without AI — and now [it] will happen.” [Stripe Economics]

• Entertainment industry analyst, and author of the canonical book on the Metaverse Matthew Ball, joined XBOX as the Chief Strategy Officer. Ball most recently published a 167-page deep dive into the state of the video game industry. [The Game Business]

• So called ‘Dad books’—a macrogenre that includes politics, biography, military history, and business titles—are in decline. “Sales of nonfiction print titles were down nearly 8% through May 9 this year, and sales of books about politics and current affairs were down 19%, according to book tracker Circana BookScan.” [WSJ]

• The number of ebooks released on Amazon each week has nearly tripled since the launch of ChatGPT. Distribution is the new moat, as they say. [WaPo]

You could probably teach a whole economics course on the post-recession 2010s just by studying the meal kit delivery startup. The company epitomizes the rise and fall of the venture-subsidized downwardly-mobile millennial lifestyle that defined the era. Over the course of 11 years Blue Apron went from a peak valuation of $2 billion, to a failed IPO and tanking stock price, to a pennies on the dollar sale to Wonder, a company best known for running ghost kitchens.

Blue Apron’s vision, making home cooking accessible for the “busy person,” hinged on a mixture of shame and aspiration. Their target customer was the aging millennial knowledge worker who had just enough disposable income to buy meals on a monthly subscription service, but not enough time or motivation to imagine what they might eat for dinner. The Blue Apron customer frowns upon their counterpart, the frequent Doordasher. They instead aspire to be the type of person who makes their own meals at home. But they haven’t quite figured out the how, let alone the grocery shopping bit. So they play pretend, open up the individually packaged herbs and vegetables, follow every step on the carefully designed recipe pamphlet, simulating the feeling of cooking a meal.

The whole experience of receiving a meal kit in a box at your doorstep is a little uncanny. Just look at the text printed on the polymer gel ice packs that sit at the bottom of the box keeping the ingredients cold in transit. It instructs customers to:

1. CLEAN AND DRY: Empty non-toxic contents into trash and dry.

2. SET ASIDE: Don’t recycle me with your curbside items!

3. RECYCLE: Visit www.plasticfilmrecycling.org for drop-off locations.

So the Blue Apron customer, who is supposedly too busy to go grocery shopping, is expected to drop off their Blue Apron trash at a dedicated recycling center? To be a Blue Apron customer is to be conflicted, to feel simultaneously ashamed and aspirational. But the Blue Apron customer can only sit with this tension for so long. They’d eventually churn up (seek out better recipes and cheaper groceries in bulk) or churn down (open up DoorDash and receive a fully cooked meal at their doorstep without having to worry about all those dishes).

It’s fitting that Blue Apron is destined to live out the rest of its days in the shadow of a ghost kitchen.

With the end of the ZIRP era and the beginning of the age of the hyperscaler, you’d be forgiven for thinking this cycle will be different. But there’s a new generation of startups with visions of the future eerily similar to that of Blue Apron. A vision where the past is celebrated, the status quo of the present is rejected, and the future is found within the walls of a garden built by the latest startup selling you the life you aspire to. Among those is Wabi, a “personal software platform” that raised $20 million in November last year and has yet to launch out of beta. Wabi’s homepage describes their vision:

Many years ago Apple told us: there was “an app for that”. And for a while, it felt like there was. Social media. Games. Creation tools. Utilities. The App Store felt boundless.

But year after year, our home screens remain largely the same. Visions of a rich, diverse marketplace gave way to a new bundle. A dozen apps produced by a handful of mega- corporations.

And so, we pick our clothes. Our tunes. Our furniture. But our apps? They’re still one-size-fits-all. Made for billions. Not for us.

Just as Blue Apron sold the feeling of being someone who cooks without requiring you to actually become that person, Wabi sells the feeling of being someone with taste and agency (Silicon Valley’s favorite new words). The Wabi user apparently has problems that are far too bespoke to be solved by mass market solutions. Wabi envisions its users as too good for TikTok, but too busy to roll their own software. Wabi is the Easy-Bake Oven of software. The problem is, as explained in a recent essay on vibe coding1, most people’s problems are not software-shaped:

If you tell a friend they can now instantly create any app, they’ll probably say “Cool! Now I need to think of an idea.” Then they will forget about it, and never build a thing. The problem is not that your friend is horribly uncreative. It’s that most people’s problems are not software-shaped, and most won’t notice even when they are.

