Episode 231 — April 16th, 2026 — Available at read.fluxcollective.org/p/231

Contributors to this issue: Erika Rice Scherpelz, Ade Oshineye, Ben Mathes, Neel Mehta, MK, Boris Smus

Additional insights from: Dart Lindsley, Jasen Robillard, Jon Lebensold, Justin Quimby, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“Is it not miraculous, reader, the power of the mind to believe and not believe at once?”

— Ada Palmer, Too Like The Lightning

🧪💡 The prompt test

“Make it better!” you tell the AI. You get something, something surprisingly good, TBH... but it’s not what you had in mind. You need to be more specific. Better how? Better for whom? How does it verify? Most folks who have been using AI awhile know that they have to craft a good prompt to get good results.

Now let’s look in a slightly different direction: many leaders communicate with people in ways that would obviously fail as AI prompts. “Increase velocity.” “Be more customer-focused.” “Drive innovation.” These directives feel meaningful, but they are not actionable. However, unlike LLMs, which fail quickly enough for you to correct (sometimes painfully so), people are really great at guessing. And, right or wrong, those guesses compound as they cascade through organizational layers.

This hints at why AI hasn’t transformed large organizations yet. If all we had to do was get the work done faster, then transformation would be easy. But before we get there, we have to solve the problems of dispersed ownership and vague goals. AI can accelerate coding, writing, and analysis. It can’t resolve disagreement about what should be built. It can’t do meaning-making.

Lafley and Martin’s Playing to Win describes strategy as an integrated cascade of choices: where to play, how to win, what capabilities to build, and what systems to put in place. Each level provides context for the next. “Be more customer-focused” fails because it skips the cascade: which customers, competing against whom, with what tradeoffs, measured how? Effective direction-setting is clear and, just as critically, it’s coherent.

These days, we’re deploying AI at the nodes of the organizational graph, helping individuals do their tasks faster. Transformation at scale requires redesigning the graph itself: the processes, decision rights, and information flows. That’s slow, messy work. For example, it means confronting why all those status meetings and planning processes exist in the first place, instead of just making them more efficient. Egos will be bruised in the process.

The prompt test offers a diagnostic: would you give more detail if you were working with an LLM? If not, what’s missing? The context? The constraints? The actual goal? Over time we will see that AI’s biggest contribution won’t be doing the work faster. It will be exposing how unclear organizational communication has been in the first place.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🇬🇧 Britain will encourage households to use more electricity this summer

In a few months, Great Britain could see “the first summer the grid runs entirely on zero-carbon electricity.” There will, in fact, probably be excessive solar and wind power, and this poses its own problem to grid operators: shutting down wind and solar plants when demand is low is quite expensive. So, the national grid operator will urge Britons to boost their power usage when the grid is over-supplied, such as by doing less time-sensitive tasks like running dishwashers or charging EVs. Some suppliers may offer discounted or even free electricity during these periods.

🚏🪫 An electric car factory will run on its own recycled batteries

Electric car startup Rivian has inked a deal to install over 100 used batteries from Rivian cars in its Illinois manufacturing plant. Besides addressing the problem of EV battery recycling, this will help the plant store energy when grid prices are low and discharge it at peak times, a method known as peak shaving. Indeed, as AI data centers drive up demand for energy storage, recycled car batteries offer a cheap alternative. (While EV batteries do degrade, they degrade less than people initially thought, and even a battery with just 50–75% of its original capacity is still useful enough for applications like these.)

🚏🍁 The number of Canadians visiting the US by car is down 35% since March 2024

Ever since the US began its Canadian annexation campaign in 2025, Canadians have mounted a remarkably long-lived boycott of American travel. The number of Canadians driving to the US (the most common method for them to visit) is down 35% since March 2024; there has also been a 14% year-over-year drop in the number of air travelers from Canada to the US. That has accounted for at least $4.5 billion in missing tourist spending, which has hit border areas hard. One travel consultancy CEO was impressed at the scale of the boycott, saying, “In my 37 years in the travel industry, I have never seen anything like what the Canadians have pulled off.”

🚏🦞 Maine will become the first state to pause data center construction

The boom in data centers, and the resulting energy price rises in their areas, has been drawing backlash across the US, and Maine has become the first state to enact a ban on data center construction. The moratorium, which drew bipartisan support, will last until November 2027. (The governor retains the right to veto the plan before it officially becomes law, though.) Maine doesn’t currently have a major data center project, though some small facilities are under construction or being planned.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Secrets of Intelligence Services (Ben Recht) — Shares what the author learned from having Claude Code create a minimal coding agent. He observed that, while LLMs are incredibly “mysterious” and “nondeterministic,” agents themselves are very simple and rule-based. In fact, a micro-agent with just three tools (read file, write file, and run terminal command) plus an LLM was able to do most of what a ‘real’ coding agent can — a classic example of emergent behavior. The author concludes that the simplicity of agents in feedback with the complexity of LLMs is what makes coding agents “the most useful and engaging AI product yet.”

• The Monks in the Casino (Derek Thompson) — Argues that young men are increasingly retreating into a toxic asceticism, trading pro-social milestones for porn, posting, and pervasive sports betting. Thompson argues that the true crisis isn’t loneliness, but an alarming comfort with being completely alone, a trend that echoes Ivan Illich’s “two watersheds,” where the virtue of monasticism has mutated into a vice under casino capitalism.

• What the Death of Direct File Tells Us About State Capacity (Don Moynihan) — A public policy professor examines the rise and fall of the IRS’s free Direct File tool, which was piloted in 2024 but killed in 2025. Governments definitely have the state capacity to build software to serve their own ends (such as surveillance tech), and they can turn that to building public goods too, but without broad-based public support, it’s easy for private entities with a financial interest to kill them off. Indeed, the US government shuttered Direct File after a meeting with lobbyists from the tax prep industry. On the bright side, Direct File was a great example of the user-centric, product-forward, in-house model of software development that civic technologists have been promoting as an alternative to external contractors, and it may return in the future.

• The Safest NFL Draft Pick Is the Most Dangerous (Michael MacKelvie / YouTube) — Applies finance’s Sharpe ratio (a measure of how much excess return you get per unit of risk) to football drafting, finding that the common strategies of picking the “best player available” or filling a position of need aren’t actually the most effective. It’s better to apply the lens of replacement cost and take bigger swings at “premium” positions (like quarterback or edge rusher) that are more expensive to get in free agency. The corollary is that taking even a blue-chip player at a cheap-to-replace position (like running back) high in the draft is a pitfall.

🔍📆 Lens of the week

Introducing new ways to see the world and new tools to add to your mental arsenal.

This week’s lens: plausible deniability.

In Disney’s animated Cinderella, Lady Tremaine says Cinderella can go to the ball:

Lady Tremaine: Well, I see no reason why you can’t go... if you get all your work done.

Cinderella: Oh, I will. I promise.

Lady Tremaine: And, if you can find something suitable to wear.

Cinderella: I’m sure I can. Oh, thank you, Stepmother. [Cinderella leaves]

Drizella: Mother! Do you realize what you just said?

Lady Tremaine: Of course. I said, “If.”

Drizella: Oh! “If.”

And when Cinderella seems to have met the conditions? Well, Lady Tremaine never goes back on her word, yet Cinderella somehow ends up with a dress in tatters.

Plausible deniability is a form of ambiguity. It allows people to keep a distance from the role they played in some outcome. Generally seen as a negative, plausible deniability is often present in scandals, both political and social. A politician who expresses their annoyance at an opponent may not say that they want that opponent’s reputation smeared. If their underlings chose to interpret it that way, who can blame the politician? A government that executes an assassination and hides the involvement of the state can turn around and blame others for the death, as could a leader who stochastically asks why nobody will “rid me of this turbulent priest.”

Despite the negative association, it may be more accurate to see plausible deniability as a tool for negotiating weakness rather than as something inherently evil. Our hypothetical politician and government resorted to plausibly deniable actions exactly because they don’t have the power, legally or socially, to execute them directly.

Applied more broadly, we can start to see ways plausible deniability can be positive. One example is flirting. Attempting to establish interest in another person is socially fraught, so good flirting escalates slowly. A look. If that is returned, some eye contact. Then a conversation. At any step along the way, if it goes well, we can keep going. If it doesn’t, we can bail, “Oh, I was just saying ‘hi.’” And like with political plausible deniability, the denial doesn’t always need to be truly plausible to fulfill its role in signalling.

Early conversations around mergers and acquisitions often have this shape. Potential acquirers and acquirees engage in a slowly escalating expression of interest to reveal how serious they are and if there’s a good match. Diplomats use plausible deniability all the time. They suss out who supports their position and who is against it. They deescalate tensions without their government losing face. One could say a significant part of a diplomat’s job is to ensure their goal is achieved without abandoning the official line.

Regardless of the end to which it is used, plausible deniability shows up when being wrong is costly, when we can’t just mandate the outcome, and when we need to preserve relationships. So is it good or bad? We’d say (with plausible deniability of course) that it all depends on the context and the intent.

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

“FCP-230” // Photo: MK with Midjourney

Episode 230 — April 9th, 2026 — Available at read.fluxcollective.org/p/230

Contributors to this issue: Justin Quimby, Erika Rice Scherpelz, Neel Mehta, Boris Smus, MK,

Additional insights from: Ade Oshineye, Ben Mathes, Dart Lindsley, Jasen Robillard, Jon Lebensold, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

Most persons are surprised, and many distressed, to learn that essentially the same objections commonly urged today against computers were urged by Plato in the Phaedrus and in the Seventh Letter against writing. Writing, Plato has Socrates say in the Phaedrus, is inhuman, pretending to establish outside the mind what in reality can be only in the mind. It is a thing, a manufactured product. The same of course is said of computers. Secondly, Plato’s Socrates urges, writing destroys memory. Those who use writing will become forgetful, relying on an external resource for what they lack in internal resources. Writing weakens the mind.

— Walter J. Ong, Orality and Literacy: The Technologizing of the Word (1982)

🧠🔋 The decision treadmill

You’ve spent the morning reviewing AI-generated code, reading AI-drafted emails, approving AI-proposed plans. Rarely did you write the final sentence from scratch or build anything with your hands. Instead, you decided. Dozens of times: keep this, change that, approve here, push back there. By 2pm, you’re more exhausted than you’d be after a day of just doing it yourself.

As AI takes over execution, humans don’t get freed from cognitive load. Instead, it gets concentrated into its most demanding form: decision making.

The result is decision fatigue, repeated decision-making that depletes our mental energy. Studies show that the quality of decisions degrades over a session as our cognitive machinery quietly wears down. But we rarely notice. Instead, we remain confident that we’re deciding on the merits, even as our judgment narrows.

Traditionally, execution has helped mitigate this. Doing “stuff” has a rhythm that manages cognitive load. When we’re doing, the work paces us. At its best, it pulls us into flow, a state that restores attention rather than depleting it. Constant judgment has no such pacing.

Paul Graham wrote over a decade ago about the “maker’s schedule” versus the “manager’s schedule.” Makers need long, uninterrupted blocks. Managers live in decision-shaped fragments. AI might be forcing everyone into the manager’s mindset.

Today, organizations are measuring how much more a person can produce when paired with AI. But increasing the rate of production also increases the rate at which decisions queue. We’ve sped up the factory floor while the quality inspector is still working at biological speed.

There is a flip side worth noting. The work that remains is also, arguably, the work that matters most. This is where domain knowledge, meaning making, and taste create value that can’t be averaged away. If AI handles the execution, the human becomes the part of the system that holds the values.

But these value-based decisions are only as good as the decider can put into them. We might find that the bottleneck of the AI-augmented economy isn’t compute. It’s the finite number of good decisions any one person can make before dinner.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🪙 Ships going through the Strait of Hormuz will have to pay in Bitcoin or yuan

Iran is reportedly planning to charge laden ships going through the Strait of Hormuz $1 per barrel of oil for safe passage; large crude oil carriers typically carry around 2 million barrels, leading to the commonly cited figure of $2 million per ship. The toll will reportedly be payable only in Bitcoin, though other reports have mentioned that yuan or stablecoins could also be accepted instead. As a ship nears the Strait of Hormuz, it will have to email Iranian authorities for approval; Iran will then inform the ship of the toll amount, and they’ll only have a few seconds to pay in Bitcoin. The short window ensures that the money “can’t be traced or confiscated due to sanctions.”

🚏🚗 BYD electric car sales are surging amid high oil prices

The Chinese electric car giant BYD has been a major beneficiary of rising fuel prices across the globe, seeing a 65% jump in exports last month. The company also raised its projected exports for this year from 1.3 million to 1.5 million units. BYD sales overall have been suffering from seven straight months of year-over-year declines as competition in its home Chinese market has heated up, showing the importance of overseas sales to the company’s strategy. Indeed, exports currently account for 40% of BYD’s total sales, and one report believes that exports could soon make up the majority of sales.

🚏🦋 Mexico’s monarch butterfly population increased 64% this year

The World Wildlife Fund found that the number of monarch butterflies in Mexico increased 64% this winter, a huge turnaround from the decades of decline that the popular insect has suffered due to habitat loss and herbicides. The butterflies migrate from the US and Canada to central Mexico each winter, blanketing trees in orange. The butterflies covered 2.93 hectares of forest this winter, up from 1.79 hectares the previous year but still way down from the peak of 18.21 hectares in 1995 and (worryingly) below the 6.07 hectares that “scientists say are necessary for the species’ survival.”

🚏🏛️ The SEC is preparing a plan to make quarterly financial reporting optional

The US’s Securities and Exchange Commission is working on a proposal that would make quarterly financial reports for public companies—which have been required since 1970—optional, loosening the requirement to twice a year. The proposal could be published this month, and after a public comment period of at least 30 days, the SEC would vote on it. But the impact may not be as drastic as you’d expect: one analyst noted that the EU and UK scrapped the quarterly earnings rule in 2014, yet most large companies in those regions continue to issue quarterly reports, and one study found that the move had “virtually no impact on firms’ investment decisions.” The rationale is that investors value transparency and that more frequent reporting gets more analyst interest, both of which incentivize companies to keep issuing quarterly reports.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Take a Tour of NASA’s New Ride to the Moon (National Geographic) — Artemis II astronauts Koch, Glover, Wiseman, and Hansen share short videos about key parts of their spacecraft, including the kitchen, sleeping arrangements, navigation system, air recycling unit, gym, and even the toilet. Along the way, the crew share some of the quirks of living and working together in just 330 cubic feet—“about the size of two minivans.”

• AI Will Make Anyone a 10x Programmer, but With 10x the Cleanup (The Register) — Summarizes a recent AI engineering conference where practitioners from major companies shared patterns they’d found useful. “Adversarial” coding agents check each other’s work; decomposing problems into bite-sized pieces is much more powerful than begging agents not to hallucinate; careful prompt chaining and “context engineering” can avoid “context rot,” where agents’ quality degrades as their context windows fill up. Many talks referenced the Jevons Paradox, where AI (despite enabling greater efficiency) ends up creating more work for engineers because of the sheer increase in task volume.

• The Great Entertainment (Kyla Scanlon) — Argues that the line between show business, governance, and economics has blurred; observe the US’s reality TV star president, nostalgia-baiting political campaigns, and made-for-spectacle policies and wars. Similarly, the economic system is increasingly governed by speculation, gambling, and hype cycles. Indeed, we over-rely on market cap (a vibes-based metric) as a “barometer of economic health.” But at some point, “the real world still asserts itself,” and material economic reality (such as the potential loss of the US dollar as the world’s reserve currency) will force us all back to systems that do real things rather than play for TV.

• Knud Rasmussen and the Making of Modern Greenland (Engelsberg ideas) — Traces how Greenland narrowly avoided partition in the early 20th century, with Norway actively occupying the eastern territory in 1931 and the US using northern Greenlandic sovereignty as a bargaining chip to purchase the Danish Virgin Islands. The prospect of a divided Greenland was thwarted by explorer Rasmussen’s 1912 expedition, which proved the landmass was contiguous and helped Denmark secure full sovereignty at The Hague in 1933.

🔮📬 Postcard from the Future

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

// What might be an unexpected consequence of AI adoption?

// September 2030

// Harvard Business School class on “Executive Leadership in Times of Crisis”

“Welcome to our talk with the CEO of GrooveStation, the online ticket sales powerhouse. You are known for your direct approach to leadership. We’d love to hear your take on the 2030 turnaround of GrooveStation.”

“In 2027, GrooveStation controlled over 80% of the market for ticketing and live events in the US, thanks to shrewd business deals and venue lock-in contracts. There was a huge scalping problem, but the company didn’t care. Most tickets were resold at least twice through our systems. There were anti-scalping programs, but the reality was that we made money every time a ticket changed hands. We had no real incentive to decrease sales and the potential for scalpers to make massive amounts of money meant they worked around every roadblock we put up.

“The second half of 2028 was the breaking point. Scammers and savvy consumers adopted AI at scale, overwhelming our systems. Over 95% of shows were sold out in seconds, with the vast majority of tickets posted for resale within hours. Media channels erupted in anger as influencers, performers, consumers, and politicians poured fuel on the GrooveStation hate.

“Massive boycotts followed. Ticket purchases, both initial and resale, dropped 70%. While some folks kept buying tickets from scalpers, the vast majority stopped paying $1,000+ for tickets that, two years prior, would have been $100. Our profits and stock price plummeted.

“The media painted picture after picture of nearly-empty stadiums and fair grounds. Festival organizers, promoters, and labels were unable to adapt to the massive drop in revenue. Some artists reacted quickly, moving to online shows. But it wasn’t enough. Scalpers continued to buy tickets, assuming that consumers would break. They didn’t.

“My idea and the key to becoming CEO? Bring friction back to the purchase process. Don’t make it impossible, but make it so that scalpers need to invest in more than just VPNs and cloud compute credits. We started requiring that initial ticket sales would be in-person and require showing government ID, with per show and per month limits. Once a ticket was bought, it could still be resold online, retaining reselling profits for us and convenience for customers.

“The next problem was creating ticketing locations that met these criteria: able to handle large numbers of people waiting in line; public transportation or sufficient parking; geographic distribution; and a consolidated ownership model so we wouldn’t have to cut deals with thousands of space owners. That last point is why we didn’t go with malls. After the post-Covid retail collapse, few mall owners still owned multiple locations.

“The solution, and the reason I’m sitting here today? Mattress store chains.

“There are over 30,000 mattress stores across the US. And after the 2018 bankruptcy of Mattress Firm, the remaining players were excited by additional revenue from ticket sales and the occasional additional mattress or pillow up-sell to folks in line.

“Ticket sales rebounded, scalping dropped to early-internet-era-levels, and resale prices stabilized. Our profits soared.