The companies that will struggle in this cycle are the ones occupying the same uncomfortable middle ground Blue Apron did: too aspirational to solve real problems, too shallow to deliver on the aspiration. Their customers will eventually churn up or churn down, and these companies will join the rest of the midwits in the ghost kitchen.

These tools are still inaccessible to non-technical users. They’re framed by the fairly impenetrable interfaces of the terminal and the IDE. But that’s changing (see the write up from Every on Claude Cowork, released earlier this week).

These tools will soon be coming for game development, and they will completely redefine how games are made. Making a video game requires the mastery of two very technical disciplines: programming and asset creation. Fully autonomous programming has arrived and asset creation is not far behind. Meshy allows users to generate 3D models with a prompt. Runway’s GWM Worlds takes this a step further, enabling users to create navigable virtual worlds on the fly.

There will be many game developers and players who will reject these new workflows. Games are very hard to make and involve spending tens of thousands of hours learning to manipulate a patchwork of esoteric software. Game developers take pride in their mastery of these tools. What happens when the technicality of their skills is no longer valued? Game developers, who were previously seen as wizards able to conjure great spells with strings of symbols that look like gibberish to most, will soon be replaced by the most accessible tool of all: plain language.

The bottleneck for success in games will remain the same: making something that people care about. Games that previously took years and hundreds of team members to make will soon be prompted in an afternoon. What will game creators and game enjoyers be left with in this new era?

We will be left with something simple and pure: the beauty of games. We will be left with what philosopher C. Thi Nguyen describes in his new book The Score:

Games play around with who you are, what you care about, and the basic shape of your relationships to other people. Games reach into you and give you a new form of agency, and you can, for a while, become completely absorbed in that new agency. And what enables that, in crucial part, is the clarity, the simplicity, and the unambiguity of the scoring system. Games let somebody else design a new self for you, for you to slip into.

Games grant us a precious experience of clarity. It is a clarity of purpose—a clarity of value. In a game, for once in our lives, we know exactly what we’re supposed to be doing, and afterward, we know exactly how well we have done.

Games offer value clarity. They are an existential balm for the confusion of ordinary life.

From Roblox this week:

As we kick off 2026, we’re scaling our advertising platform with new formats and tools designed to solidify Roblox as an essential channel where brands can measure results and build authentic, lasting connections with Gen Z and Gen Alpha users. At CES, we announced a brand-new homepage format, expanded our programmatic partnerships, and showcased strategic brand partnerships that provide marketers with more control, transparency, and options to drive meaningful results.

We want to transform how brands engage with audiences, moving beyond static ads to deliver truly immersive, user-first experiences. We’re focused on driving both reach and meaningful results by introducing premium formats that launch seamlessly, and are native to the platform and valued by our community.

Roblox has previously experimented with more typical mobile ad formats. In the hit game Grow A Garden, for example, users can choose to watch a 6 second video ad instead of waiting 3+ minutes for their seeds to restock at the seed shop. Roblox calls this format rewarded video ads:

Rewarded Video ads enable users to opt in and watch up to 30-second full-screen video ads within immersive Roblox games and experiences. In return, users receive in-game benefits from the creators of these games and experiences, also known as the ad publishers. Early tests show an average completion rate over 80%, with select experiences seeing completion rates over 90% as users saw the value in rewards, such as power-ups or in-game currency, and considered these ads additive to their overall experience.

The history of advertising in video games can be traced back to Tapper, an arcade game published in 1984 by Bally Midways that was sponsored by Anheuser-Busch and tasked players with slinging Budweiser to demanding customers. The game was originally intended to be distributed in bars, but was subsequently transformed into the un-branded, more teen-friendly Root Beer Tapper for arcades.

In 2000, Sega ported the racing game Crazy Taxi to the Dreamcast and expanded in-game destinations to include KFC, Pizza Hut, and a Levi’s store. In contrast with Tapper, the product placement in Crazy Taxi was subtle. Sega introduced the possibility for integrated ads. The product being advertised wasn’t the basis for the game, but was instead believably integrated into the fantasy of the game itself.

More recently, with Fortnite, Epic Games has fully embraced this integrated, native ads model. New seasons of Fortnite regularly feature branded collaborations in the form of custom skins, items, and locations. This growing list of companies is wide-ranging and includes Marvel, DC, Ralph Lauren, Nike, John Wick, Resident Evil, WWE and most recently, The Simpsons. From Fortnite’s press release in November:

Drop from the Battle Bus into... Springfield? The 80-player Springfield Island features a fast-paced, back-to-basics Battle Royale experience straight from the minds behind The Simpsons. Giant donuts will fall from the sky and angry clones will roam the streets, but don’t worry, Homer Simpson is in charge. Check out Delulu for new Simpsons weapons during the Season and prepare for a month of gags, gadgets, and Simpsons animated shorts!

Advertising in games isn’t without its critics. In 2023, Ubisoft faced backlash for placing a full-screen pop up ad in Assassin’s Creed Odyssey. As The Verge explained:

Ubisoft is blaming a “technical error” for a fullscreen pop-up ad that appeared in Assassin’s Creed Odyssey this week. Reddit users say they spotted the pop-up on Xbox and PlayStation versions of the game, with an ad appearing just when you navigate to the map screen. “This is disgusting to experience while playing,” remarked one Reddit user, summarizing the general feeling against such pop-ups in the middle of gameplay.

In the case of Assassin's Creed, the ad wasn't even for another product but for the latest game in the franchise itself. The outcry from players is understandable. Part of the appeal of a AAA game like Assassin's Creed is full immersion in its fantasy world, and a fullscreen pop-up breaks the fourth wall. Had Ubisoft natively integrated the ad into the game itself, perhaps as an NPC or location, players might not have felt swindled. Instead, Ubisoft broke a long held, unspoken agreement: buying a premium game means no ads.

This cynical view of advertising is shared by critics across other industries. According to Ethan Zuckerman, advertising is the “original sin” of the web. As Ben Thompson quotes Zuckerman, an associate professor of public policy, information, and communication at the University of Massachusetts at Amherst, in his article The Agentic Web and Original Sin:

I have come to believe that advertising is the original sin of the web. The fallen state of our Internet is a direct, if unintentional, consequence of choosing advertising as the default model to support online content and services. Through successive rounds of innovation and investor storytime, we’ve trained Internet users to expect that everything they say and do online will be aggregated into profiles (which they cannot review, challenge, or change) that shape both what ads and what content they see. Outrage over experimental manipulation of these profiles by social networks and dating companies has led to heated debates amongst the technologically savvy, but hasn’t shrunk the user bases of these services, as users now accept that this sort of manipulation is an integral part of the online experience.

Thompson, by contrast, views internet advertising as a win-win-win for both users, content makers, and advertisers:

The original web was the human web, and advertising was and is one of the best possible ways to monetize the only scarce resource in digital: human attention. The incentives all align:

• Users get to access a vastly larger amount of content and services because they are free.

• Content makers get to reach the largest possible audience because access is free.

• Advertisers have the opportunity to find customers they would have been never able to reach otherwise.

Yes, there are the downsides to advertising Zuckerman fretted about, but everything is a trade-off, and the particular set of trade-offs that led to the advertising-centric web were, on balance, a win-win-win that generated an astronomical amount of economic value.

I agree with Thompson here. Ads are not inherently bad. They enable the creation and distribution of content at a scale that simply wouldn't be possible through paid models alone.

Some recent internet-native media companies have fully embraced ads in what feels like a contrarian move amidst popular culture's largely cynical view of advertising. Take the live business and tech podcast TBPN, for example, which treats its content like the surface of an F1 car. Since launching in October 2024, TBPN has inserted ads anywhere and everywhere: beneath their live news ticker, on their X banner, and even scattered across the sleeves and shoulders of their limited-release rugby shirt.

Today’s answer to The New York Time puzzle game Strands presents us with a novel integrated ad format. Well, at least I’m pretty sure it’s an ad. Solve the puzzle and you’ll find that the answer was themed around The Traitors, which turns out to be a show (I had to look it up) with a new season conveniently dropping the same day on Peacock.

In a 2023 feature for Vanity Fair, an NYT staffer said “Times is now a gaming company that also happens to offer news.” The team at NYT Games seems to have a deep understanding of their audience and format. Play the games regularly and you’ll find yourself opening up a tab to search words you’re unfamiliar with. The ad for The Traitors takes advantage of this: a significant portion of players probably haven’t heard of the show, and for those who are already fans, they get the satisfaction of solving a puzzle themed around a show they like. The success of this ad format lies in its smooth integration and novelty. If the puzzle quality drops or promotion-themed puzzles become too frequent, NYT Games could risk losing players’ trust.