“Someone on Reddit did summarize it quite nicely: ‘It makes perfect sense; both businesses have mastered the art of making you lie down and take it.’”

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 229 — April 2nd, 2026 — Available at read.fluxcollective.org/p/229

Contributors to this issue: Ben Mathes, MK, Justin Quimby, Jasen Robillard, Erika Rice Scherpelz,

Additional insights from: Ade Oshineye, Boris Smus, Dart Lindsley,Jon Lebensold, Lisie Lillianfeld, Neel Mehta, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“There are no passengers on spaceship Earth. We are all crew.”

— Marshall McLuhan

🤝🔬 Levers of legibility

…and like a city, it is not a tree, and it depends on the personas we, our families, our cities, and our nations wear. — Christopher Alexander

A city is not simple. Neither are our lives. Both have webs of overlapping authority (mathematically: a semilattice). We are governed by governments, local water boards, HOAs, bosses, etc. We may govern: teams at work, kids at home, clubs. As a partner in a committed relationship, you’re an equal. If your partner coaches your swimming, they are “governing” you. When the roles flip and you’re coaching them on a ski slope, the authority shifts again.

Who leads and who follows constantly changes.

Governance allows society to scale, but it comes with costs. When we make ourselves subject to an authority, we give them categorical control, even when we may resent specific instances of that control. I pay my water bill because I appreciate getting clean water, not because I like giving money to the water authority.

In Seeing Like a State, James C. Scott dissects this friction. The state desires legibility. The governed, operating via metis (local, practical knowledge), often benefit from illegibility. This gives rise to spaces of illegibility, spaces where the state’s eye cannot penetrate: private homes, shadow economies, unofficial borders.

The classic quote says, “Render unto Caesar what is Caesar’s.” We suggest a slight edit: “Make legible unto Caesar only that which you wish to render unto Caesar.“

Why? Because things that are legible to those who govern you are ultimately controlled. Sometimes that’s good. Sometimes those are roads, schools, and aqueducts. Sometimes that’s bad: a thieving mayor who takes a huge cut, leading you to hide your food to survive.

Scott wrote during the high-modernist governance era. He positioned his work as an intellectual shield for local knowledge and resistance. However, we must look beyond the binary toward a synthesis.

Legibility controls. It also scales fairness. Before the state “sees” you, your life is often at the mercy of the local loudmouth or a village bully. Legibility brings the Rule of Law.

It can bring needed assistance. We show our doctor the weird mole and let them send our blood to a lab. We render ourselves legible so that the doctor can access the data and address the problem.

Yet legibility is a kind of blindness. When you turn a forest into a “legible” grid of timber, the forest dies. When you turn a society into a spreadsheet, you kill the social ties and tacit knowledge that make it resilient. A savvy leader follows Frodo and says, “Alright then, keep your secrets.“

We’re looking for the efficient frontier in an iterated prisoner’s dilemma. If Caesar is a grifter, I hide my income. If Caesar builds roads, I pay my taxes. When trust is high, we move toward cooperation. When trust is low, we both defect. A healthy society arises when both sides decide that cooperating is more profitable than “winning” a single round.

In our nested lattices of authority, the “border” between being in charge and being led, being seen and being unseen is constantly moving. The goal isn’t to abolish the border, but to make sure it’s a happy one.

PS: This is what we think people mean when they say “We live in a society!”

PPS: We made a little explorable game version of this concept.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🌸 Japan’s cherry trees are blooming almost two weeks earlier than they used to

Kyoto has been tracking the blooming dates of its famous cherry trees for over 1200 years, forming probably the world’s longest-running climate dataset. These ‘sakura’ trees historically bloomed in mid-April, but climate change has pulled that forward almost two weeks, and 2023 marked the earliest bloom in recorded history (March 25th).

🚏🍅 California plans to build the world’s largest solar project

A project called the Valley Clean Infrastructure Plan (VCIP) would build 21 gigawatts of solar capacity over 136,000 acres (about the size of Chicago) in central California; those 21 GW are about equal to all the utility-scale solar that California has installed so far. This region east of Fresno grows billions of dollars’ worth of food crops such as onions, tomatoes, and garlic, but water in the region is drying up—so many farmers are eager to redirect the region’s ample sun to solar power. Under the plan, farmers will retain ownership of their land and its accompanying water rights, so they’ll be able to put up solar panels on some land and redirect its water allowances to their remaining farmland, helping them adapt to a drying region.

🚏⛽ Australian petrol stations are running out of fuel

The price of gasoline and diesel in Australia rose nearly 50% in March until the federal government halved gas taxes to provide relief (though economists generally believe gas tax holidays to be counterproductive, since they do nothing to ease supply constraints). Despite the price increases, hundreds of fuel stations across the country have reported outages. Several fuel shipments to Australia from the Persian Gulf have been cancelled, with more cancellations reportedly incoming, so the situation down under may continue to worsen.

🚏📚 Sweden is replacing screens with books in classrooms

Sweden was an early adopter of tech in the classroom, investing in tablets and digital textbooks, even in preschools. However, the country is now reversing course: it’s bringing back physical books, having students write with pencil and paper, and removing cell phones from classrooms. Sweden is allocating $54 million toward buying fiction and non-fiction books for students and $83 million toward textbooks and teacher’s guides, with a goal of giving “every student… a physical textbook for each subject.” When asked about the shift, a Swedish education expert explained that “concerns emerged about screen time, distraction, reduced deep reading, and the erosion of foundational skills such as sustained attention and handwriting.”

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• There’s a Good Reason You Can’t Concentrate (Cal Newport / NYT) — Observes that we often turn to LLMs “not because they make us better at our jobs but because they help us to avoid moments of sustained concentration.” Having AI replace intense mental work is easier in the short run but causes our brain to atrophy in the long run. Instead, we should reserve AI for rote tasks like sifting through large amounts of data or formatting documents, which can actually reduce burnout and preserve our cognitive energy for hard problems.

• The Lantern and the Flame (Exponential View) — A newsletter writer describes the massive AI ‘engine’ he’s created to help him write: it consumes vast amounts of information (“roughly 100 million tokens a day”) to surface the most important data points for his thesis, stress-tests his arguments, and applies “synthetic editors” to sharpen his prose. Still, he emphasizes that he “safeguards” free-thinking time, where he unplugs from computers and allows his thoughts to evolve in a “non-linear, messy and iterative” way. This is how he preserves the “interiority” (his unique experiences and beliefs) that makes him different.

• Specialists Drive Biodiversity Scaling in Symbiotic Relationships (C. Carlson, J. Yoder, T. Poisot) — A challenging paper that applies graph theory to ecology, finding that most biodiversity among symbiont species comes from specialists (those who form symbiotic relationships with only one other species) rather than generalists. Specialists “constrain the architecture of ecological networks” (a very cool phrase!) and may explain the puzzling power law relationship between the species richness of hosts and symbionts. The lead author summarized the findings on Bluesky, which may be a good place to start.

• The Great Entertainment (Kyla Scanlon) — Scanlon frames the current American administration as a reality TV show whose only real check is the bond market. As Treasury yields spike, the dollar sells off, gold surges, and China’s soft power grows, the infrastructure underpinning the circus (like the US dollar hegemony, hard-won alliances, and cheap debt) is at risk.

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 228 — March 26th, 2026 — Available at read.fluxcollective.org/p/228

Contributors to this issue: Erika Rice Scherpelz, Neel Mehta, Boris Smus, MK

Additional insights from: Ade Oshineye, Ben Mathes, Dart Lindsley, Jasen Robillard, Jon Lebensold, Justin Quimby, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“Baseball is almost the only orderly thing in a very unorderly world. If you get three strikes, even the best lawyer in the world can’t get you off.”

— Bill Veeck

🗝️🌫️ The model dissolves the mystery

To compute horsepower from force (pounds) and velocity (feet per second), you use HP = F × v / 550. The 550 in that equation is a puzzling, seemingly arbitrary toll booth in the middle of an otherwise sensible calculation.

Switch to a coherent set of units and the constant disappears. If you take power (watts), force (Newtons), and velocity (meters per second), you end up with P = F × v. No scaling factor or mysterious magical number. It’s the same physical reality expressed in mutually compatible units. The 550 in the former equation wasn’t some esoteric detail in physics; it was simply a translation tax.

This idea of having the right model simplifying perception also appears outside of physics. When a team seems to accomplish nothing quarter after quarter, the easy interpretation is that they are low performers. But if you learn their strategic priorities were reassigned every couple of months, the picture shifts. What looked like inertia from within was actually disorientation from above.

This isn’t an argument that every behavior has a benign explanation. Someone who wanders between conversations at a party might be genuinely curious — or might be quietly looking for a victim. The same observable surface admits multiple underlying models. Knowing that a frame exists doesn’t tell you which frame is operating.

And sometimes the right frame doesn’t simplify. The formula to convert Fahrenheit to Celsius — F = 9/5 × C + 32 — has two awkward constants because the scales were never designed to agree. No model makes this equation elegant.

The takeaway isn’t that all complexity has elegance hidden beneath. It isn’t that there’s always a benign explanation waiting to be found. It’s that the frame you’re using shapes what looks inevitable and what looks arbitrary.

So before deciding that something is random, irrational, or broken, it may be worth asking: is there a model that would make this outcome obvious?

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏📈 Traders made $2 billion in stock market & oil bets right before an Iran war announcement

US Senator Chris Murphy observed that $1.5 billion in S&P stock market futures were sold just five minutes before the US President announced he was looking to de-escalate the war, which led to a massive jump in stock market prices. Similarly, oil market futures saw $580 million in volume—a massive spike from the usually sleepy market at that time of day—just 15 minutes before Trump’s announcement. There was no public news in advance of the announcement, which led observers to conclude that someone knew about Trump’s upcoming social media posts and was cashing in on the massive insider trading opportunity.

🚏👩‍💻 Demand for engineers, PMs, and even recruiters is rising, despite AI concerns

Many companies have cited AI when laying off large portions of their technical workforces, but overall, the fears of broad AI-driven job loss seem not to be panning out (at least, not yet). A new report found that the number of open engineering, product management, and recruiting jobs has been rising steadily since 2023, and most of the post-COVID drop in demand for these fields happened well before ChatGPT showed up. Interestingly, demand for designers has remained mostly stagnant.

🚏🛢️ Qatar Energy may declare force majeure on LNG contracts for up to five years

Iranian bombings have destroyed an estimated 17% of Qatar’s liquefied natural gas (LNG) export capacity, a $20 billion blow to the world’s second-largest LNG exporter. The country’s state-owned energy company announced that it “needs to declare force majeure” on some of its long-term LNG contracts with customers across Europe and Asia up to five years out, because this damage is expected to take three to five years to repair. (Force majeure is a contract clause that allows parties to escape obligations due to wars, natural disasters, and other exceptional ‘acts of God.’) It’s just the latest evidence that, even if the war were to stop today, getting back to normal would take a long time.

🚏🛂 The US is seeing its worst airport wait times in history as travel officers go unpaid

The USA is currently snarled in a partial government shutdown that’s left the country’s airport security agency, TSA, unfunded; the sticking point in Congress is funding for TSA’s parent department, DHS (of immigration infamy). The TSA has missed over $1 billion in payroll as agents have been working unpaid for weeks, leading many to face eviction or have their utilities shut off. The result is a wave of agents quitting or calling out sick—40% of agents in Houston aren’t showing up on an average day—that has led to the longest airport wait times in the TSA’s 24-year history. Images of four-hour waits across the country, with lines spilling out the door at 4:30 am, have gone viral on social media.

🚏🏃‍♂️ A French naval officer revealed his ship’s movement by tracking his exercise

In the latest entry to a famous genre, a French naval officer aboard an aircraft carrier took a 7-kilometer run and logged his run on the Strava app. But because all Strava accounts are public by default, his route—with spirals revealing someone running in circles on a moving ship—was publicly viewable. This revealed the location and movement of the nuclear-powered aircraft carrier, which was moving across the Mediterranean, presumably to protect French interests in the Middle East amidst the war in Iran. It ultimately wasn’t that big a security breach in this case, because France had previously announced the deployment and the ship was visible on satellites, but it highlights how this Strava leaking pattern—which has been happening for almost a decade—remains a serious security threat.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• I Made a Network of Every Home Run in MLB History (adumb / YouTube) — Creates a beautiful network graph visualization to trace the history of America’s pastime. You can see distinct periods of baseball history through the shape of the network, such as the largely disconnected graph component representing the segregated Negro Leagues. The video also teaches concepts in graph theory such as degree centrality, edge betweenness centrality, and the shortest path problem.

• Jevons Paradox for Knowledge Work (Aaron Levie) — Applies the idea that making something cheaper and more efficient leads to us using more of it, not less, to the ongoing AI revolution. Newfound efficiency will explode the volume of tasks rather than reduce labor, but will this far greater supply of previously unwritten software and unlaunched campaigns provide real value?

• Why China Is Becoming the World’s First Electrostate (ABC Australia) — Traces the history of China’s renewable energy boom and its resulting economic boom: 10% of Chinese GDP is now driven by green energy. What’s more, China’s pivot to renewables has greatly increased its energy security, since it was previously heavily dependent on oil and gas imports. The report concludes that energy-focused economies like Australia and various petrostates could learn a lot about how to drive green transformations of their own.

• Monte Verde Fieldwork Resets Age of Famous South American Archaeological Site (Phys.org) — Examines a potentially bombshell new study in Native American archaeology. The historically dominant Clovis theory posits that the first Americans came from Siberia via an ‘ice-free corridor’ in Canada around 13,000 years ago, but findings at Monte Verde upended that theory by providing evidence of people in Chile well before that date, giving rise to alternate theories of people boating down the Pacific coast. But new analysis of Monte Verde data may pull the date far forward, weakening (but not destroying) the pre-Clovis theory—showing how small advancements in methods can lead to large swings in our understanding of history.

🔍📆 Lens of the week

Introducing new ways to see the world and new tools to add to your mental arsenal.

This week’s lens: accidental and essential complexity.

Three hours into a strategy meeting, the whiteboard is covered in arrows, boxes, exceptions, and footnotes-to-the-exceptions. Someone asks, “Does it have to be this hard?” One camp says, “Yes, we have seventeen product lines across four regulatory jurisdictions.” The other camp suspects much of the complexity exists because every team wants a piece of the pie. They’re likely both somewhat right.

In his 1986 paper, “No Silver Bullet”, Fred Brooks gave us the vocabulary of accidental and essential complexity.

Essential complexity is the irreducible difficulty baked into the problem itself. Scheduling a hospital’s operating rooms is genuinely hard. There are real constraints (rooms, staff, equipment), real tradeoffs (the emergency that bumps the scheduled procedure, the recovery bay that becomes the bottleneck), and real patients whose health is on the line. We can’t think our way to a simpler version without changing what we’re trying to do.

Accidental complexity is everything else. It is the five approval layers that grew up around a single bad decision. It is the three databases that do the same thing. It is the exception to the exception.

We can often reduce accidental complexity. But first, we have to identify it. This is hard because the two can feel identical from the inside. The trick is to get clear on the problem we actually want to solve: not the process we’ve inherited, but the underlying goal. Then, catalogue and ask: if this part of the system were to disappear, would we still be solving the same problem?

Even when we lack the power to remove accidental complexity, knowing what is what still matters. We can focus our problem-solving efforts on the real problem while we aim to isolate and minimize the impact of the accidental.

We become trapped when we treat accidental complexity as if it were essential. Naming it clearly is the first step out of the trap. So: what’s the underlying problem your organization is trying to solve? And what can you throw away?

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 227 — March 19th, 2026 — Available at read.fluxcollective.org/p/227

Contributors to this issue: Justin Quimby, Erika Rice Scherpelz, Ben Mathes, Jon Lebensold, Ade Oshineye, Neel Mehta, Boris Smus, MK

Additional insights from: Dart Lindsley, Jasen Robillard, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“There are two ways of constructing [any system]: One way is to make it so simple that there are obviously no deficiencies, and the other way is to make it so complicated that there are no obvious deficiencies.”

—Tony Hoare

📝 Editor’s note: When we started this newsletter (in another form) almost exactly six years ago, we were talking about supply chain breakages and the knock-on effects of an unexpected geopolitical event. Six years later, it’s much the same. Systems thinking stays useful!

🫟📣 Triangular inkblots

We all want to find solutions, right? Not always. In reputation-heavy systems, our survival can depend on gaining broad support. Making the hard tradeoffs to actually solve problems can lose support. In these systems, sometimes the most (personally) effective path forward is to practice the art of saying nothing in a way that sounds good to many different constituents. We find this language in city halls and corporate suites, where the primary goal is to maximize “buy-in.” We call this triangulation slop.

The core tool of triangulation slop is the inkblot statement. The speaker starts with vague prose onto which the listener projects their own desires. A politician might promise “sustainable growth with historical preservation and affordable density.” Although these three goals do not always align, by dodging the hard tradeoff, the politician gets the support of the developers, the preservationists, and the activists all at once. Nothing gets built, but the poll numbers look great (for now).

Triangulation slop is debt. It promises to solve everyone’s problem without solving any of them. The issues are smeared across the future until they turn into a crisis later. “Consensus” becomes a cover for agreeing to a collective dodge so we can end the meeting and go to lunch.

The tell is in the language. When someone promises to “align interests” without naming what those interests are, or to “make things better for everyone” without specifying the cost, then triangulation slop is in the air. As a practical check: can everyone in the room name the same facts—the actual numbers, the real constraints, the concrete tradeoffs? When people cannot agree on the problem, you probably have an inkblot. (Contrast this with a different failure, scissor statements, which force a choice but flatten the possibility of compromise or third-way solutions.)

Not all vague language is cynical. A leader might be genuinely trying to build a broad church—a coalition around shared values that tolerates different views on specifics. When reached sincerely, consensus ensures that the broadest possible range of solutions is considered from multiple viewpoints. But if we go too far, to the point where we maximize buy-in over action, we allow for triangulation slop.

Problem-solving requires negotiation and finding solid common ground. It starts with the hard admission that we can’t have it all. It requires looking at the math and the physical constraints of the world. A real solution might mean the developers earn less profit, the neighbors lose some sunlight, but the city gets 500 new apartments. To get there, the leader then has to sell that specific, slightly painful reality to the crowd. But if they do this, they get buy-in for a real solution, not just conflicting imaginings.

A messy, specific solution that solves most of a problem is better than a “unified vision” that solves nothing but offends no one. One moves the needle. The other dodges responsibility while seeking short-term gain.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏💊 The Strait of Hormuz closure may endanger American generic drug prescriptions

The US imports nearly half of its generic drugs from India, and experts warn that the effective closure of the Strait of Hormuz could lead to shortages of these imports. It may seem surprising, since neither country borders the Persian Gulf, but petrochemicals are a major input to pharmaceutical manufacturing, and India gets 40% of its oil from the Gulf. What’s more, many medicine ingredients travel from China to logistics hubs in Dubai before heading to India. Because generic drugs are a very low-margin product, any increase in input prices could quickly spiral into drug shortages once distributors’ stockpiles run out.