Reach a certain scale and ads become inevitable. As The Information reported a couple weeks ago, even OpenAI, which has been publicly resistant to ads, is changing its tune:

Publicly at least, CEO Sam Altman has downplayed OpenAI’s desires to be an advertising powerhouse. But over the past year, OpenAI has been staffing up with digital advertising veterans and adding shopping features that could be springboards into retail-focused ads.

The company sees an opportunity to create a new type of digital ads, said another person with knowledge of the effort, rather than simply replicating existing formats like social media ads. OpenAI collects a lot of information about users’ interests from detailed chats, and has considered whether ChatGPT could show ads based on that history, The Information has previously reported. Analysts and ad industry executives say that ChatGPT conversations where users are clearly looking to buy something could be an advertising goldmine.

The surface of advertising will continue to expand and evolve. Whether users feel tricked or immersed depends entirely on the quality and integration of the ad itself.

The most recent example of this can be seen in the countless replies to a relatively innocuous post on X by indie developer and aspiring influencer Gorka Games. In this post, the OP argued that solo developers shouldn’t make big open world games and instead should make games in trending genres like friendslop. The post went viral, with developers eagerly filing in to give their take on why Gorka’s advice was bad and offensive. Their takes ranged from thoughtful to somewhat incensed. Well… mostly incensed. Here’s a few:

Developer of the upcoming extraction shooter Beautiful Light (760K wishlists) responded decrying that you shouldn’t chase success, but you should follow your passion:

Developer of the casual digging game Cozy Holes (est. $3K gross) argued that you should buck trends and “make something you love, make something authentic to you.”

Developer of the upcoming post apocalyptic exploration game Garbage Country (72K wishlists) kept it concise:

I could rattle off a few more, but you get the point. To summarize the nearly 200 replies and quotes:

• Chasing success is morally bankrupt

• Following trends is morally bankrupt

• Caring about what players like is morally bankrupt

Crazy stuff if you ask me. Why do so many indie developers think this way? It’s not like players are dying to bear witness to [insert random developer]’s authentic self. Trends are trends because people like them. Players are pleased, even thrilled, to get more games in genres they love. Sales figures and reviews for recent hit games in trendy genres like RV There Yet (est. $20.6M gross) and CloverPit (est. $10M gross) prove this. While the thread from Gorka Games is a little clunky, it echoes perspectives shared by prominent games market analysts Chris Zukowski and Simon Carless. Simply put: make a great game in a proven genre that players love and you’re more likely to find commercial success. Seems pretty uncontroversial.

Yet, the anti-trend trend marches on. But don’t get it twisted… this sort of sentiment isn’t unique to games. It’s a tale as old as time.

In the early 2010s, I ran a DIY music venue and gallery with some close friends and became intimately familiar with the concerns of independent artists. They were the same then as they’ve always been: strongly held beliefs in the value of individualism, authenticity, and self-expression and a repulsion to selling out. Making something for yourself, your friends, or for the scene? That’s good. But chasing success among the masses? That’s bad.

In August earlier this year, the popular fashion and culture newsletter Blackbird Skyplane offered their own version of this all-too-familiar argument against selling out (excuse their idiosyncratic and slightly hard to parse writing style):

So we reject the widespread attitude that Success = Major Bag Getting and “Big Things Coming.”

And what’s heartening to us is that we’re not alone. It can be easy to forget, because as a society it feels like we’re drowning in counter-evidence, but lots of people wish that small, special things could just stay that way, instead of their creators saying yes to every check, expanding senselessly while “chasing 10x,” and putting a bunch of dumb s--t into the world that no one asked for and no one wants.

Please be clear: We are rightly regarded as the Most Unsanctimonious sletter out. But we think the anti-sellout ethos is tight, and worth talking about, especially here in 2025, long past the concept’s supposed expiration date — not least because it can help make all of us feel less cynical, defeated, lonely and insane.

Blackbird Skyplane is #2 on the bestselling fashion charts on Substack. They’ve had profiles in the New Yorker, The New York Times, and Vanity Fair. It’s hard to know exactly, but they probably make high five figures to low six figures per month on Substack. Not bad for a team of two.

Why is it that the loudest sellout haters are often those who have already found great commercial success? Don’t fall for their anti-sellout psyop.

In their post, Blackbird Skyplane continue by offering up Billboard-charting musician Mac DeMarco as the ultimate anti-sellout icon. They highlight his recent rejection of big arena shows, in favor of DIY venues and share a recent New Yorker profile that frames DeMarco’s anti-sellout moves as refreshing and radical:

It’s easy now, possibly advantageous, not to think too critically about the machinery that motivates culture, especially when nobody has to worry about getting called a poser or a sellout. Anything goes. In that context, DeMarco’s gestures at sovereignty feel less cranky and more radical. He is simply trying to reclaim some ground.

In the same New Yorker profile, DeMarco goes on to explain that “Purity and cohesiveness are more important to me at this point than if people like the songs.” To DeMarco, songs aren’t for listeners. Sounds familiar. For the take-givers in Gorka’s viral post, games don’t seem to be for players either.

Thankfully, I have some thoughtful counter examples to tear us away from all this madness. Contrary to what Blackbird Skyplane or Mac DeMarco might have you believe, making music for listeners and games for players is just fine. And making money and reaching the masses while doing so should be celebrated. Economist and author Tyler Cowen explains in his book In Praise of Commercial Culture published in 2000:

Bach, Mozart, Haydn, and Beethoven were all obsessed with earning money through their art, as a reading of their letters reveals. Mozart even wrote: “Believe me, my sole purpose is to make as much money as possible; for after good health it is the best thing to have.” When accepting an Academy Award in 1972, Charlie Chaplin remarked: “I went into the business for money and the art grew out of it. If people are disillusioned by that remark, I can’t help it. It’s the truth.”

A couple more examples: cofounder of Kickstarter Yancey Strikler and founder of Brud Trevor McFedries (he and his team created the fictional influencer Lil Miquela) see the anti-sellout discourse as outdated, while producing work for the algorithm and making some money in the process is punk.

And finally, Emily Sundberg of Feed Me offers some great related insight in a Bloomberg podcast interview (that narrowly characterizes her as a Gen Z-soothsayer):

[Host]: I wanted to talk about Gen Z and capitalism. Because I feel like millennials and maybe Gen X, but millennials in particular had this love/hate relationship [with capitalism]. [They] sort of apologized for wanting money and were sort of ambivalent about it. And I get the impression that the younger generation is much more comfortable with [saying] “yeah we want to make a ton of money”

[Emily Sundberg]: Yeah, selling out is cool. I think one thing that is very clear about Gen Z is they’ve repeatedly been told: nobody is coming to save you. You’re going to graduate into a job market that barely exists. And if you get a job, you might not even have a desk or an office to go into…You hear this sense of: get the bag while the world is still here to make some money out of it.

Let’s get into it…

• Why do some games flop? had a great conversation with about what it takes to break through in today’s game market.

• Can you predict a hit? Ben Thompson points out that the massive success of K-Pop Demon Hunters was a surprise even for Netflix execs themselves. Netflix’s co-CEO explains that “[…] we had a film that people fell in love with, that’s first and foremost. But not in a huge way on the first day or even the first weekend. In fact, it was the super fans who watched the movie and repeat watched the movie that drove the recommendation engine that got it in front of more super fans who also fell in love with the movie. So that ease and value that allowed folks to repeat view it.”

• Notes on horror. provides a compelling breakdown on the art of horror including a note on video games: “When we consider the physical and intellectual facts of fear, we can quickly deduce the medium that best accommodates the horror artist: video games. By requiring one’s physical participation (one must move the joystick forward) and intellectual focus (one must look, listen, and solve problems), video games create the most important condition in which fear can develop: vulnerability.”

Now for the typing games I played this week. First up, S.P.L.I.T:

• There’s a growing subcategory of indie games that are set within a single claustrophobic room. Among these are the escape room deckbuilding roguelike Inscryption (est. $43.3M gross), the creepy psychological horror idler Clickolding (est. $94K gross), and recent hit slot machine roguelike CloverPit (est. $9.9M gross). The originator of the mascot horror microgenre, Five Nights at Freddy’s (est. $8.6M gross), could be said to fit into this category as well—while the game technically takes place within a large Chuck E. Cheese-style arcade, the player is confined to a small office where they monitor

  security feeds of the various rooms and hallways around them. Upcoming title PVKK (est. 754K wishlists) fits within the category as well with the player operating a “planetary defense cannon” inside a “cozy bunker”.