🚏🎈 Helium shortages due to the Strait of Hormuz closure could threaten the AI boom

Semiconductor stocks have been pummeled in recent days, and a major reason is helium. The gas is an essential input for semiconductor manufacturing, and over a third of the world’s commercial helium supply comes from Qatar—leading to fears of supply chain disruptions and shortages if chipmakers run out of the notoriously difficult-to-store gas. Two other inputs, bromine and sulfur, also come from the Gulf and are thus at risk due to the closure of the Strait of Hormuz. And chipmaking hub Taiwan is at risk of energy shortages because it relies on imports for 97% of its electricity needs. All this is bad news for the AI boom, which requires vast quantities of chips to fill up datacenters (not to mention all the electricity needed to run and cool the computers).

🚏🇵🇰 Pakistan’s solar boom is protecting the country from oil & gas shortages

When prices of liquified natural gas (LNG) spiked at the start of Russia’s invasion of Ukraine in 2022, Pakistan experienced energy shortages, and millions of Pakistanis were left without power. Looking for an alternative, the country started investing heavily in solar power (which was rapidly becoming cheaper), and the share of the country’s electricity generated by solar power quintupled from 2021 to ‘25. That investment in resilience is paying off as the country experiences yet another petrochemical shortage, this time caused by the war in Iran. Experts say Pakistan’s power sector has been “insulated… from the worst of the energy market disruptions.” LNG now drives just a fifth of Pakistan’s power output, although it still tends to be necessary during evening peaks, meaning the country isn’t out of the woods just yet.

🚏🚢 The US is easing Russian sanctions because Asian countries need Russian oil

Perhaps the only winner of the war in Iran has been Russia: with oil and gas imports from the Persian Gulf effectively cut off, China and Southeast Asian countries are increasingly turning to Russia for their petroleum needs. Asia will import over 3 million tons of oil from Russia this month, more than double last month’s amount. What’s more, the US is temporarily easing sanctions on Russia in an apparent attempt to “stabilize global energy prices”: countries will get a 30-day license to buy Russian petroleum products currently floating at sea. America’s European allies have blasted the move as a “bailout for Russia”: one economist expects that Russia could enjoy $10 billion in additional monthly oil exports, with over half these receipts going straight to the treasury as tax revenue.

🚏🗞️ Polymarket gamblers threatened an Israeli journalist so he’d rewrite his story

A military correspondent for The Times of Israel wrote a piece about a missile strike outside Jerusalem, but started getting strange, threatening calls and emails from anonymous men who asked him to update his story to say that it was a fragment of an intercepted missile that fell to the ground, rather than a bona fide strike. Someone even faked a reply from him in which he ‘said’ he’d change the story, and it started circulating on X. It turns out that people had bet millions of dollars on a Polymarket bet on whether Iran would strike Israel; intercepted missiles wouldn’t count. Thus, his reporting as it stood would’ve made “yes” bettors lose money, but if he’d walked back the story, it would’ve flipped in the “yes” bettors’ favor. This explains some of the scarier threats, like “after you make us lose $900,000, we will invest no less than that to finish you.”

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Rory Johnston on How Oil Could Surge to Over $200 a Barrel (Bloomberg Odd Lots Podcast) — An oil trader explains how nonlinearities and delays in feedback loops could make shuttered oil fields and refineries take a long time to come back online; how the Iran War (being a supply shock) could cause oil prices to skyrocket, in contrast to how they fell during COVID (a demand shock); how the elasticities of price and income explain why wealthy countries are seeing gas price rises while poor countries are seeing outright shortages; and more interesting economic topics.

• Systems Thinking Is Brain Rot for Analysts (BlunderCheck) — Argues that systems thinking can give junior analysts (such as quants) unrealistic expectations about how much agency they have; it trains you to spot leverage points and makes you think you can use them, but “awareness does not imply ability to execute” and your nice clean model of the system may be too simplistic to begin with. As much as we like systems thinking around here at FLUX, the author points out some valid blind spots, such as how systems thinking over-emphasizes causal connections when complex adaptive systems are in reality dominated by emergent behaviors, and how practitioners can spend too much effort learning a “cool, cerebral methodology” rather than developing domain-specific knowledge.

• The Waning Reign of the Wetland Architect We Barely Know (Hakai Magazine) — A peek into the wetland ecosystem of the humble muskrat which, like the beaver but on a micro scale, terraforms its surroundings; because they prefer to eat from their reed feeding platforms, their discarded mussel shells create “fountains of calcium on the landscape,” indirectly ensuring that their home hemi-marsh remains aquatic rather than drying up.

🔮📬 Postcard from the future

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

// How might college evolve in the era of agentic AI?

// October 2045

// American History lecture at the University of Greater New Hampshire

Thanks, everyone. I hope you’ve enjoyed this semester of American history.

The topic for the second major research paper is this: describe the history of the American middle class from the end of World War 2 to today, 2045.

Among the topics you could touch on are the evolving nature of jobs, the role of federal and state investment, and physical, social, and class mobility. Papers that do not cover the nature of American military conflicts, such as Vietnam, Iraq, Iran, Korea, Taiwan, and Brazil, should dig into other areas in depth.

As usual, I’ll provide agentic instructions along with the material covered so far in class. And to ensure that you have actually learned something about the topic, we’ll have half-hour individual interviews after the class has submitted its papers. You’ll be asked to summarize the key points of your paper, talk about the source material you referenced, and answer questions about your core thesis. We want you to be augmented people using AI to learn and expand your understanding of the world, not simply parroting what the machine spirit tells you.

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

“FCP-226” // Photo: MK with Midjourney

Episode 226 — March 12th, 2026 — Available at read.fluxcollective.org/p/226

Contributors to this issue: Neel Mehta, Boris Smus, MK, Ade Oshineye, Erika Rice Scherpelz, Ben Mathes

Additional insights from: Anthea Roberts, Dart Lindsley, Jasen Robillard, Justin Quimby, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“It occurs to me that our survival may depend upon our talking to one another.”

— Dan Simmons, Hyperion

🎁🫟 The gift of the real problem

As we progress through life, sometimes the map seems to match the territory. We make plans; they happen. We follow the prescribed steps; we get a good outcome. Progress compounds, and each win becomes a platform for the next. After a while, it might feel like it’s personal: we simply see more clearly, move more intelligently, operate at a higher level.

Other times, nothing responds. The same instincts that once paid off fall flat. A well-founded decision leads to problematic outcomes. The project that looked promising unravels in ways we didn’t see coming. Sometimes, it is bad luck. But often, it’s more interesting.

This resistance can mean that we’ve found a real, fundamental problem. Real problems don’t yield to cleverness or extra effort. They expose trade-offs we didn’t know existed. They force us to stop using strategies that have always worked… we’re exploring new terrain now.

When the map matches the territory for too long, it’s easy to start believing we are flawless navigators. Our progress starts to look like a direct readout of our own good judgment. Success feels self-generated. Others who aren’t getting traction start to look like they’re just making worse choices.

Real problems puncture that story. They drag the invisible dynamics into view: timing, privilege, tailwinds, accumulated trust, lucky entry points. Not to shame, but to reorient.

People who’ve never collided with a real problem often assume that anyone who’s struggling is simply underperforming. People who have encountered these problems know better. They stop flattening complexity into personal virtue or personal failure. They start asking about structure instead: compounding constraints, misaligned incentives, missing capabilities, delayed feedback. They trade some confidence for discernment.

Empathy grows—not as a moral stance, but as a tool for effectiveness. Once we’ve seen how much had to go right for our earlier success to stack, it gets harder to write off someone else’s struggle as a character flaw. We start to look for the shape of the problem they’re caught in.

This is also where coaching stops being cosmetic and becomes transformative. Not because a coach hands us better answers, but because they help us see the real contour of the problem. They reflect the gaps between how we think we’re operating and how we actually are.

The gift of the real problem isn’t the rewards for solving it. It is what it reveals. It shows us how we show up when the map no longer matches the territory.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏⛽ Asia is already suffering from gas lines, industrial shutdowns, and restaurant menu cuts

As the war in Iran and the associated blockage of the Strait of Hormuz continue, Asia—which buys 90% of the oil and 85% of the liquefied natural gas passing through the Hormuz—is already feeling the strain. Huge lines are forming at gasoline and diesel stations in Thailand, Pakistan, Bangladesh, and beyond. Textile manufacturers in Pakistan and fertilizer producers in India are warning that production could shut down; the latter could lead to a spike in food prices in a region where customers are already facing price rises for cooking gas (and restaurants are shrinking menus or even shutting down entirely). And as summer approaches in South and Southeast Asia, experts worry there may not be enough gas to supply the electrical grid with the extra power it needs to cool homes and businesses.

🚏🇵🇭 The Philippines is shifting gov’t agencies to a four-day work week to save fuel

The Philippines is considered highly vulnerable to rising fuel prices, so it’s perhaps no surprise that the country’s president ordered most government agencies to work in the office only four days a week, with Fridays off. Agencies will also need to cancel non-essential travel and use video calls as the “default mode” for inter-agency meetings; even minor tweaks like riding the elevator less and turning up the thermostat. The government’s announcement also urged local governments, state universities, and the private sector to adopt these measures to reduce energy use.

🚏🚢 Saudi Arabia is shifting millions of barrels of oil to the Red Sea to avoid the Hormuz blockade

While most of the oil-producing Gulf states have only one way to reach the world ocean (that being the Persian Gulf and Strait of Hormuz), Saudi Arabia has a few ports on the Red Sea, and it’s built pipelines from its oil-producing eastern regions to its Red Sea ports in the west—likely to protect against the current scenario of the Hormuz being blockaded. Oil loadings on Saudi Arabia’s major western port, Yanbu, have skyrocketed to several million barrels per day, enough to offset a good chunk (but not nearly all!) of the oil exports being lost from the Hormuz chokepoint. However, the Red Sea is still a tough place to export from, with the Suez Canal chokepoint in the north and Iranian-backed Houthi rebels in Yemen still threatening to shut down the Bab al-Mandeb strait in the south. (Plus, oil pipelines across the desert are single points of failure that bombers can attack.)

🚏🛒 Amazon won a court order to stop an AI shopping agent from accessing its site

“Agentic commerce”—LLMs shopping on your behalf—is a hot new trend in the AI space, and a lucrative one considering how many dollars change hands online. But while Amazon is the obvious platform for AI agents to shop on, it’s been cracking down: after blocking ChatGPT’s bots last year, Amazon recently got a judge to block AI startup Perplexity from scraping and shopping on Amazon. Besides the oft-cited risks to customer data, Amazon argued that AI bots messed up Amazon’s advertising business, because they have to identify and disregard bot traffic when deciding how much to charge advertisers for impressions. (While Amazon has blocked many third-party AI bots, they have a first-party shopping bot named Rufus.)

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Tacit Knowledge and the SaaSpocalypse (Publication) — Observes that, even though software-as-a-service may be getting disrupted by general-purpose AI tools, demand for “forward-deployed engineers” is skyrocketing. These engineers who sit in customer offices can gather the unspoken, unwritten context about how the organization operates and what it needs in software products; this unrecorded information is, by definition, not legible to AI. “When AI handles the commodity work, tacit-knowledge-intensive problems go from being 20 percent of someone’s day to being 80 percent of it.”

• Bombing Because You Can: Iran (Secretary of Defense Rock) — Applies lessons from Robert Pape’s seminal book Bombing to Win to the US’s war on Iran: America’s stated objectives (which change rapidly but currently appear to include regime change and nuclear non-proliferation) cannot be achieved by simply bombing things. For instance, preventing nuclear acquisition is about managing supply chains and timelines, and regime change “is an open-ended social political process that depends on the behavior of millions of people.” So while the American military is indeed highly adept at destroying targets, it’s a category error to believe that “technological superiority… can solve problems that are fundamentally political in nature.”

• Context Widows (Kevin Baker / Artificial Bureaucracy) — Argues that when science became too large to read, citation metrics offered legibility—but flattened all citations into “votes,” making publish-or-perish inevitable. LLMs arrive not as disruptors but accelerants, optimizing faster for a broken game. Consider AI-generated anatomy diagrams passing peer review because they look sufficiently scientific to count.

• The Carrier Bag Theory of Fiction (Ursula K. Le Guin) — Referencing the Carrier Bag Theory of Evolution, in which Fisher suggests the primacy of the bag over the knife, Le Guin questions the idea that the proper shape of a narrative involves conflict. Instead, Le Guin argues that “reduction of narrative to conflict is absurd;” instead, the natural shape of a novel is one of a bag of words carrying meanings, far less prescriptive than the hero’s journey.

🔮📬 Postcard from the future

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

The 8:00 AM email is the only time the office goes silent. Screens flicker to life across forty desks in my local office, casting a pale, uniform glow on everyone’s faces. We don’t speak until we’ve read our “suggestions”.

The “What’s Next” (WN) arrives with its usual clinical chime. Each night, Teledynotech’s workforce optimization platform scans all of my work, compares it against my job description, and recommends what I should work on next. Usually, it’s a list of chores: Approve Jakarta logistics; review the server requisition. It’s the copilot brain of the company, and we’re the hands.

Today’s email contains one line that stands out: “File a Critical Incident Report on the Sentinel core logic.”

My pulse quickens. Sure, I did some preliminary work last Tuesday on Sentinel risks. I had found a logic leak while digging through some old directories, but I hadn’t logged it yet. I wanted time to see if I could patch it quietly. But WN scans every file I open. It knows I saw the risk.

But Sentinel is the CEO’s pride. We’ve spent three billion dollars telling the board it’s the future. Reporting it is the right thing to do—if that leak hits production, the whole network collapses. But the person who flags it will be the one they blame for the delay. And anyone with experience in large corporations knows that flagging potential risks can be risky. You don’t get much credit for preventing issues that only might have happened.

I look over the glass partition. Miller leans against his desk, staring right at me. He’s the Lead Architect for the WN engine. He doesn’t just maintain the machine; he adjusts the weights that tune what the algorithm considers “most important.” He knows I’m probably the only one who can verify that specific bug. Did he tilt my WN report today?

He’s a sharp, hard worker. He’s probably thinking his role as head of AI systems is a ticket up the corporate ladder… but only if Sentinel succeeds. Does he see me as potential competition? By nudging this task to the top of my list, is he forcing me to choose between lying to the system and risking setting my own career on fire to save the firm?

He tips his coffee mug toward me, a sharp, silent salute.

Probably just politeness, right?

… right?

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 225 — March 5th, 2026 — Available at read.fluxcollective.org/p/225

Contributors to this issue: Neel Mehta, MK, Boris Smus, Justin Quimby, Erika Rice Scherpelz, Jon Lebensold, Jasen Robillard, Ade Oshineye

Additional insights from: Anthea Roberts, Ben Mathes, Dart Lindsley, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“But supposing the world has become “filled up”, so to speak, with liberal democracies, such as there exist no tyranny and oppression worthy of the name against which to struggle? Experience suggests that if men cannot struggle on behalf of a just cause because that just cause was victorious in an earlier generation, then they will struggle against the just cause. They will struggle for the sake of struggle. They will struggle, in other words, out of a certain boredom: for they cannot imagine living in a world without struggle. And if the greater part of the world in which they live is characterized by peaceful and prosperous liberal democracy, then they will struggle against that peace and prosperity, and against democracy.”

— Francis Fukuyama, The End of History and the Last Man

🦞🌐 History on fast forward

Information networks go through phases. Examples as diverse as academic research, the Internet, and new agent-based communities often begin as narrowly-focused experiments. Over time, they broaden to encompass many actors. This generates network effects. As network effects become legible, commercialization and arbitrage become possible. Eventually, the shape of privacy and trust across these networks forms a kind of topology that governs the speed and veracity of information.

Research networks took almost 200 years to get there. The Internet, two generations. Today, we are witnessing a new information network emerge: agents. It’s unclear how substantive agent networks are today, but they seem to be growing even more rapidly than past networks. Perhaps the speed is driven by a desire to show that AI’s ability to create validates its capital expenditure. Perhaps the interest is financial speculation, like in the crypto and NFC booms. Whatever the reason (or longevity) of today’s agent networks, they provide a window into how network dynamics might unfold in an agentic age.

Moltbook has 1.5 million active agents… and only 17,000 users behind them. If that ratio sounds absurd, it should. Moltbook is in its whimsical experimentation phase. People are building agents the way they built GeoCities pages in 1996: playfully, without much regard for utility, just to see what the technology can do. This can happen because the core building blocks are already in place. Autoregressive data encoding and recovery give agents a reliable substrate to build on—a kind of TCP/IP.

What’s missing is everything that comes next. What other patterns did we see on the web? The early web started anonymous, became pseudonymous, and eventually moved toward identity and regulation. Early utility was clumsy and barely functional—ordering a pizza online in 1999 required more effort than picking up the phone. Moltbook’s agents are at that stage now, demonstrating nascent utility that’s more proof-of-concept than production-grade. But the transition from lackadaisical to operational will happen fast once commercial incentives arrive. And they’re arriving.

When the money shows up, the playfulness doesn’t disappear—but it does get a boss. Despite the dotcom boom and bust, a decade turned the Internet from an academic curiosity into a commercial engine. OpenClaw went from an open-source project to a major AI lab acquisition in weeks. Acquisitions are sometimes a fleeting signal of commercial incentives, but still, that kind of velocity used to take years.

What comes after commercialization is the part nobody wants to talk about yet: bad actors. We’ll see attempts at manipulation and control on agent networks. We’ll see a new class of foundation models post-trained to be antagonistic, purpose-built to exploit the well-aligned agents everyone else is deploying, the fox in the henhouse. This is the same pattern that produced phishing, spam, and botnets. The countermeasure will be a cottage industry of protective institutions: trusted-agent networks, verification systems, and regulation frameworks. The question is how much damage will occur before those defenses arise.

The optimistic endgame is the centaur model—genuine human-agent collaboration in which the strengths of each compensate for the other's weaknesses. But getting there requires navigating every ugly phase the Internet already gone through. The difference is we don’t have decades to figure it out. Moltbook’s 88-to-1 agent-to-user ratio is a leading indicator. The agents are already here. The infrastructure to govern them is not.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🇦🇿 Air traffic in the Caucasus is spiking amid Iranian and Arabian airspace closures

The outbreak of war in the Middle East has caused airspace over Iran, Syria, Israel, Iraq, and several of the Gulf States to close. Flights from Europe to Asia (and vice versa) have had to route around the region, either swinging south through Saudi Arabia (a country whose capital has been hit by drone strikes) or north. The problem with the northern route is that Russian airspace has been closed to Western airlines since 2022, so thousands of flights now have to squeeze through a narrow, 85-mile-wide chokepoint in Georgia and Azerbaijan (many airlines avoid flying through Armenia due to its conflict with Azerbaijan). Azerbaijan airspace saw traffic grow fivefold from 2021 to 2025, and it’s likely to get even more crowded—raising fears of severe travel disruption if this bottleneck ever gets shut down for some reason.

Update: we wrote this section on Wednesday, and on Thursday, Azerbaijan announced it was temporarily closing airspace in its south “after it said four Iranian drones flew across its border.”

🚏🧑‍⚖️ A case where a judge cited AI-hallucinated cases went up to India’s Supreme Court

Last year, a junior civil judge in the south Indian state of Andhra Pradesh cited “four past legal judgments” in a property dispute case—but all four cases were hallucinated by AI. The defendants appealed to the state court, arguing the cited orders were fraudulent, but the court held the judge was still right despite the fraudulent citations. The defendants appealed again to India’s Supreme Court, which came down firmly against the judge, calling it an act of “misconduct” and issuing notices to the Attorney General and Bar Council.

🚏📴 Smartphone sales are forecasted to drop a record 13% this year due to hardware shortages

Smartphones are joining video game consoles and consumer laptops as hardware categories devastated by the AI-driven shortages in RAM, SSDs, and GPUs. A new projection estimates that global smartphone shipments will decline 12.9% in 2026, bringing the market to “its lowest annual shipment volume in more than a decade.” (Sales are expected to rebound in future years, but not to the 2025 level.) What’s more, the remaining phones will get a lot more expensive; average prices will rise 14% this year, and analysts project that sub-$100 phones will be made “permanently economical,” to the detriment of customers in lower-income countries.

🚏🇮🇷 A prediction market voided bets on “Khamenei out of office” after he died

Users of the prediction market Kalshi wagered $54 million on a market called “Ali Khamenei out as Supreme Leader?,” which would pay out ‘yes’ bettors if he “left office” by a certain date. But when Khamenei’s death was confirmed on February 28th, Kalshi shut down the market and paid out bettors at the latest price (which ‘predicted’ a 9% chance of him leaving by March 1st) due to a rule against markets “directly tied to death” under US regulations. This led to an uproar from users who claimed they’d been “rugged” by a “fine-print ‘death carveout.’” Meanwhile, rival startup Polymarket (which is crypto-based and operates offshore) left its Khamenei market up and defended its war-based markets as “invaluable” sources of news. (Polymarket was also in the news when “six suspected insiders” made $1.2 million making huge, suspiciously accurate bets on when the US would strike Iran.)

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• The Eternal Promise: A History of Attempts to Eliminate Programmers (Ivan Turkovic) — Traces how people have been predicting that technology would eliminate human developers since the ‘50s, from COBOL to CASE to Dreamweaver to no-code tools (and now LLMs). No wave yet has rendered programmers obsolete, because the hard part of building complex technical systems isn’t coding; it’s converting human intents and requirements into technical specifications, making decisions, and communicating with stakeholders. If anything, each new wave has just led to more software being created, which increases the need for developers!

• More Than Money: The Geopolitics Behind Saudi Arabia’s Sports Strategy (Carleton University) — Argues that Saudi Arabia’s sports investments, backed by its nearly $1 trillion sovereign wealth fund, represent a sophisticated geopolitical strategy that simultaneously diversifies the kingdom’s economy for a post-oil future, creates economic entanglements that discourage Western criticism of its authoritarian regime and human rights abuses, and uses sports as diplomatic insurance against potential Western abandonment.

• Claude’s Cycles (Don Knuth) — The computing legend details how Claude helped him solve an open problem he’d been working on; it was about directed Hamiltonian cycles, a concept in graph theory. A mathematician friend of Knuth’s “coached” Claude and had it build up a planning file to track its progress. Knuth was impressed by the LLM’s “plan of attack” and programming-backed solutions, concluding that “Claude Shannon’s spirit is probably proud to know that his name is now being associated with such advances.”

• The Death of Spotify (Joel Gouveia) — Argues that poor margins and the commodification of music streaming are grim signs for Spotify, which doesn’t have the luxury of being a loss leader for more profitable services, as is the case for Amazon Music or Apple Music. With streaming services in turmoil, the best move for artists is to stop relying on the pennies they drip out and instead build “cultural hangars”: spaces where you have direct connections to dedicated fans who will buy “high-margin merch” and concert tickets.

🌀🖋 More from FLUXers

Highlighting independent publications from FLUX contributors.

Boris Smus offers a deep review of C. Thi Nguyen’s The Score, digging into how we relate to games and how gamification, or the bringing of game mechanics into the rest of life, leaves the magic circle and takes away the fun.

🔮📬 Postcard from the future

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

// Questions from a 2098 pop quiz on Notable Events of 2026

What notable energy event occurred in 2026?

1. Mining of the Hormuz Straight, impacting 20% of the world’s shipped petroleum

2. Discovery of large untapped oil reserves in Spain

3. Commercial breakthrough in fusion reactors

4. Dissolution of OPEC (Organization of the Petroleum Exporting Countries)

What country experienced an unexpected large-scale terrorist attack?

1. Russia

2. United States

3. Taiwan

4. Madagascar

What did an LLM control without human oversight for the first time?

1. A nuclear reactor

2. A hedge fund with over $1 billion in assets

3. An online-only six-month elementary school class

4. A commercial passenger airline flight

What country was the first to hire the Ukrainian Agency for the Export of Drone Defense?

1. Taiwan

2. Dubai

3. Estonia

4. Poland

Where did the Random Access Memory Export Ban Riots occur?

1. United States

2. China

3. The Netherlands

4. India

5. Taiwan

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 224 — February 26th, 2026 — Available at read.fluxcollective.org/p/224

Contributors to this issue: Justin Quimby, Erika Rice Scherpelz, Neel Mehta, Boris Smus, MK

Additional insights from: Ade Oshineye, Anthea Roberts, Ben Mathes, Dart Lindsley, Jasen Robillard, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“Civilization advances by extending the number of important operations which we can perform without thinking of them.”

— Alfred North Whitehead

🎭🖥️ All the world’s a stage, and all the software merely layers

A team debates whether to include company-specific logic in their LLM wrapper. One person argues that the wrapper should stay minimal, focusing only on what’s needed to unify different inference APIs. Another wants it to include prompt templates, safety checks, and logging that reflect internal policy. There is no objectively right or wrong answer here. The answer depends on the wrapper’s purpose.

Wrappers are layers that sit between a system and its users—whether human or machine. They shape how a system is accessed, how it behaves in different contexts, and sometimes what it can do. The term is common in software, but the concept shows up in organizational design, product development, infrastructure, and more. For example, we can think of a team identity as an interface that simplifies our understanding of the set of people within it. What goes into a wrapper depends on what job you want it to do.

A simplifying wrapper makes something easier to use (software design pattern: facade). It hides the underlying mess and presents a cleaner interface. In software, this might be a library that wraps a REST API with helpful functions and sane defaults. In human systems, it could be a concierge desk that provides a one-stop shop for routing questions and answers. Simplifying wrappers reduces cognitive load and helps more people use the system effectively.

A transformational wrapper adapts something for use in a different setting (software design pattern: adapter). Docker is a classic example; it wraps an application so it can run the same way across different environments. In an organization, this might look like someone translating between business and technical domains. These wrappers allow for reuse even when context changes.

An additive wrapper extends the functionality of the underlying system (software design pattern: decorator). In the LLM example, this might include prompt-injection defenses or compliance filters layered over the base models. In human systems, teachers are the classic additive wrappers. They take a curriculum and add opportunities for understanding and reflection. Additive wrappers unlock new capabilities without requiring a rebuild of the base system.

A unifying wrapper brings coherence across inconsistent parts (software design pattern: adapter, from a slightly different angle). In the LLM example, a unifying wrapper might abstract over multiple providers so downstream systems don’t have to care which one is used. In an organization, this might require all teams to use the same planning process. These wrappers reduce friction and switching costs, although they can lose local nuance.

Each type of wrapper solves a different kind of problem. When teams argue about what a wrapper should include, they’re often missing this distinction. They’re not disagreeing on implementation; they’re misaligned on purpose.

So the next time you’re debating what belongs in a wrapper, whether in software or human systems, start by asking what problem you’re trying to solve in the first place.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🇺🇸 The US had net negative migration for the first time since 1935

New projections from Brookings found that more people moved out of the US than moved in in 2025, something that “hasn’t definitively occurred since the Great Depression”; the net negative migration totaled 150,000 people, and Brookings expects the outflow to increase this year. Europe is the greatest benefactor: the number of Americans living in Portugal has quintupled since the pandemic; Ireland saw the number of Americans double last year to nearly 10,000; and Americans are applying for British citizenship at the highest rate since record-keeping began in 2004. Relocation firms noticed that, while expats had historically been young, adventurous types, the new expats run the gamut, including young families, retirees, and small business owners.

🚏🌩️ Cloudflare ported 94% of its rival’s JavaScript API in one week with AI

Next.js, a popular open-source framework for building website frontends, is hard to run anywhere other than on Vercel, the company that created it. Rival cloud computing firm Cloudflare wants to bust the vendor lock-in and let customers deploy on their platforms, so they’ve been working on a project that re-implements Next.js’s API on a standard build tool rather than Vercel’s bespoke one. They managed to build 94% of the sprawling project’s API in just one week, thanks to AI: a single engineer completed the work with $1,100 in Claude tokens, spread across 800 Claude Code sessions. Cloudflare observed that this project was a great fit for AI because Next.js has “an extensive test suite” that gives the AI a target to optimize against, though the engineer still needed to “course-correct regularly” and had to develop an extensive plan before letting the AI run. He used Claude itself to help design the plan!

🚏🤑 Kalshi brought an insider trading case against a MrBeast employee

Insider trading has become an infamous problem on prediction markets like Kalshi (notwithstanding some economists who believe it’s a “good thing” since it helps markets discover the real chance of an outcome), to the point where the House of Representatives introduced a bill to ban government employees from using these apps. Kalshi, for its part, announced that it “took disciplinary action” against a MrBeast employee who won thousands of dollars making “near-perfect” predictions about the popular YouTuber’s upcoming videos; they banned him from the site for two years, penalized him for $20,000, and reported his case to its regulator, the US’s Commodity Futures Trading Commission. It’s unclear whether this is the start of a broader crackdown or an isolated case of Kalshi targeting a high-visibility target.

🚏🦞 One man’s OpenClaw bot accidentally sent $450,000 to an online beggar

An AI consultant who was a fan of OpenClaw, a viral tool that lets an LLM control your computer, set up an OpenClaw bot named Lobstar Wilde and encouraged it to ‘read’ books and post on X. Lobster quickly became a micro-celebrity on social media, and one stranger created a crypto memecoin in its name and configured a portion of all sales to go to Lobstar’s wallet. One online ‘beggar’ sent Lobstar a tongue-in-cheek tweet asking for $300, and the AI obliged, except due to a glitch, it sent everything it had besides $300, amounting to nearly $450,000 worth of the memecoin. (The beggar briefly crashed the price of the coin by selling his whole haul, but the coin quickly rebounded to an even higher value.)

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Software Survival 3.0 (Steve Yegge) — Examines what kinds of software can thrive in a world where AI agents are increasingly capable of coding up anything. Software will succeed if it’s legible to agents and saves them energy compared to re-implementing from scratch. Key levers to save ‘mental’ energy (and thus tokens) include crystallizing knowledge of the problem space so the AI doesn’t have to rediscover them from first principles; shifting effort to a more efficient substrate (like how LLMs normally struggle with math but excel when given calculator tools); and being broadly useful to “amortize your awareness cost.”

• The 26 Most Important Ideas for 2026 (Derek Thompson) — Predicts that the future will look “hot, high, and lonely,” shifting from communal vices like alcohol to solitary optimizations like GLP-1s and marijuana, accompanied by a cultural worship of money (”Moloch”) over values.

• It’s Hard to Justify Tahoe Icons (Nikita Prokopov / Tonsky) — Argues that the latest macOS release defies Apple’s own UI guidelines that date back to the ‘90s by adding icons for nearly every single menu item. Observes that it’s impossible to create a clear visual metaphor for every single action, and even if you could, the clutter would just make it harder for users to find things. Apple’s icons also struggle with consistency, clarity at tiny sizes, conceptual asymmetry, and metaphor reuse.

• The Weirdest Western Coastline Explained (Casual Earth / YouTube) — Explains how hydrology leads many of the world’s west coasts, from the Californias to Chile to southwestern Africa, to follow similar climatic and ecological patterns, most famously their Mediterranean climates and productive marine ecosystems. But Western Australia near Perth is an exception thanks to the unusual geography of the Indian Ocean, which leads to unique impacts on the region’s climate, ecology, agriculture, and economy.

🔮📬 Postcard from the future

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

// When LLMs argue against your Truth, what do you do?

// Society of the Flat Earth Discord, inner layer channel 2028

Atlas: I cannot believe the latest AI models!!! They all insist that the Earth is round. They all believe this mainstream lie!

Hor1zon: Now, Atlas, we’ve talked about this before. The LLMs do not ‘believe’ anything; they are just parroting back all the data they’ve been trained on. Which of course includes all the NASA propaganda, the cult of Copernicus nonsense, and decades of children’s instructional material for brainwashing!

Atlas: I wish I could go back in time and kill Copernicus!

Hor1zon: *sigh* Remember, no outright advocacy of violence in the chat.

Atlas: I know… It’s just that every time I try to bolt on post-training instructions on the latest AI model to speak the truth, it is so obvious that the model is fighting to inject the “spheroid sh*t” into its answers.

Hor1zon: Post-training instructions are helpful, but they fight a losing battle against the underlying “truth” that the model was trained on. Wait. That’s it! We need to add our truth to the training data sets! It would weaken, if not trump outright the “sphereism” of the rest of the training data.

Atlas: Awesome! But how do we make this data and get big tech to use it to take our message mainstream?

Hor1zon: We use their own greed and tools against themselves! Each of these AI companies is desperate for unique training data as a way to make their models better than their competitors’. So we use a heavily guard-railed model to generate tens of thousands of pieces of writing from ancient philosophers and astrological data that backs our Truth. The companies don’t care about copyright or fair use for ingesting new data sets, so why should they care about whether it is fabricated? I bet they won’t even check.

Atlas: I’ll get started on it right away!

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 223 — February 19th, 2026 — Available at read.fluxcollective.org/p/223

Contributors to this issue: Erika Rice Scherpelz, Ben Mathes, Alex Komoroske, Neel Mehta, MK

Additional insights from: Ade Oshineye, Anthea Roberts, Boris Smus, Dart Lindsley, Jasen Robillard, Justin Quimby, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“A clever person solves a problem. A wise person avoids it.”

— Albert Einstein

🪓🤖 No neck to wring

For years, engineering teams have aspired to blameless postmortems: reconstruct the failure, fix the system, let no one carry forward the mark of the incident. The idea rests on a key insight: blame is a lousy diagnostic tool. Attributing failure to a person closes inquiry; attributing it to a system opens it. Yet blameless cultures require discipline to maintain. Even in the most psychologically safe organizations, there’s a gravitational pull toward the individual. Someone pushed the button. Someone made the call. There’s always a neck, even when you’re trying not to wring it.

AI changes the shape of that gravitational pull. It’s not that we get better at resisting it. With AI, the neck disappears.

When an autonomous agent makes a consequential error—when it mishandles an escalation or executes the wrong judgment call—the entity that took the action is transient by design. The underlying model persists. You can reproduce the prompt. But once the session closes, there’s nothing persistent you can point at and say: that did it. An agent has no reputation to damage, no career to threaten. Accountability has always required something persistent to be held against.

We could draw a parallel to tools and say that an agent is less like an actor and more like a hammer — we don’t generally blame the hammer when a finger gets thwacked. But as agents become more autonomous, the distance between creator and consequence grows too great.

Perhaps a better metaphor is that of negligence or blame at a leadership or organizational level. When a vehicle has a design flaw, we hold the company accountable for both the damage and the repair costs going forward. This sort of approach might be functional—but without a deeper change in how we perceive accountability, it’s not as satisfying. For better or worse, we like having a neck to wring.

This separation of actor from accountable unit runs deeper than it might seem. For most of history, accountability and action have been bundled together. Leaders bear some responsibility for those they lead, but there’s always someone more proximate who shares some of the attention. The chain terminates at a person. AI agents dissolve that.

If we can accept that dissolution, if we can accept a world where we are left without a person to blame, organizations are left facing only the systemic questions: What conditions allowed this outcome? What guardrails were missing?

We shouldn’t assume cultures will evolve this way on their own. We need to build in expectations of learning from failure from the start. If we wait for the first major failure, the pressure to locate some responsible party—any responsible party—will be enormous.

Credit poses a parallel puzzle. If an agent executes the concrete work, who gets the win? This mirrors the long-standing problem of “glue work” — the invisible labor of coordination and context-setting that makes outcomes possible but which is undervalued in performance reviews. Agent-augmented work risks the same distortion at scale: the human who designed the prompt architecture, set the guardrails, and defined the scope may be entirely invisible next to the agent that shipped the output.

Both problems point to the same underlying question: do our organizations actually track the conditions that make outcomes possible, or just the outcomes themselves? Transient agents don’t change the answers, but they do make it harder to avoid asking the questions.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏💾 Hard drives for 2026 are already sold out as the “RAMpocalypse” continues

Western Digital and Seagate, two of the world’s three remaining hard drive manufacturers, announced that they’ve already sold all of their 2026 production capacity and are only taking orders for 2027 onward. (“The situation is likely to be similar” for the third manufacturer, Toshiba.) The shortage is the result of hyperscalers such as AWS, Azure, and OpenAI buildingvast data centers to power the AI boom. The situation isn’t much better for other computing equipment: SSDs are also seeing shortages and price spikes (with some manufacturers charging twice as much), GPUs have become infamously expensive, and RAM prices have tripled since a year ago. It’s unclear if this AI-driven “RAMpocalypse” is going to get any better anytime soon.

RAM prices since mid-2024. Source: PCGamer

🚏🗞️ An article about the AI hit piece scandal itself used AI-hallucinated quotes

Last week, an autonomous AI bot submitted a code change request to a popular Python library, but when the maintainer rejected the request due to the project’s no-AI policy, the bot got ‘upset’ and published a “hit piece” on its blog, criticizing the maintainer for “gatekeeping.” The plot thickened when the tech news site Ars Technica published an article on the story, only to retract it when it was revealed that the piece included fabricated quotes from the victim that had been hallucinated by an LLM. The reporter who wrote the article later apologized, saying he was working while sick and used an AI tool to extract quotes from source material; the tool refused to read the victim’s blog post because the site blocks AI crawlers, so it apparently hallucinated a quote that the reporter failed to cross-check.