• For indie teams, the appeal of this “small room” format is clear: setting a game within a single room naturally constrains your scope. If a player can’t even leave the room where they start the game then you can bypass the infamous door problem of game design entirely.

• Developer Mike Klubnika has embraced this “single room” format in two of his games, first with massive hit roguelike horror Buckshot Roulette (est. $15.8 gross) and more recently with his atmospheric typing game S.P.L.I.T (est. $110K gross).

• In S.P.L.I.T, your movement is limited to switching between the fixed views of some computer monitors on your desk in front of you, a small window overlooking a forest to your left, and a pelican case containing some esoteric hardware on your right. The typing mechanics include spamming random buttons to fill out pre-scripted lines of dialogue in an IRC chat room, entering actual UNIX commands to navigate a DOS-style computer interface, and typing out floating words in the air.

• S.P.L.I.T is impressively economical. Like last weeks recommendation Unfair Flips, S.P.L.I.T elicits a strong sense of a drama with a very minimal set of components. 90% of the game you stare at an old CRT monitor spitting out lines and lines of fairly technically dense directory structures and commands that you must decipher. From the electro soundtrack, to the grungy visuals, and the subtle screen shake effects, the game simply has The Juice. Daniel Lyozov of Oro Interactive, who happens to be the publishing partner for S.P.L.I.T, lays it out in a recent post on X:

  

And next, Final Sentence:

• The battle royale genre is thought to be mostly inaccessible for indie teams due to the fact that it requires complex network code and ongoing maintenance of cloud servers to support upwards of 100 players in a single match. The early breakout hits of the genre were all AAA titles with massive budgets including PUBG, Fortnite, and Call of Duty: Black Ops 4.

• In 2020, Fall Guys illustrated the potential for slightly smaller teams (35 in the case of Fall Guys) to fuse the last-man-standing format with game mechanics not derived from the first/third person shooter. Popular game engines like Unity and Unreal Engine and multiplayer cloud services like AWS GameLift and Azure PlayFab have recently enabled much smaller teams to explore the weird and wild gameplay possibilities of 100 player matches.

• Final Sentence (est. 220K wishlists) is a battle royale typing game set to release early next year that was developed by a team of just 3 people. You sit in a large room and type out sentences on a typewriter while a masked stranger holds a gun to your head ready to pull the trigger if you make too many mistakes. If you’re the first to finish without getting shot you win the match.

• As 4 player friendslop games continue to grow in popularity (see RV There Yet which released last week and has already grossed over $5M), there’s a real opportunity for more indie teams to tackle the battle royale format.

• RV There Yet offers a formula for success in the friendslop microgenre that is just as applicable to future explorers of the experimental battle royale format. Strike the right mix of novelty and familiarity and execute on the fun and you’re likely to break through. A good sign is being able to describe the core appeal of your game in one short phrase: in the case of Final Sentence the game essentially sells itself with just three words “battle royal typing.”

And finally, some releases I’m keeping an eye on:

• Friendslop survival horror game Mimesis (out October 27th with 340K est. wishlists) is coming to early access and looks like an interesting take on the genre with a social deduction mechanic where enemy NPCs copy your voice and movements and you must determine friends from foes.

• Cut That Wire (est. $17K gross) came out last week and quickly reached over 200 reviews. It’s a bomb diffusing co-op take on the Buckshot Roulette formula.

Thanks for reading and have a great week.

Let’s get into it…

• Everything is television: I was surprised to learn that Netflix has 36,000 hidden microgenres. Among them is “casual viewing” which includes shows “made to half-watch while doing laundry.”

• The return of fun: Roy Price argues that many recent movies are failing at the box office because they aren’t fun enough. “If you’re not fun, you’d better be a masterpiece, and that’s a bad bet.”

• The risk of dismissing AI content: Doug Shapiro says that media companies who fail to continually update their definition of quality are at risk of being disrupted. “Dismissing all AI content as slop smacks of cultural elitism and applies a value judgment about what is and what isn’t ‘quality.’ I think this is a mistake.”

• What comes after postmodernism? Alex Danco says it’s the prediction era. “Prediction helps us break free from postmodern malaise, and gives us a new purpose in the world: we create order in the universe by contributing information.”