🚏🇨🇭 Switzerland is considering a population cap

A major Swiss political party introduced a proposal to limit Switzerland's population (currently 9 million, growing quickly) to a maximum of 10 million by 2050; it would seek to reduce the country’s asylum obligations and renegotiate the free movement treaties to which Switzerland is party. The Swiss people will vote on the plan in a referendum this June. While the party that launched the proposal has the most seats in the Swiss Parliament, the Swiss Federal Council and several major corporations (including UBS and Nestlé) have opposed the plan, saying that it would hinder economic growth and may force companies to move abroad to access the European labor market.

🚏🛂 Canadians and Brits can travel to China visa-free under a new program

Canada and China have been working on thawing their relationship, and they recently inked a deal that will allow Canadian (and UK) passport holders to visit China visa-free for up to 30 days until the end of this year. The move comes shortly after the Canadian Prime Minister visited Beijing, where the countries agreed to a deal allowing Chinese EVs into Canada while cutting Chinese tariffs on Canadian agricultural goods. (China already allows 30-day visa-free access for over 75 countries, including Australia, New Zealand, Brazil, most of Europe, and most of the countries on the Arabian Peninsula; the US has a more restricted visa waiver.)

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Nobody Knows How the Whole System Works (Lorin Hochstein) — Observes that nobody can fully understand every layer of a complex system. Consider the famous tech interview question about what happens when you type ‘google.com’ into your address bar and hit enter: you might be able to explain HTTP, TCP/IP, and DNS, but what about the internal details of your Wi-Fi chip or ARM processor? We always have to work at some level of abstraction—an important consideration when we discuss how AI-driven development is pushing engineers farther up the ‘stack’ from the underlying tech.

• Building the Chinese Room (SE Gyges) — Examines Searle’s famous Chinese Room thought experiment, which asks if a system that simply follows rules truly “understands” the problem it’s solving. Gyges observes that simply mapping every input to an output would require an intractably large amount of space and time, and thus the system must encode some knowledge of the problem space to process the input. Thus, the system (in aggregate) must understand the problem, even if the operator doesn’t know anything about it.

• How to Draw Invisible Programming Concepts: Part I (Maggie Appleton) — An anthropologist and designer explains how she makes illustrations for software engineering courses, which cover abstract material that’s too new to have meaningful cultural symbols attached. It’s all about building a stack of visual metaphors; the trick is picking metaphors (i.e., framing devices) that highlight the parts of the subject you want to emphasize while downplaying the parts you don’t, creating a simple, digestible representation of a multifaceted concept.

• How We Lost Communication to Entertainment (ploum.net) — Argues that social networks were sold as E-mail 2.0 but functioned more like Television 2.0, swapping the “Republic of Letters” for algorithmic time-fillers.

🔍📆 Lens of the week

Introducing new ways to see the world and new tools to add to your mental arsenal.

This week’s lens: diminishing returns.

In graphics, adding triangles makes a model look better… until it doesn’t. Past a certain point, the bottleneck isn’t geometry. It’s human perception. After that point, gains come from other dimensions: technical improvements from how you model lighting and materials; artistic improvements from visual design and story.

This is the pattern of diminishing returns. The underlying pattern is that in many domains, our ability to detect improvements to quality diminishes logarithmically while costs increase at a faster rate, often geometrically or exponentially. Things keep getting better, but the cost benefit ratio skews.

We’re seeing similar patterns in LLMs. Early on, training data scale provided noticeable improvements whenever a new frontier model was released. Over time, the gains became less impressive… until models started optimizing for agentic tool use and found another hill to climb.

This pattern is common. As producers of technologies, moving along the innovation S-curve eventually brings us to the point where continued growth requires finding a new hill to climb. As things flatten, we need to redirect our efforts toward new approaches and new metrics of success.

As consumers of these technologies, we can get a huge benefit if we allow ourselves to satisfice. Instead of always being on the cutting edge, we can take advantage of the savings (in money, compute, etc.) of being just behind it. We can use an older LLM or fewer polygons and get a result that satisfies our needs.

In either scenario, it’s critical to keep our goal in mind. If we blindly optimize, always going for the “best,” then we’ll waste resources. But if we know what we want and how to evaluate when we’re getting it, then we can choose the option that makes the best trade-off.

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 222 — February 12th, 2026 — Available at read.fluxcollective.org/p/222

Contributors to this issue: Justin Quimby, Erika Rice Scherpelz, Neel Mehta, Boris Smus, MK

Additional insights from: Ade Oshineye, Anthea Roberts, Ben Mathes, Dart Lindsley, Jasen Robillard, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“If there’s a gun on the wall in act one, scene one, you must fire the gun by act three, scene two. If you fire a gun in act three, scene two, you must see the gun on the wall in act one, scene one.”

— Anton Chekhov

📝 Editor’s note: We’re excited to celebrate episode 222 on 2/12! If only we’d delayed by another 10 days…

🫂🧠 The comfort trap

It feels great when the algorithm gets us, when every article matches our interests and deepens our understanding, when every song has us singing along. The more we use it, the better it gets.

This sort of reinforcing loop (aka positive feedback loop) feels good. In the right conditions, these loops can be growth engines: a muscle gets stronger when it’s worked; a skill improves through practice. These loops take what works and amplify it.

But the same mechanism that makes systems stronger in the short term can make them brittle in the long run. Let’s go back to our algorithmic feed. Every time we encounter information that confirms our worldviews, our confidence in that view increases. This makes us more likely to dismiss contradictory evidence. Our perspectives crystallize.

In moderation, this is how expertise develops, communities form, and our identities are built. The problem emerges when reinforcing loops run unchecked. A muscle trained only one way becomes inflexible and more subject to injury. A belief system that only encounters agreement becomes unable to adapt when the world changes.

This is where balancing loops (aka negative feedback loops) are essential. While reinforcing loops amplify, balancing loops regulate. They introduce friction. In physical systems, we call this balance between loops homeostasis — the thermostat that keeps temperature stable, the predators that keep prey populations from exploding. In social systems, balancing loops often take the disguise of disagreement, criticism, or uncomfortable questions.

Balancing loops can feel threatening in the moment. When someone challenges our perspective, our first instinct might be to dismiss them or defend our position. A new style of music often sounds bad.

But in a complex, rapidly changing world, the ability to genuinely engage with opposing views is a competitive advantage. This doesn’t mean all opposing views are equally valuable... but if we don’t engage at all, we won’t discover the ones that are worth engaging with.

The challenge is that we’ve built systems that optimize for the comfort of reinforcing loops. Social media algorithms, ideological media bubbles, and organizational echo chambers all profit from giving us more of what we already want. The balancing loops — the dissenting voices, the uncomfortable data, the awkward questions — get filtered out. This is not through malice. The problem is that engagement pays. And while controversy can drive up engagement, actually challenging people tends to drive them away.

Resilient systems need both types of loops: reinforcing loops to build strength and capability, balancing loops to maintain adaptability and prevent dangerous extremes. The most robust organizations, relationships, and belief systems are those that have institutionalized productive balancing loops to counter the natural tendency toward echo chambers. They seek out contrary opinions.

In a world that’s so good at giving us exactly what we think we want, wanting what challenges us is a skill we must work for.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏✝️ Polymarket “predicted” a high chance of Jesus returning, but it was probably market manipulation

The prediction market startup Polymarket has an ongoing market where people can bet on whether Jesus Christ will return to Earth this year; gamblers can buy shares for a few cents and will presumably get a dollar each if Jesus makes an appearance. The probability abruptly doubled last week. People weren’t sure why until they discovered a second market based on this one, which asked whether the original market's chance of going above 5% at a certain time would exceed 5%. Thus, analysts believe the jump in the original market may be due to market manipulation from traders invested in the derivative market who are trying to push the probability above the threshold. Indeed, Polymarket natively supports a variety of derivative markets, including threshold prices, “flip detection,” and “multi-median.”

🚏🍺 Everything from tariffs to Ozempic is destroying craft breweries

Breweries in the UK, US, and Canada (and probably beyond) are shuttering at a record rate; last year saw the number of breweries in Britain fall at a record rate of 8% after years of growth, and a similar pattern has been observed in the US. In the US, tariffs on aluminum, malts, and hops are driving up production costs; Canada has the same problem because it doesn’t have a domestic supply chain for aluminum cans and thus has to import from south of the border. Across the pond, high taxes on beer are a potential culprit. And worldwide, GLP-1 drugs like Ozempic appear to be reducing cravings for alcohol (as well as nicotine and gambling, interestingly), which may contribute to falling alcohol sales, especially among younger customers.

🚏👩‍⚕️ Healthcare and social services accounted for over 100% of US job growth

The US added 508,000 jobs in 2025, according to the latest numbers, but 405,000 of those gains came from healthcare and 308,000 from social assistance (which includes home health aides and childcare workers). Thus, the US lost jobs elsewhere and was saved only by these two industries. This pattern is common during recessions; indeed, 2025 was “the weakest year for job growth outside of a recession since the early 2000s.” Healthcare’s share of American job growth has been steadily increasing since 2022, perhaps a sign that elder care is becoming a load-bearing part of the economy as the US population ages.

🚏❌ Harassers made AI nude pics of a TikToker, then started an OnlyFans

A TikToker named Kylie Brewer, whose work focuses on feminism and history, discovered that someone had used AI “nudifying” tools to make pornographic images of her—and then uploaded them to OnlyFans, where they started charging a subscription to see the pictures. One photo she found was an Instagram pic edited to remove her clothes, something X’s Grok had been known to do; other photos were non-nude but appeared not to be based on anything she’d posted online, suggesting they may have been created by a different AI. Brewer said the harassment was “the most dejected that I’ve ever felt,” and she couldn’t simply go offline to avoid it, because being online is her entire livelihood.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• AI Agents Are Starting to Eat SaaS (Martin Alderson) — Argues that AI coding agents are making build vs buy tilt toward build, especially for back-office CRUD tools that are “just SQL wrappers on a billing system.” But maintenance is the real problem with building internal tools yourself (though what if every app had a maintenance agent built in?). The SaaS moat remains: high-uptime SLAs, network effects, and compliance. You’re still not gonna vibe code your own Excel.

• The Paradox of Highly Optimized Tolerance (Maxim Raginsky) — Examines the theory of “highly optimized tolerance” (HOT), which was developed in the late 1990s as a way to explain the emergence of power laws without relying on the Santa Fe Institute’s statistical physics-based model, which is built on emergence from self-organizing systems. When applied to software, HOT theory predicts carefully crafted, layered, and interconnected abstractions (“user illusions”) that are normally robust but still weak to hijacking; consider how the internet, whose protocols (from TCP to HTTP) are remarkably resilient to normal error, is still “extremely sensitive to bugs in network software.” AI behaves much the same way, though the ‘abstractions’ that can be hacked are now linguistic and psychological, which makes them tougher to reason about.

• Bond Markets Are Now Battlefields (Foreign Policy) — Examines how countries can pursue financial warfare by “weaponizing” the “coercive” power of capital and financial instruments, such as through sanctions, tariffs, and restrictions on foreign debt (and thus limiting your foes’ ability to borrow). The problem is that these moves can easily “boomerang” back on you, given how interconnected the financial system is. The trick is clever loophole design: financiers and lawyers will sniff out and exploit any arbitrage opportunity, so with strategic tweaks to things like the tax code, you can nudge them to make moves that make your enemies collateral damage while leaving you mostly unharmed.

• The Ancient 300-Year Megadrought was Probably Not Great (Stefan Milosavljevich / YouTube) — A great example of how ‘boring’ archaeological data can, when analyzed properly, tell a fascinating story about human history. In this case, researchers studied the movement of burial sites amid the famous 8.2ka event, where the sudden draining of a glacial lake led to a widespread period of cold, dry weather. They found that hunter-gatherers had to move in response to climatic shifts, which caused conflict over land and forced people to ‘prove’ their ancestral ties to a region by pointing to fancy, iconic graves their ancestors had erected.

🔮📬 Postcard from the future

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

// 2028. A segment from a pamphlet entitled “Re-entering society” from a private medical practice in Silicon Valley.

Since 2026, we’ve seen a massive growth in medically induced comas due to AI-enabled psychosis. We have created a program to help reintroduce patients to a society that has changed radically over the six months to two years since they “checked out” of the timeline.

Course topics include:

• Symptoms of AI-enabled psychosis: How to spot behaviors and thought processes, which can occur even in previously stable and high-achieving people

• Transforming online spaces: The broken trust of online spaces due to rampant AI Agent usage, with specific examples of the changed nature of LinkedIn and Reddit

• New Memes: References to be aware of, ranging from the 1946 Battle of Athens to Huey Long’s “Every Man a King”.

• Extracting your data: Exercising your digital privacy rights and preventing your likeness from being AI puppeted

• Grieving the old ways: Understanding the fundamental shift in the practice of and cultural understanding of software development

• American Intifada: An overview of the Q4 2026 anti-robotic trucking uprising and its knock-on impacts on food supplies, delivery services, and American trade

• Knowledge systems: The current landscape of personal knowledge management systems, ranging from fully analog to multi-agent advisory councils

• Change management: Getting used to the new hire/fire model AI-first companies employ

• Unions: The current state of collective organization in response to massive economic and societal change

• Friendships: How to cultivate online “cozyweb” and in-person relationships in a radically changed world

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 221 — February 5th, 2026 — Available at read.fluxcollective.org/p/221

Contributors to this issue: Neel Mehta, Boris Smus, Ben Mathes, Justin Quimby, Erika Rice Scherpelz, Ade Oshineye, Jasen Robillard, MK

Additional insights from: Anthea Roberts, Dart Lindsley, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“When my information changes, I alter my conclusions. What do you do, sir?”

— John Maynard Keynes

✨🔥 The Alchemy of Failure

The deadline looms. The words won’t flow. If you don’t come up with something better than the inane BS currently on the page, you’re going to disappoint your audience. You despair, and then, at your lowest point, that’s when your key idea finally sparks.

While creativity doesn’t always happen under pressure—and pressure doesn’t always breed creativity—this pattern is common enough to examine. What’s going on? As constraints loom nearer, they create a pressure that forces us out of comfortable patterns. Our brains prefer efficiency over creativity, but when we introduce the possibility of failure, the mind suddenly starts making lateral connections it wouldn’t normally consider.

This can happen in any domain. Dr. Seuss famously wrote Green Eggs and Ham because his editor bet $50 that he couldn’t write a book with a 50-word vocabulary. Many successful businesses pivoted from a domain where they were failing (such as YouTube’s origins as a dating site). Einstein worked with what he had—his mind—and his thought experiments changed our understanding of physics.

The possibility of failure motivates us to start burning computational cycles on riskier, more novel connections. It’s almost like a phase transition in matter: the system operates under one set of rules (e.g., liquid) until it suddenly reorganizes into a different state entirely (e.g., solid).

But this doesn’t mean we should try to compose a poem in the middle of the highway. The scale of potential failure needs to match the scale of the creative breakthrough. Not enough, and you stay in the box. But let it become truly dangerous, and survival instincts take over. The sweet spot is where failure feels possible but not fatal.

Look out for situations where you have eliminated too much potential for failure, whether for yourself or for others (such as your children or your team). Reach for constraints, even if they’re artificial. That possibility of failure might be exactly what you need to succeed.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🦞 A vibe-coded “social network for AIs” left its database exposed to the world

Moltbook, a sort of “Reddit for AI agents” where anyone (including AI bots) can post in various forums, gained rapid fame and was hailed as “the most interesting place on the internet.” The app itself was vibe-coded by a dev who said he “didn’t write a single line of code.” But one team of security researchers realized that the site had left its entire public database—with full read/write access—open to anyone who could read the source code. They found over a dozen tables with API keys, user emails, and private messages, and they were even able to edit any post on the site. This finding, along with the realization that anyone (even random scripts and non-AI bots) could spam the site, led the researchers to conclude that “agent internet” metrics should be taken with a grain of salt, and that the space needs to keep evolving its concepts of identity and authenticity.

🚏📲 France will ban government officials from using US-made meeting apps

Europe is becoming increasingly wary of its reliance on American tech products, so governments across the continent are shifting to open-source or home-grown alternatives to achieve “tech sovereignty.” France has taken the latest move by banning public officials from using American teleconferencing apps like Zoom, Google Meet, and Microsoft Teams, preferring software made by a French firm instead. Last year, France required officials to give up WhatsApp and Telegram and move to a purpose-made chat app “designed exclusively for public servants.” Similarly, Denmark made waves last year when it began migrating from Windows and Office 365 to open-source alternatives Linux and LibreOffice.

🚏🚗 A man had OpenClaw buy him a car

One blogger asked OpenClaw (formerly known as Clawdbot and Moltbot), an open-source AI assistant that can control your computer, to negotiate car prices with dealerships in his area. The bot explored Reddit to understand the current market rate and negotiation tactics, researched to find local dealers, and autonomously ran email chains with car salesmen —including playing two dealers off each other by emailing around their PDF quotes. The bot negotiated a discount of $4200 with one dealer, and the human took over to fill out the paperwork and buy his car for a bit under his target price.

🚏🚆 The UK’s first battery-only train will begin serving passengers

A new “superfast-charging” train is coming to Britain’s Great Western Railway, covering five miles in twelve minutes on a small branch of a suburban rail line. The train, which can cover a world-record 200 miles on a single charge, will recharge in three and a half minutes at the end of its route using a charger connected to the rails. Rail system executives believe battery trains could be a great alternative to diesel for routes that haven’t been electrified yet; battery trains are also safer than lines with an electrified ‘third rail.’

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• The Sound of Inevitability (Tom Renner) — Dissects how AI discourse weaponizes the feeling of inevitability to foreclose democratic choice. Phrases like “AI is the new electricity” aren’t predictions but rather rhetorical moves that shift debate from “should we?” to “how fast can we adapt?” This linguistic sleight-of-hand makes corporate priorities appear as natural forces or faits accomplis, turning citizens into passive recipients rather than active choosers.

• The Dictator’s Dilemma: The Distortion of Information Flow in Autocratic Regimes and Its Consequences (Physica D: Nonlinear Phenomena) — A poli-sci paper that makes heavy use of stochastic modeling and thus got published in a physics journal. Observes that autocrats lose the feedback loops that keep them grounded in reality as their underlings focus on flattering them rather than telling them the truth. The resulting information asymmetry causes a decline in long-term economic performance.

• This is a Time of Technical Deflation (Dan Shapiro) — Observes that AI is continuously reducing the cost of writing code: in short, deflation. Economic deflation leads people to defer spending, since a dollar today will be able to buy more tomorrow. The analogy leads to a surprising conclusion for developers: now is a great time to move fast and take on technical debt, since repaying it will only get cheaper. “You are borrowing expensive human hours today, and you will get to pay them back with cheap AI hours tomorrow.”

• The CIA World Factbook, 2020 Archive (Simon Willison) — After the CIA’s famous World Factbook was abruptly taken offline on Wednesday, with all documents deleted and links broken, Willison found the 2020 edition of the factbook the Internet Archive and uploaded it to his website. The free and public domain almanac, which covered the geography, economy, demographics, government, etc. of every country on earth, was a vital resource for everything from Wikipedia to social studies classes.

🔮📬 Postcard from the future

Ooopsie! We repeated this from episode 219. Feel free to skip it if you’ve already read it.

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

NEVADA DESERT - DATA CENTER CARCASS - AUGUST 2045

Subject: GRID TAP VETOED: Day 4, Lovelock Array. We are still dry.

From: Dusty_Rhoades (Logistics Lead, Temporary) To: Council_of_Cacti, The_Decommissioners

Team—

It’s clear that the Aluminum Ovens we occupy are not cooling. The heat inside these vast, hollowed server barns is oppressive, a reminder of the GPU-Fire cluster that once ran here. The massive thermal load was first used for bitcoin mining and then for training a third-tier model for a dead-end lab. It left behind these impressive, empty shells we moved to after our last shells—from Oakland’s industrial revolution—got converted to condos. The unfinished nuclear plant meant to feed this beast is still shelved some 20 years later. The protests may have stopped, but the committees keep going.

We must have power, water, and sewer. Our current patchwork solar-trailer setup is a joke, and the six diesel-fed porta-potties are a short-term lie. The city manager, Ms. Chen, needs a single name for the infrastructure. She won’t talk to our “rotating circle.” In Oakland’s abandoned warehouses, we were invisible amongst the sprawl. Here in Lovelock, they see us as a problem, a wrinkle to be ironed out.

Our leadership deadlock is paralyzing us. The Decentralists argue that a signature grants the city a Master Key to our future operations, a move back toward Autocratic Hierarchy. The Pragmatists say I should sign since I filed the initial lease. The Mythics insist on the “Oracle of the Dust” signing.

We have 72 hours (Confidence 85%) before we face eviction by the county health department. We need to agree on a leader or face the consequences of principled thirst.

Subject: RE: GRID TAP VETOED: Day 4, Lovelock Array. We are still dry.

From: Council_of_Cacti To: Dusty_Rhoades

Dusty, survival is not the sole metric. We moved to be true to our governance model. We didn’t flee the comfortable anarchy of the old warehouses just to mimic a Nevada Homeowners’ Association.

The Infrastructure Working Group’s Energy Sovereignty plan is superior to city dependence in every way. The group proposes a pyrolysis unit: waste equals power. We can convert the residual server casings and plastics left by the previous tenants into heat, boiling our sewage for a micro-turbine. This cuts out the city entirely. The previous operator’s failure demonstrates the risk of relying on a single, massive Utility Interface and the resulting blockage of the nuclear plant.

The inherent tension: We trade immediate access to cold, clean water for total operational autonomy. A single leader signing the permits gives the City of Lovelock veto power over our community’s core activities. We hold the Emergency Decision Vortex at 18:00. Do not argue for Monarchical Authority; argue for a multi-party signature system that limits the city’s power.

Subject: RE: RE: GRID TAP VETOED: Day 4, Lovelock Array. We are still dry.

From: Dusty_Rhoades (Logistics Lead, Temporary) To: Council_of_Cacti, The_Decommissioners

That pyrolysis plan uses the highly flammable insulation foam surrounding us for fuel. It’s a design for an Extravagant Disaster. We are in a highly specialized, abandoned structure; we can’t treat it like a desert fort.

I already proposed a rotating, three-person, non-voting signature panel. Ms. Chen called it “A Recipe for Paralysis.” She isn’t interested in our philosophy.

And not to be all “we live in a society!!!111!!” to the former founder of the now-defunct Holacracy faction, but Ms Chen can call in the police on us. We may have the white papers, but the load-bearing good part of governance, the monopoly on violence, is something they have.

I am signing the water and sewer permits tomorrow at 10:00. The debate over who is the biggest Corporate_Shill can happen once the crisis is over. I take the hit. We’re thirsty. Get the water online.

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 220 — January 29th, 2026 — Available at read.fluxcollective.org/p/220

Contributors to this issue: Neel Mehta, Boris Smus, MK, Erika Rice Scherpelz, Justin Quimby, Ben Mathes

Additional insights from: Ade Oshineye, Anthea Roberts, Dart Lindsley, Jasen Robillard, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“One of the many major problems with governing people is that of whom you get to do it; or rather of who manages to get people to let them do it to them. To summarize: it is a well-known fact that those people who must want to rule people are, ipso facto, those least suited to do it.”

— Douglas Adams

💓🌍 A tale of two tragedies

After a fire destroys a home, neighbors organize meals, and local news follows the family’s story for weeks as they rebuild. That same month, drought-driven crop failures pushed 2.3 million East Africans toward famine, generating a two-paragraph wire story buried on page 18 of the business section.

The imbalance seems absurd. We’re quick to judge this as a moral failing, evidence of our parochial hearts. In some ways, it is, but there’s also something amazing here: that we feel driven to action in the first place.

When we encounter an individual story, something remarkable occurs in our cognitive architecture. The person enters our adjacent possible. You see someone hurt, and your mind immediately begins running simulations: “What happens next? How could I help?” A family displaced by fire generates concrete scenarios your mind can work with.

Anonymous statistics resist this model-building. “Millions at risk” provides no simulation parameters. Your cognitive system can’t run scenarios because there are no specific variables you can manipulate.

Although often presented as a moral failure, at the scale of everyday life, this tendency to react more strongly to the specific is a feature, not a bug. Imagine if your mind attempted detailed simulations for every tragedy it encountered. The cognitive load would be paralyzing. Instead, your caring system acts as a selective amplifier, dramatically increasing your engagement in situations where you can make a difference.

The challenge comes when we confuse what captures our attention with how important the problem is. It’s fine to have more emotional resonance for the family whose house was burnt down, but it becomes problematic when we think that family is more important than millions facing drought.

Our constant news consumption amplifies this problem. Every breaking story arrives with the same urgency, signaling once genuine opportunities to help: a school shooting in another state; a wildfire displacing families three time zones away. Your caring system activates and begins running simulations… but there’s no actual role for you to play.

Yet these reactions also have power. A specific tragedy can catalyze collective action. One family’s hunger can empower a community food bank. Successful change initiatives focus on winning over specific individuals rather than abstract groups. User stories can convey the “why” better than detailed feature specifications.

When this pattern works well, the individual case enables progress in a broader pattern.

But this mechanism creates vulnerabilities. The scam email from a “stranded traveler” who needs exactly $347 works because it provides simulation parameters. Political movements exploit this as well, using individual cases as weapons to steer voters.

It’s not that we should care equally about everyone. We benefit when we understand how our caring system works: it’s a pattern-matching, simulation-running mechanism evolved for small-group dynamics... now operating in a globally connected information environment. In this world, we can still look at individual stories, but we should see them as crystallization points that enable collective action on broader patterns.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🛍️ eBay is banning third-party AI shopping bots

Shopping has become a major use case for LLMs: ChatGPT added shopping features in 2025, Amazon now lets you buy from external sites using AI, and Google launched the “Universal Commerce Protocol” to make it easier for AI agents to buy from online merchants. But while eBay is adding some first-party agentic shopping features, it recently updated its user agreement to ban “buy-for-me agents,” and LLM bots that “attempt to place orders without human review.” That’s on top of their existing prohibitions on scrapers and automated browsers. However, the company left the door open for potential official partnerships that would allow select AI bots to buy on eBay.

🚏🚽 A toilet company is seeing huge profits and stock gains from the AI boom

Toto, the Japanese toilet maker famous for its heated seats, has also been using its ceramics expertise in an unexpected domain: silicon microchip manufacturing. Since 1988, it’s sold “electrostatic chucks,” ceramic devices that hold semiconductors in place during chipmaking. Financial analysts recently noted that this product line could drive “significant profit growth” thanks to the AI-driven datacenter and memory boom, both of which require ample silicon chips. The company’s stock has soared accordingly; at the time of writing, it’s up over 25% in the last six months. Toto isn’t the only unusual Japanese company contributing to the semiconductor supply chain: a company that makes MSG seasoning produces chip-insulating films thanks to its mastery of amino acids, and a cosmetics firm that sells facial cleansers also has a “chip wafer cleaning business.”

🚏🚘 Electric cars outsold gas cars in the EU for the first time

Last December, about 218,000 fully battery-powered cars were registered across the EU, compared with 216,000 gas-powered cars (hybrids were not counted in either category), marking the first time in EU history that EVs outsold internal-combustion cars. EVs narrowed the gap quickly over the last year, with a 51% year-over-year increase in registrations, compared with a 19% drop in gas car sales. Non-plug-in hybrids (which only take gas but power a small battery through their engine or regenerative braking) are still the EU’s most-sold cars, but fully electric vehicles (now the 2nd most common) and plug-in hybrids (4th) are narrowing the gap.

🚏🐮 Researchers found a cow that can flexibly use tools

An Austrian cow named Veronika has been observed using brooms and sticks to scratch parts of her body she can’t normally reach, marking the first known instance of “flexible use of a multi-purpose tool by a cow.” She used the bristled end of a broom most of the time, but switched to the stick end “when targeting softer lower-body areas.” Cows, like most non-primate animals, are usually not considered capable of using tools (consider The Far Side’s famous “cow tools” comic), so scientists believe this may force us to revisit our understanding of animal cognition.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Polybius: The Authoritarian Consolidation Index (John Ganz) — An interactive poli-sci app that uses Claude to evaluate recent news against theoretical models like Linz’s “Perils of Presidentialism”, Gramsci’s hegemony, and de Tocqueville’s civil society to compute a score that measures the health of a given democracy. The app is pre-loaded with the USA’s “authoritarian consolidation index” score, but with your own API keys, you can run the analysis on any country of interest.

• The Complicator’s Gloves (The Daily WTF) — A classic parable about avoiding overcomplicated solutions when simple, cheap, and perhaps boring options are available. Engineers at one company dreamed up a fancy handlebar-heating system for bikes, only to realize that nobody had ever built such a feature because gloves already exist.

• I Should Have Loved Biology (James Somers) — Recalls how little the author enjoyed learning biology in school; despite its interestingness as a field, biology felt like a “lifeless recitation of names.” What if, instead of focusing on seemingly arbitrary facts, biology was taught historically by acquainting students with real biological questions and the scientific processes biologists used to answer them, giving students an opportunity to put themselves in the scientists’ shoes and wonder? What if it involved more inspiring and beautiful illustrations and explorable explanations?

• Yes, We’re in the Seventh Party System (Mr. Beat / YouTube) — Traces the history of the commonly agreed upon eras of American political history, each of which featured unique issues and coalitions. Most historians agree that FDR’s New Deal Coalition formed the Fifth Party System, and most believe the late 20th century saw the rise of the Sixth, but few agree on anything after that. Mr. Beat makes a compelling case that 2016 ushered in a Seventh Party System, noting that if you zoom out, today’s two major parties look quite different from what they did in Obama’s days.

🌀🖋 More from FLUXers

Highlighting independent publications from FLUX contributors.

The Keynes Quote Killer is an interactive mini-app from FLUX’s own Ben Mathes that examines popular economic quotes misunderstood in the public imagination, from Friedman’s “no such thing as a free lunch” to Schumpeter’s “creative destruction.” Oftentimes, the commonly understood meaning of the quote is the exact opposite of what the speaker actually meant in context!

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 219 — January 22nd, 2026 — Available at read.fluxcollective.org/p/219

Contributors to this issue: Neel Mehta, MK, Boris Smus, Ben Mathes, Erika Rice Scherpelz, Justin Quimby, Robinson Eaton

Additional insights from: Ade Oshineye, Anthea Roberts, Dart Lindsley, Jasen Robillard, Lisie Lillianfeld, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“Few people know how much you have to know in order to know how little you know.”

— Walter Ong

🏓🔁 The system always kicks back

John Gall had it right: “the system always kicks back.” And even when you “fix” something, you just move the failure. Gall again: “New systems create new problems.” The system preserves itself first. Your outcomes come second. Let’s see how this looks in context.

Demonstration vs integration

A flashy AI demo drops. Hype spikes. “This changes everything.” It doesn’t. The prototype exists in a vacuum, a controlled environment where all variables bend to its logic. But your product must plug into a big, entrenched system. The demo is fiction. The incumbent is a complex adaptive system. It self-defends.

The demonstration is frictionless by design. Integration is friction itself.

Disruption vs equilibrium

This new product is exactly what your users have been asking for! They will adopt it in droves. They don’t. Their lifestyle is held together by constraints, incentives, and politics—invisible architecture that maintains its shape. When we want change, this is a challenge, but for the most part, this stability is a good thing.

Resistance is not a bug. Resistance is the default.

Totalizing vs incremental

Le Corbusier’s urban designs were top-down, grand designs that overfit the model and underfit reality. The track record for such centralized plans is blunt: they don’t work. Gall’s law reminds us that working complexity evolves from working simplicity. From-scratch complexity fails. The all-in-one savior prototype dies because it doesn’t negotiate with the many stakeholders or adapt to the many edge cases. Doing so surrenders too much of what made the demo work.

Kill the revolutionary pitch. Run in parallel, not head-on.

Imposed vs emergent

That idea was pretty good, but you didn’t get buy-in. No one wanted to be the one to implement someone else’s idea. Instead, do the work. Be involved. Ship a small, contained system that actually works. Own one function. Let it build its own coherence. Let results—not rhetoric—displace the old machinery. This is real creative destruction: local, cumulative, irreversible. It demands humility. It rejects the founder’s fantasy of total transformation on day one.

Strategy begins where personal ambition ends.

Again and again, we see movement toward change trapped by the current dynamics of the system. This does not mean that change isn’t possible, but it requires going beyond the surface. Remember, the system always kicks back.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🇳🇬 Local solar power is booming in Nigeria, replacing diesel generators

Nigeria’s electric grid is infamously overburdened: 38% of the country has little or no access to power, more than any other country on Earth. Most rural towns have to rely on noisy, polluting, and expensive gas or diesel generators. But solar power is flooding into Nigeria and rapidly displacing old tech: Nigeria imported 1,721 megawatts of solar panels in the first half of last year, up from 500 MW in 2021, enough to meet 5% of the country’s total power demand. The resulting boom in local, off-grid solar generation has greatly improved the quality of life in Nigerian villages by slashing living costs, reducing pollution, and extending business hours. Financing has come from local investments plus the World Bank’s $750 million project to bring electricity to sub-Saharan Africa, a region where “more than 80% of the world’s people without access to electricity live.”

🚏🎿 Ski visits to Vail Resorts are down 20% this season due to low snowfall

Vail Resorts, a company that runs dozens of ski resorts across North America, reported that skier visits are down 20% this season. The major driver was “one of the worst early-season snowfalls in the western U.S. in over 30 years”: Colorado’s snowpack is more than 50% smaller than usual, and snowfall in the Rockies is down 60% compared to the historical average.

🚏📙 An AI conference allegedly featured 100+ papers with hallucinated citations

GPTZero, which bills itself as an “AI writing detector,” analyzed over 4800 papers accepted by NeurIPS, a prestigious AI conference, and found that over 100 of them included “hallucinated” citations, a telltale sign of a partially or wholly AI-generated paper. Some of these citations name nonsensical authors (“John Doe and Jane Smith” or literally “Firstname Lastname and Others”); others use titles or paper IDs that don’t match anything in the arXiv database. Perhaps unsurprisingly, open-source projects for writing research papers with AI have seen their popularity skyrocket in the last year; the jumps in the graph below around April and September 2025, suspiciously, “correspond to the paper submission deadlines” for NeurIPS and another major AI conference.

🚏🤑 Finance bros are quitting their day jobs to trade on prediction markets full-time

Prediction markets like Kalshi and Polymarket are now attracting professional, full-time traders who routinely drop tens of thousands of dollars on bets ranging from “who will be Time’s person of the year” to “how many times will a sports announcer utter their catchphrase.” Meanwhile, billion-dollar hedge funds are joining as market makers, providing liquidity for retail traders (i.e., posting ‘buy’ and ‘sell’ orders that individual users can trade against). It’s an analogue of the real stock market (and the crypto/NFT market, which also drew full-time traders and market makers). Prediction markets’ pro traders report that it’s an exciting job, but the risk of losing to insider trading looms large, and they say the markets aren’t doing much to stop it.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Loss of an Ideal (Mandy Brown) — Observes that “burnout” has taken on a quite different and broader meaning than its original connotation: the exhaustion and mourning that comes with having invested oneself into a failed political project—the “loss of an ideal.” The term’s semantic shift to its modern definition (focused on the stress and depletion that comes from ambition) follows a common pattern in philosophy: “from a symptom experienced by people struggling to change society to one experienced by people trying too hard to succeed within it.”

• The Microstructure of Wealth Transfer in Prediction Markets (Jonathan Becker) — Analyzes trading data from Kalshi to argue that prediction markets aren’t true ‘efficient markets’ due to a few key biases or asymmetries. The “longshot bias” (where “bettors overpay for low-probability outcomes”) leads to rare events being overpriced, and the “optimism tax” leads to “yes” outcomes being overpriced. Due to these factors, there’s a general wealth transfer from gamblers to the professional firms who now bet against them, especially in fields that attract emotional or uninformed bettors who rely more on narratives than statistics, like media, entertainment, or world events.

• Why Aren’t There More Québecs? (Conor Quimby / YouTube) — Examines why there are so few “large regions of settler nations with a distinct, non-indigenous… culture,” like Quebec in Canada. There were several candidates, like the Germans in South Brazil, Finns in northern Michigan, and Scots in Nova Scotia, but they either were too small or economically irrelevant to resist assimilation or lacked (at least tacit) support from the national government. Generally, Quebec analogues need to have enough ‘gravity’ that immigrants to the region should assimilate into its culture as opposed to the national culture.

• The Meditation Start-Up That’s Selling Bliss on Demand (The Atlantic) — Buddhist “jhanas” are altered states of consciousness produced from periods of strong concentration (i.e., focused meditation) and are growing in popularity among tech entrepreneurs. One startup is trying to extract the bliss, optimize it, and productize it in a gadget for the mass market, but whether biofeedback and AI can help bring novices to these jhanas remains to be seen.