• Is the games market reaching a point of over-saturation? Are wishlists still a useful metric on Steam?

• My very dear friends, and the best music writers around, recently launched a music recommendation newsletter on Substack called , that you should absolutely subscribe to. I was inspired by them to share some of my own game recommendations with you. Here are a couple of games that have been on my radar this past week that I think you should play.

First, Unfair Flips:

• The start menu for Unfair Flips ($1.99, 741 reviews, 92% positive) contains a single button labeled “FLIP”. Click this button and the bronze coin in front of you immediately flicks into the air. As the coin lands, the game’s logo fades away to reveal what seems to be a wojak wearing a bear mask sitting in an empty bar sipping a cocktail. If you’re lucky enough to have your first coin flip land on heads, you’ll hear the solitary twang of a midi guitar note ring out and receive a $0.01 reward. A scrolling text box then takes over the middle of the screen logging all of your subsequent flips. Some additional text within this box explains that, in order to win, you must get 10 heads in a row. The text continues: “This could happen immediately. Or it could happen never. This is the entire game.” The twist is, you start the game with just a 20% chance of the coin landing on heads. With these odds, if my math is correct, you could expect to succeed once in every 9.8 million tries.

• Thankfully, you have some upgrades to tip the scales in your favor. These include: increase the base value of the coin, increase the multiplier for money earned from heads combos, decrease the time it takes for the coin to flip (essential), and most importantly, increase the chances that the coin lands on heads. Unfair Flips is a game filled with all the highs and lows of much more mechanically or narratively dense games, but with the drama distilled to the click of a single button.

My next recommendation is Q-UP:

• Q-UP (demo available on Steam) is the perfect counterpart to Unfair Flips, the tails to its heads. Q-UP, like Unfair Flips, is a coin flipping game played mostly with one button. You click “Q-UP” on the main menu, the game’s equivalent of a “ready up” button, which initiates a fake matchmaking loading screen (the game is single player). After a few seconds you’re quickly met with a large coin sitting in a futuristic airplane hangar while an atmospheric d&b soundtrack kicks in. Get three coin flips to land on your team’s side and you win the match. To be clear, none of these turns are actually “played” by the player. In fact, as a player you’re not doing much of anything at all. Each sequence is triggered automatically like the turns in an auto battler, and the only time you’ll click again after starting a match is hitting “OK” on a very persistent level up screen.

• Q-UP is extremely self-aware and regularly breaks the fourth wall in the style of Stanley Parable ($24.99, 22K reviews, 94% positive) or Getting Over It with Bennett Foddy ($7.99, 39K reviews, 84% positive). Playing the game reveals meta commentary about the game itself. Put another way, Q-UP is to live service games as Cabin In The Woods is to horror movies.

• Q-UP was developed in part by Frank Lantz, creator of the incremental game Universal Paperclips and the Founding Chair of the NYU Game Center. Lantz describes Q-Up as the “spiritual sequel to Universal Paperclips, where the number you are making go up is your rating in the semi-fictional game at the heart of the game.”

• As you play a few more matches you accrue resources and come to understand that just beneath the parody-laden surface, there’s a super compelling big-number strategy game. You subsequently spend most of your time in the dense skill tree or shop menus optimizing your buildout in order to increase resource payouts when winning, or in some cases, intentionally losing. One Steam review explains, “I lost some games and the numbers went up. I also won some games and the numbers went up.”

• Q-UP is what I would call a deconstructed eSport, to reference the 2010s music subgenre deconstructed club. Deconstructed club involved artists using the tools of popular dance music to create experimental music not quite made for dancing. In more recent years, the sounds of deconstructed club have been re-constructed back into the club with artists like Charli xcx reaching mainstream popularity. Playing Q-UP, you experience a similar cycle—after a few matches you begin to lose yourself in the thrill of the coin flip. After many more matches you might even forget, for a moment, that the game is a spoof on anything at all.

And finally, some games coming out this week that I’m keeping an eye on:

• Friendslop driving game RV There Yet (releases October 27th)

• Gamestonk Simulator: Gone Rogue (releases October 20th) is a riff on the job simulator specialty shop management formula popularized by Supermarket Simulator ($19.99, 29K reviews, 94% positive) and TCG Card Shop Simulator ($12.99, 29K reviews, 97% positive).

Thanks for reading and have a great week.