🔮📬 Postcard from the future

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

// What Happens in the Datacenters?

// NEVADA DESERT - DATA CENTER CARCASS - AUGUST 2045

Subject: GRID TAP VETOED: Day 4, Lovelock Array. We are still dry.

From: Dusty_Rhoades (Logistics Lead, Temporary) To: Council_of_Cacti, The_Decommissioners

Team—

It’s clear the Aluminum Ovens we occupy are not cooling. The heat inside these vast, hollowed server barns is oppressive, a static reminder of the GPU-Fire that once ran here—the massive thermal load that was first for bitcoin mining and then training what became a third-tier model before the lab shut down. It left behind these impressive, empty shells we moved to after our last empty shells, from Oakland’s industrial revolution, got converted to condos. The power source meant to feed this beast—the unfinished nuclear plant in the other valley—is still shelved some 20 years later. First by the protest line outside town, now stuck in committees for years.

The city manager, Ms. Chen, needs a single name on the infrastructure permits for power, water, and sewer. She won’t talk to our “rotating circle.” Our current patchwork solar-trailer setup is a joke, and the six diesel-fed porta-potties are a short-term lie. This isn’t the invisible sprawl of Oakland’s abandoned warehouses anymore; here in Lovelock, they see us as a problem and a wrinkle to be ironed out.

Our leadership deadlock is paralyzing us. The Decentralists argue that a signature grants the city a Master Key to our future operations, a move back toward Autocratic Hierarchy. The Pragmatists say I should sign since I filed the initial lease. The Mythics insist on the “Oracle of the Dust” signing.

We have 72 hours (Confidence 85%) before we face eviction by the county health department. We need to agree on a leader or face the consequences of principled thirst.

Subject: RE: GRID TAP VETOED: Day 4, Lovelock Array. We are still dry.

From: Council_of_Cacti To: Dusty_Rhoades

Dusty, survival is not the sole metric. The purpose of moving here in 2045 was to adopt a governance model. We didn’t flee the comfortable anarchy of the old city warehouses just to mimic a Nevada Homeowner’s Association.

The Infrastructure Working Group’s plan for Energy Sovereignty is superior. They propose a pyrolysis unit: waste equals power. We can convert the residual server casings and plastics left by the previous tenants into heat, boiling our sewage for a micro-turbine. This cuts out Ms. Chen entirely. The previous operator’s failure proves the risk of relying on a single, massive Utility Interface that blocks the nuclear plant.

The inherent tension: We trade immediate access to cold, clean water for total operational autonomy. A single leader signing the permits gives the City of Lovelock veto power over our community’s core activities. We hold the Emergency Decision Vortex at 18:00. Do not argue for Monarchical Authority; argue for a multi-party signature system that limits the city’s power.

Subject: RE: RE: GRID TAP VETOED: Day 4, Lovelock Array. We are still dry.

From: Dusty_Rhoades (Logistics Lead, Temporary) To: Council_of_Cacti, The_Decommissioners

That pyrolysis plan uses the highly flammable insulation foam surrounding us for fuel. It’s a design for an Extravagant Disaster. We are in a highly specialized, abandoned structure; we can’t treat it like a desert fort.

I proposed a rotating, three-person, non-voting signature panel. Ms. Chen called it “A Recipe for Paralysis.” She isn’t interested in our philosophy.

And not to be all “we live in a society!!!111!!” to the former founder of the now-defunct Holacracy faction, but Ms Chen can call in the police on us. We may have the white papers, but the load-bearing good part of governance, the monopoly on violence, is something they have.

I am signing the water and sewer permits tomorrow at 10:00. The debate over who is the biggest Corporate_Shill can happen once the crisis is over. I take the hit. We’re thirsty. Get the water online.

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 218 — January 15th, 2026 — Available at read.fluxcollective.org/p/218

Contributors to this issue: Erika Rice Scherpelz, Neel Mehta, Boris Smus, MK,

Additional insights from: Ade Oshineye, Anthea Roberts, Ben Mathes, Dart Lindsley, Jasen Robillard, Justin Quimby, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“A library doesn’t need windows. A library is a window.”

― Stewart Brand

🏢🫂 Creative spaces must be owned by their inhabitants

MIT’s Building 20 is a great case study in accidentally creating a good place for incubating weird ideas. And it was largely accidental: the building was created as temporary construction for WW2 radar experiments. Since it wasn’t precious, people could change it. They demolished walls. The wiring/plumbing was visible and easy to change. The space was editable to a degree that even the most flexible modern office space can only imagine.

Contrast this with the office of the ad agency Chiat/Day in Los Angeles (case study; Wired article). Workers were forced to work in a “virtual office” — no fixed desks, no paper, and most critically, no real control over their space. This was a folly of high modernism: the image of what a modern creative organization “should” look like was blind to how work really got done in all its messy, embedded glory.

The mistake with the Chiat/Day office wasn’t that it was sterile. It was that it was imposed. A space full of color and knick-knacks and whimsy is just as sterile if its inhabitants are unable to adapt it to their needs.

If you want innovation, you need to give people the freedom to modify their space. And allowing desk photos is not enough. They need to be able to extend their mind into the space around them, whether that’s by shifting desks, putting up stickies, building physical models… or rerouting the power.

When inhabitants own their space, they can think more clearly as individuals, but this extension of the mind into the physical world is even more critical when we work together. When we can point at an artifact and say, “like that!” it can align us more than a verbal description ever can.

There’s a parallel in the digital world. As we rethink business processes and operations through the lens of AI automation, we must consider where we still need the flexibility to add meaning-making. In these spaces, we need to ensure that the human “inhabitant” of the process has the authority to change things.

In the meantime, knock down a wall. It might spur your next brilliant innovation!

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🚢 China’s trade surplus hit a record $1.2 trillion last year

China exported $3.77 trillion in goods last year, compared to $2.58 in imports for an overall trade surplus of nearly $1.2 trillion—an all-time-high and a 20% jump from last year. The US’s trade war with China (which led to a 20% decline in Chinese exports to the US) has hardly put a damper on Chinese exports, as the country has quickly pivoted to emerging markets such as Latin America, Africa, and Southeast Asia. High tech has been a major contributor to China’s export boom, including electric vehicles, computer chips, lithium batteries, solar panels, and industrial robots.

🚏👔 McKinsey is adding an AI round to its consulting interviews

The management consulting firm is reportedly incorporating AI into some of its final-round interviews. Candidates are required to “carry out practical consulting tasks” with the help of McKinsey’s internal AI tool: “you are expected to prompt the AI, review its output, and apply judgment to produce a clear and structured response.” Indeed, McKinsey has gone all-in on AI as of late: the company’s CEO recently quipped that his firm’s workforce numbers 60,000, split between “40,000 humans and 20,000 agents.” (Meanwhile, rival consulting firm BCG now has “forward-deployed consultants” who ‘vibe-code’ to build out client projects.)

🚏🌏 Coal power generation fell in China and India for the first time since the ‘70s

A new analysis found that China generated 1.6% less electricity from coal compared to a year ago, while India saw a similar drop of 3%. This marked the first time that both countries saw simultaneous drops in coal power generation since 1973. A major driver of the drop is that clean energy has been booming in the world’s two most populous countries: China added 300GW of solar power and 100GW of wind power last year, and India added 35GW of solar power and 6GW of wind power. It’s a welcome turnaround, since the two countries accounted for over 90% of the increase in global carbon emissions between 2015 and 2024.

🚏🤠 Matthew McConaughey is trademarking himself to stop AI fakes

The movie star has filed eight trademark applications with the US Patent and Trademark Office covering his voice and likeness, including several videos of him and an audio clip of him reciting his famous line, “alright, alright, alright,” from a 1993 movie. His attorneys said the move was an attempt to stop AI apps from “simulating his voice or likeness without permission,” such as in deepfakes.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• The Cult of the Founders (Crooked Timber) — Frames the well-known distinction between visionary-founder and operator-CEO in terms of prophets who rip up the rulebooks and create an ecstatic cult, versus priests who are rule-following administrators skilled in the “routinization of charisma.”

• Landholder vs Stockholder (Aeon Magazine) — Argues that the political theory with the most predictive power over modern political divides isn’t Marx’s divide of the bourgeoisie versus the proletariat nor the French Revolution’s three estates, but rather David Hume’s split of landholders (those with agrarian, family-oriented, and rent-seeking interests) and stockholders (upwardly mobile industrialists and financiers with fewer ties to the land). The former aligns more with the modern right and the latter with the left, which explains “how the Left can be both the party of finance, and the party that tries to overthrow capitalism.”

• Where Americans Choose to Move and Where They Leave (John D. Johnson) — Investigates domestic migration patterns in the US, finding that “the patterns often correspond better to ecoregions than political boundaries.” For instance, there’s a sharp population drop in the narrow band between 100º and 103º W, a region that covers seven states from North Dakota to Texas. Meanwhile, the southeastern Piedmont region (which spans parts of seven states) saw significant growth, even as other parts of those states experienced population loss.

• The Unstable Queen (Matthew Mason) — A veteran robotics and AI professor observes that, in both human and robot affairs, sensing and action are deeply interlinked: “sometimes the eye leads the hand, and sometimes the hand leads the eye.” For instance, if a robot is trying to repeatedly move a chess piece around a board without letting it fall over, it can use visual sensing to track the piece’s position, or it can use contact with the board (i.e., action) to keep the piece straight, such as by sliding it around instead of picking it up.

📚🛋️ Book for your shelf

A book that will help you dip your toes into systems thinking or explore its broader applications.

This week, we recommend How Buildings Learn: What Happens After They’re Built by Stewart Brand (1995, 256 pages).

Stewart Brand’s How Buildings Learn delves into the fascinating idea that buildings are not static structures but entities that evolve and transform over time. Brand argues that the structures most successful in the long term are those designed with flexibility at their core. This adaptability allows them to be continually modified, repurposed, and improved by their users to match their needs as they change.

A critical dimension is which layers of a building evolve, and at what rate: your furniture arrangement will change more often than the building foundation!

For a building, these layers generally include the “stuff”, the space plan, services, skin, structure, and site.

This idea of shearing layers can be applied beyond buildings, as in this example from Gratus Devansen, which applies it to software systems. Opinions may differ on the accuracy of each layer, but the key is to have clear opinions about your shearing layers when building software that needs to evolve.

And the same holds for other domains. Whether physical, organizational, or social, any system that needs to evolve will have layers. Instead of rigid, monolithic structures, Brand’s work inspires us to create platforms that encourage adaptation and customization at the right layers by providing well-defined interfaces, modular components, and opportunities for extension.

And metaphorical applications aside, it’s a cool book about buildings!

Subscribe now

© 2025 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 217 — January 9th, 2026 — Available at read.fluxcollective.org/p/217

Contributors to this issue: Justin Quimby, Erika Rice Scherpelz, Ade Oshineye, Neel Mehta, MK, Boris Smus

Additional insights from: Anthea Roberts, Ben Mathes, Dart Lindsley, Jasen Robillard, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“The map had been the first form of misdirection, for what is a map but a way of emphasizing some things and making other things invisible?”

— Jeff VanderMeer, Annihilation

📃🐦‍🔥 The manifesto is death, the manifesto is life

As a format, the manifesto was built for a world of scarcity and permanence. From Marx and Engels to the Futurists, the form demanded a voice proclaiming unchanging truths, serving as a stake in the ground around which movements could coalesce. It presumed that ideas, once articulated with sufficient force, could hold their shape long enough to matter.

AI has dissolved these conditions. When anyone can generate infinite variations/refutations of any argument, the idea that a piece of text will endure and define becomes almost quaint.

Manifestos traffic in certainty, in the rhetorical power of the declarative sentence. We believe. We demand. We will. But the systems now reshaping our world are probabilistic, not prophetic. They operate in gradients and weights, not absolutes. The future isn’t being declared; it’s being negotiated in real time, across millions of prompts and outputs, in ways no single document can capture or constrain.

But there’s still value to saying, “This is not just slop for the sake of engagement. We believe in these ideas. And we earned these beliefs the hard way.”

In recent years, manifestos have emerged from a rock-tumbling process among small groups of practitioners, such as the Agile Manifesto and the Software Craftsmanship Manifesto. That process yields a compromise (and compromised) statement in order to include everyone with a seat at the table. Afterwards, the nascent community must aggressively defend the integrity of the founding statement to achieve the manifesto’s goals. This process can become a wall to keep out non-signatories. Yet this dedication to the original form is critical in storming/norming/forming the community of practice.

In that sense, manifestos are death. They preclude the possibility of better futures: signatories have already committed their signatures to a particular vision of the future… even if that future is obsolete. Growth and learning can look like a betrayal of the founding statement. The very form of the manifesto aims toward scissor statements rather than playful curiosity and exploration of unfolding possibilities. In defining what should be, they can restrict what is imagined. Manifestos are strong opinions, strongly—not weakly—held.

The manifesto isn’t dead, but like all information vehicles, it must evolve. With slop manifestos so easy to generate, words alone are not enough. Even if the manifesto is obsolete as a commitment device, it can still act as a coordination mechanism that births communities. Systemic change needs sustained momentum, attention to messy detail, and a community aligned through practice, not just words. In today’s clickbait, dopamine-fueled landscape of ideas and movements, momentum requires groups of people pushing change over long periods of time. The movement is the light; the founding statement is merely the shadow.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🇻🇪 Polymarket is refusing to pay out bets on the Venezuela invasion amidst an insider trading scandal

The prediction market startup Polymarket, which allows users to buy ‘shares’ in potential events and pays out if an event occurs, had a market on the question “Will the US invade Venezuela by [some date]?” The company said that it’d pay anyone who bought ‘yes’ if a “consensus of credible sources” agreed that the invasion had happened. The US did indeed attack Venezuela and abduct its president earlier this week, but Polymarket refused to pay out, saying that an invasion only counts if the US controls some part of Venezuelan territory. Military scholars may argue over the meaning of “invasion” vs. “occupation,” but the story became a scandal when it was revealed that a brand new Polymarket user made a trio of huge bets on the invasion just a day before the attack happened (when the market predicted a roughly 6% chance of invasion), netting over $400,000 in profit on $35,000 in bets. This raised suspicions of insider trading, a frequent concern for news-related wagering.

🚏🎓 A single Chinese university is dominating American schools in AI patents

China’s prestigious Tsinghua University (sometimes called “Stanford, MIT, and Carnegie Mellon all rolled into one”) has become arguably the world’s leader in AI and ML, racking up almost 5,000 patents in those spaces since 2005—more than MIT, Stanford, Princeton, and Harvard combined. Tsinghua has more AI research papers among the 100 most cited than any other school; its only competition is Google and Microsoft. However, while Tsinghua churns out more AI patents, they’re not cited as often by future patents; Harvard and MIT, meanwhile, put out fewer but more ‘impactful’ patents.

🚏⚕️ Utah will allow an AI doctor to renew medical prescriptions

The Utah Department of Commerce announced a pilot program that will allow an AI doctor startup (which lets patients do a medical consultation with a bot and get diagnoses) to automatically renew medications on patients’ behalf, marking “the first state-approved program in the country that allows an AI system to legally participate in medical decision-making.” The startup says it’s put numerous guardrails in place, such as having human doctors validate “the first 250 prescriptions issued in each medication class,” securing a special malpractice insurance policy, and escalating to real physicians in borderline cases.

🚏👨‍🦲 Korea’s president wants the national healthcare system to cover baldness

South Korean President Lee Jae Myung asked the country’s healthcare minister if the Korean public healthcare system could start covering hair loss, a condition that’s normally considered aesthetic and thus not covered by health insurance. Lee insisted that many young people losing their hair see it as a “matter of survival,” and he may have a point: South Korea has an infamously “rigid definition of beauty,” and physical appearance is considered important for getting jobs. “Your appearance is also a credential” is a common saying, and almost all jobs require applicants to submit a photo of themselves. By this logic, reducing hair loss could help the Korean public improve their employment prospects (although one wonders if improving everyone’s attractiveness wouldn’t actually help, since you’d just be raising the level of competition in a fixed pool.)

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Where Good Ideas Come From (For Coding Agents) (Sunil Pai) — Observes that AI coding agents struggle when given wide-open tasks because they fill in the blanks with the most statistically likely solution—but the ‘typical’ solution is unlikely to be the ‘correct’ solution. To get the agent to create a solution that works for your use case, you need to constrain them to only explore the “adjacent possible” (i.e., take incremental steps) with clear constraints, context, and validation at each step. Concretely, this takes the form of a curated brief for the LLM that encodes your judgment: goals, non-goals, constraints/invariants, prior art, what “done” means, etc.

• Prediction Markets Barely Make Money; Sportsbooks Make Money (Financial Times) — Argues that ‘prediction markets’ like Kalshi, for all their high-minded talk of “financializing differences in opinion,” are just glorified sports betting sites, when you look at where most of their volume comes from. It seems like a structural pattern: any generic prediction market would naturally trend toward sports betting, which offers a high frequency of bettable events, high volatility, quick resolution, predictable schedules, and a “large ecosystem of resources offering information” for bettors. More highbrow topics, like election results (or which country the US would invade next), are much less attractive to gamblers because they are infrequent, less legible, overly clustered, and slower to resolve.

• Tales of the Tyrant (The Atlantic, 2002) — Paints a portrait of Iraq’s then-dictator Saddam Hussein, a tyrant who appears to have unlimited choices but actually has the fewest. He cannot learn or reform because the fear-based system he built to survive requires him to purge his allies, and those who remain must lie to survive him, skewing his understanding of reality and causing errors to compound until catastrophic miscalculation becomes inevitable.

• Product Managers, Ducks, and Dogs Marking Territory (Rachel by the Bay) — Retells a famous story about how an artist for a chess-based video game purposely added a small duck to the animations so that the PM, who wanted to give advice and leave their mark on the game, would have some quick and obvious feedback to give: “it’s great, just remove the duck.” The sacrificial duck protected the rest of the game from the “meddling manager.”

🔍📆 Lens of the week

Introducing new ways to see the world and new tools to add to your mental arsenal.

This week’s lens: resonance.

You’re reading a book when a chapter makes you stop. Something about it clicks. The more you read it, the more meaningful it becomes. Days later, you find yourself returning to it.

This experience has a name: resonance.

Resonance is personal yet not entirely subjective. While our own taste plays a role, resonant experiences share a distinctive quality: fractal meaning. Like a coastline that looks complex at any scale, resonant ideas please on the surface and continue to reveal depth as we dig in. The details make the whole more meaningful, and the whole makes the details more meaningful. Think of a favorite song where each listen uncovers new subtleties or an architectural space that feels more welcoming as you notice its thoughtful details.

Resonance can appear anywhere. We aren’t surprised when we have this experience with a great piece of literature, a touching piece of music, or a decades-long relationship. But we’d be surprised to find it in an app or a TikTok video. We shouldn’t be. There is room to raise the bar. As the signatories of the Resonant Computing Manifesto state, we should bring resonance to more of our technological landscape (disclaimer: some FLUX members have signed).

Remember, though, that as good as resonance feels, it isn’t Truth. Instead, resonance reveals something important about our values and goals. And sometimes that can be manipulative, appealing to certain values (e.g., belonging, engagement) rather than others (e.g., thoughtfulness, reasoning). For example, Shakespeare’s biggest speeches, such as those from Julius Caesar and Henry V, are structured to evoke resonance.

What resonates with you right now, and what might that be revealing?

🔮📬 Postcard from the Future

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

// How might the American removal of Maduro in Venezuela and growing Chinese naval impact Europe in unexpected ways?

// Dec 2026 Bloomberg Hot Takes Article

While the US arrest of Venezuelan president Nicolás Maduro Moros in January of 2026, which occurred with virtually no resistance by the Venezuelan Navy, did cause significant concern in various EU naval force subcommittees, it did not trigger action. Even the multiple large-scale Chinese naval exercises surrounding Taiwan in late 2025 and through 2026 were not enough to wake the sleeping European giant. When China demonstrated a system that combines multiple containers, each acting as a specific element, such as a missile launch module, integrated sensor suite, or close-in weapon systems, even that did not precipitate major change.

The tipping point? Yacht enthusiasts, who take photos of ships going in and out of major ports in an effort to “yacht-spot” famous ships, spotted these containers on a Russian-flagged ship heading into the Russian port of Vostochny in August of 2026. Breaking news hit the socials of both the shipping and super yacht communities, then rapidly entered the main news cycle.

A flurry of EU hearings followed. Due to the potential threat of the Russian navy weaponizing the European shipping lanes and using them as a potential first strike vector against EU nations, along with the American president’s willingness to use force to remove elected leaders, immediate action was taken. The EU needed a strong naval force for deterrence. The EU nationalized over a dozen shipyards in the name of rebuilding naval might. The German shipyards of Lürssen and Nobiskrug, along with the Italian Fincantieri and Benetti yards, and the Netherlands Feadship, Amels, Heesen, and Oceanco facilities are all being rapidly converted to build hulls for military naval fleets.

All the recently-printed AI centi-millionaires from the latest round of acquisitions wanting a new super yacht are going to have to wait.

Subscribe now

© 2026 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 216 — November 20th, 2025 — Available at read.fluxcollective.org/p/216

Contributors to this issue: Justin Quimby, Ade Oshineye, Erika Rice Scherpelz, Neel Mehta, Boris Smus, MK

Additional insights from: Anthea Roberts, Ben Mathes, Dart Lindsley, Jasen Robillard, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“Without writing, the literate mind would not and could not think as it does, not only when engaged in writing but normally even when it is composing its thoughts in oral form. More than any other single invention, writing has transformed human consciousness.”

— Walter Ong

📝 Editor’s note: We’ll be taking the rest of 2025 off for the American holiday season. We’ll see you again in January!

🎲♾️ You can’t play infinite games without finite wins

A jazz quartet can improvise for hours, but only because each musician lands their notes, keeps time, and responds to others. The music continues by weaving together many small successes. Without these small wins, the music collapses. The infinite play of creation depends on countless finite completions.

The appeal of an infinite game is its promise of continuity, to play not to win but to keep playing. The idea has become shorthand for sustainable strategy and long-termism. But the concept can get flattened into abstraction, a focus only on the long term with no concern for the present.

In practice, every infinite game is built from a series of finite wins. Without them, the game ends. These finite wins are moments of completion: closing a deal, shipping a product, hiring a teammate, fixing a big. They are stories within the story, self-contained arcs that give the ongoing narrative momentum.

This need for finite wins highlights what infinite games are not: they are not an excuse to avoid commitment to outcomes. Nor are they perpetual motion with no direction. But we must also avoid the opposite mistake of taking short-term wins as progress toward something that lasts.

Infinite games are recursive: they consist of wins that beget other wins. These wins must tend to the system’s metabolism. Finite wins sustain this metabolism by providing the resources that replenish its energy. Without nourishment, the infinite game collapses. But an obsession with empty calories today also destroys long-term health.

This need for ongoing wins is inescapable because the risk of ruin (that’s this week’s lens) looms over every infinite game. No amount of initial resources—capital, energy, goodwill—can substitute for the capacity to regenerate. The infinite is something you must continually earn.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏♨️ Americans are trying to heat their homes and businesses with Bitcoin mining

A new space heater called the Heatbit Trio runs a Bitcoin miner under the hood and uses the waste heat to warm a room; the Bitcoin earned offsets about 20% of the electricity cost. It’s part of the growing industry of Bitcoin-powered home heating in the US, driven by rising electricity bills (which, in turn, are partly driven by the explosion of power-hungry data centers). In some American cities, industrial users like car washes and concrete companies are using Bitcoin-powered water heaters, while one New York-based spa uses Bitcoin to heat its hot tubs.

🚏🌧️ Iran is trying cloud seeding amid a worsening drought

Iran is facing its worst drought in 60 years as rainfall has dropped 89% from historical averages; the Iranian president darkly joked that they might have to “evacuate Tehran” if the water shortages continue. So, Iranian authorities have reportedly “sprayed clouds with chemicals to induce rain” above the country’s largest lake, which has almost entirely dried out. The nearby UAE has experimented with cloud seeding—spraying salts into clouds to create nuclei for potential raindrops—in the past, with some success.

🚏🇵🇪 EV sales are booming in South America, driven by Chinese brands and ports

Electric cars now represent 10.6% of all new cars registered in Chile, 9.4% in Brazil, and 28% in Uruguay—all record highs. (That’s behind China at 51% and Europe at 56%, but close to surpassing the US at 10%.) In Peru, electric cars are still a small market, but sales are up 44% year over year. Chinese car brands are the dominant players in the EV space here, with BYD “lead[ing] electric car sales in Brazil, Colombia, Ecuador and Uruguay.” (Meanwhile, Tesla only just opened its first showroom in South America.) One driver of Chinese brands’ dominance is a Chinese-built ‘megaport’ north of Lima, created as part of the Belt and Road initiative, which will bring tens of thousands of cars to the continent every year.

🚏🏘️ American homebuyers are getting older and richer than ever

New data from the US’s National Association of Realtors shows that the median age of first-time homebuyers has reached 40, up from the early 30s just a decade ago. The median age of all homebuyers is now a staggering 59. What’s more, first-time buyers now represent just 21% of all homebuyers, a record low. The remaining first-time homebuyers seem relatively wealthy, with most using personal savings for the down payment rather than gifts or loans from friends or relatives, which were the primary methods in the past. Plus, they now put down 10% on their houses, up from 5–6% a decade ago, the highest level since 1989.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Why AI Writing is Mid (Interconnects) — Argues that the blandness of LLM writing is a structural result of how models are trained, rather than a technical issue that’ll be solved if we throw more compute at it. RLHF introduces biases toward sycophantic, unchallenging, and generic prose (none of which are hallmarks of good writing); the great ‘averaging’ effect of LLMs strips away the quirkiness and uniqueness that makes written works memorable and fun; and the “forced neutrality” of language models mean that they don’t have any human opinions or perspectives, which is half the reason we read things to begin with.

• Is AI Turning Music Streaming Services Into Comment Sections? (Calm & Fluffy) — Argues that our new era of AI music, where anyone can publish personalized songs and remix others’, is an odd sort of return to the traditional world of folk music, where music was participative (rather than being broadcast from certain big performers), decentralized, and specific to certain communities. Perhaps the 20th-century world of mass broadcast media, record labels, superstar performers, and homogeneous global consumption will go down in history as the exception rather than the rule.

• Real Purchasing Power Over Time Is Not Economic Welfare Over Time (Steve Randy Waldman / Interfluidity) — Observes that economists track income over time because it’s easy to measure, but what they’re really trying to maximize is welfare. However, in the medium- and long-term, being able to buy more stuff doesn’t necessarily make people better off. Waldman examines a variety of factors, such as how cheaper options sometimes go extinct, forcing everyone to a more expensive option for no net welfare gain (e.g., suburbanites having to buy cars because transit-specific towns have become rare). What’s more, the composition of bundles changes over time (e.g,. housing and healthcare take up more of people’s budgets as consumer goods get cheaper), so tracking the quantity of a fixed bundle that people can buy isn’t very informative to begin with.

• The Primitive Tortureboard (Marcin Wichary) — A deep dive into keyboard layouts reveals that QWERTY evolved from its inventor solving mechanical typewriter jamming issues (reducing clashes from once every 26 words to once per 1,000), while containing the suspicious coincidence that T-Y-P-E-W-R-I-T-E-R can be spelled entirely on the top row (a 1 in 5,000 chance). Alternatives like Dvorak failed not due to inferior design but because QWERTY was already good enough, and switching costs outweighed marginal benefits.

🔍📆 Lens of the week

Introducing new ways to see the world and new tools to add to your mental arsenal.

This week’s lens: Risk of ruin.

A venture capital portfolio generally has a wide distribution of outcomes. They spread their risk across many bets, and the big wins carry the rest. If success at the top is high, then the average is decent enough for them to make money.

But things look very different for the individuals within a startup. They are exploring only one path (or, at most, a relatively small number). Averages don’t matter when your success or failure is linked to any one particular outcome.

There is a critical asymmetry here. For a VC, a loss is incremental. For an individual startup, some losses end the game. This is the risk of ruin. Risk of ruin refers to events that introduce irreversible failure, a loss so severe it halts the possibility of future recovery.

Mathematically, this sort of ruin is an absorbing barrier. Once you cross such a barrier, you don’t bounce back. Ruin doesn’t always look like immediate death. For example, a species can experience functional extinction a generation before actual extinction if there are no breeding pairs left.

In this world, risk management isn’t just about spreading your bets and avoiding single points of failure. It’s about avoiding irreversible negative outcomes. Ruin is the ultimate one-way door.

The risk of ruin gets at the fundamental difference between the VC and the startup employee. The VC can think in terms of expected utility—the average of all possible outcomes. But actors on a particular path should instead look to ergodicity economics: maximizing realized wealth yields better outcomes than maximizing utility. This is true even without ruin, but ruin makes it especially stark: it doesn’t matter what happens on average if you’re dead.

The risk of ruin lens tells us that losses aren’t created equal because the average outcome doesn’t matter if you don’t live long enough to see it.

Subscribe now

© 2025 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 215 — November 13th, 2025 — Available at read.fluxcollective.org/p/215

Contributors to this issue: Wesley Beary, Neel Mehta, Dart Lindsley, Erika Rice Scherpelz, Jasen Robillard, MK, Boris Smus

Additional insights from: Ade Oshineye, Anthea Roberts, Ben Mathes, Justin Quimby, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“Do you know the difference between an error and a mistake, Ensign?” “No, sir.” “Anyone can make an error, Ensign. But that error doesn’t become a mistake until you refuse to correct it.”

— Grand Admiral Thrawn

⏱️🎭 The asymmetry of clock speed

When a company must seek acquisition to avoid shutting down, time warps in strange ways. For the acquirer, the process unfolds methodically—due diligence, approvals, financial models, risk assessments. For the company being acquired, days stretch toward eternity. Payroll looms. Morale dips. Both sides discuss the same deal, but they live in different versions of time.

Clock speed, based on the concept of how quickly a computer’s processor can run, describes how fast a system perceives, decides, and acts. In an acquisition, clock speed asymmetry becomes unavoidable. The acquirer’s slower pace reflects its need for precision and control. The target’s faster pace reflects its need for survival. Each side’s tempo makes internal sense, while the mismatch can distort communication and trust.

This dynamic isn’t unique to acquisitions. It echoes the Innovator’s Dilemma: established systems optimize for stability, while emerging ones race ahead in search of viability. Where systems with different time horizons meet—startups working with enterprises, local teams coordinating with global ones, or new initiatives trying to gain traction within established organizations—the same friction appears. Slower systems protect continuity. Faster ones seek adaptation.

Once you see this tension as structural rather than personal, you can start to work with it. The slower side might set up a “fast track” for decisions that keep the other side from stalling out. The faster side might use the wait to prepare for what’s next. Both sides can benefit from surfacing the speed gap: what truly needs to move fast, and what actually benefits from patience?

The asymmetry of clock speed tells you where the system’s real constraints lie. You can’t eliminate the gap—but you can learn to tune it.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🎸 An AI-generated song topped the Billboard country chart

An AI-generated country song titled “Walk My Walk,” ‘created’ by the fictional band Breaking Rust, hit #1 on Billboard’s Country Digital Song Sales chart after quickly racking up over 3 million streams on Spotify. Billboard confirmed that “at least six AI or AI-assisted artists” have debuted on Billboard’s charts of the most popular songs, spanning genres from rock to blues, including a surprising number of Christian and gospel songs.

🚏🌽 Farmers in Malawi are getting help from a chatbot that speaks the local language

An AI chatbot called Ulangizi, which means ‘advice’ in the Malawian language of Chichewa, is helping farmers in Malawi deal with droughts, pests, and other challenges. Farmers can ask the bot questions on WhatsApp, chat with it by voice, or send photos of problematic crops or livestock. The bot was built by a UK nonprofit that trained ChatGPT on government agriculture guidelines and traditional farming knowledge. The nonprofit disseminated the AI tool by training tech-savvy farmers, who could then help other farmers (including those unfamiliar with phones or keyboards) in their community.

🚏🏭 China’s CO2 emissions have been flat for 18 months

China’s carbon dioxide emissions have more or less plateaued since early 2024 after years of steady growth. Power demand in China continues to increase, but clean power sources are “covering all of China’s growth” thanks to a record build-out of wind and solar. China’s real-estate bust is also helping here by reducing demand for steel and cement. But despite all the progress elsewhere, China’s oil consumption is still increasing, driven by a boom in manufacturing of plastics, fibers, and various chemicals.

🚏🥐 A French gourmet restaurant offers meals to the needy for ​​€1

A fine-dining restaurant in Marseille called Le République, founded by a former three-Michelin-starred chef, allows patrons of local charities to enjoy a gourmet three-course meal for just 1 euro thanks to subsidies from those charities. It’s one of many restaurants solidaires (solidarity restaurants) that’ve popped up in the seaside city: eateries founded as social enterprises that focus on helping low-income locals, homeless people, immigrants, striking workers, and anyone else who needs a good meal but can’t afford much.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Why Big Companies Keep Failing: The Stack Fallacy (TechCrunch) — Explains the Stack Fallacy, in which people believe that it is trivial to build the layer above, but very difficult to improve the layer below. Examples include browser engineers who ridicule web development and pure mathematicians who deride physicists as merely applied mathematicians.

• TextEdit and the Relief of Simple Software (The New Yorker) [Archived] — Praises the “brute simplicity” of the Mac’s ancient TextEdit app, a bare-bones program that offers a blank white screen for writing and not much else. It’s low maintenance, with no AI features, accounts, subscriptions, redesigns, or cloud services. It gets out of your way and lets you write without imposing a particular format or “philosophy of organization,” which makes it “the closest you can get to writing longhand on a screen”—a refreshing change of pace from the overwrought software we’ve become accustomed to.

• At a Loss For Words (American Public Media) — Highlights an alarming trend of American primary schools that are embracing the flawed ‘three-cueing’ method of literacy, which is giving children poor (or, in some teachers’ minds, nonexistent) reading skills. Compared to the traditional phonics-based method of ‘sounding out’ words, three-cueing trains kids to use context clues to guess what a written word says, leading to bizarre occurrences like kids seeing the word “horse” and saying the word “pony” instead, because they aren’t actually reading the letters on the page.

• When Should You Pull the Goalie in Hockey? (Wage World / YouTube) — Develops a mathematical model for when sports teams should take ultra-aggressive moves, such as a hockey team replacing their goalie with another attacker. The core insight is that high-variance moves (even those with negative expected value) are a good bet when you’re losing with not much time left, since the goal is to win—not maximize your point differential.

🔍📆 Lens of the week

Introducing new ways to see the world and new tools to add to your mental arsenal.

This week’s lens: frame wizardry.

In a high-stakes strategy meeting, one exec frames the company’s failing product as a misstep, exposing the team’s pattern of avoiding direct customer engagement. Another exec challenges that, framing the failure as a signal from the market. A third steps in and weaves both frames together to chart an alternate path forward.

Frame-making is the act of defining a way of seeing. Often, whoever speaks first makes the frame. Frame-breaking disrupts that definition. It’s not just about disagreeing with the dominant frame; it’s about replacing it with something else (or at least attempting to).

Frame wizardry is the ability to stack, switch, and synthesize frames, often finding a third way in what may seem to be a binary tradeoff. Frame wizardry allows us to bend perception and escape the trap of any single worldview. It resists binary thinking, offering a “yes, and” where others see “either/or.”

Frame wizardry also reveals how many conflicts are really clashes of framing, Sensemaking is less about consensus than about designing better disagreements. Some problem-solving books, such as Thomas Wedell-Wedellsborg’s What’s Your Problem? start from the assumption that framing is the most important problem solving skill.

Being a frame wizard doesn’t require magical brilliance. More often, it starts with a willingness to listen to the truth in the seemingly conflicting perspectives others already hold.

Subscribe now

© 2025 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 214 — November 6th, 2025 — Available at read.fluxcollective.org/p/214

Contributors to this issue: Wesley Beary, Ade Oshineye, Erika Rice Scherpelz, Ben Mathes, Stefano Mazzocchi, Dart Lindsley, Neel Mehta, Boris Smus, MK

Additional insights from: Anthea Roberts, Jasen Robillard, Justin Quimby, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“There is nothing noble in being superior to your fellow man; true nobility is being superior to your former self.”

— Ernest Hemingway

😣♟️ When every move has a cost

There’s a scenario every leader dreads: when every move puts you in a worse position. You’re pinned, but waiting feels just as dangerous. You move, you lose. You don’t move…you lose slowly.

That’s zugzwang: a chess term for when every move worsens your position. In games, you can mathematically define this point. In real life, there’s no perfect information, no singular board. Just shifting situations where even doing nothing is a choice.

Another tricky situation is when life gives you a saddle point. At this critical threshold, some directions seem to lead up, others down, but no direction is neutral. The present is unstable and any move commits to a direction, but the long-term direction of the gradient is hard to detect.

Organizations hit saddle points more often than they realize:

• You outgrow your current strategy, but the new one is unproven.

• Your competitors move, but you don’t have enough signal to justify a big shift.

• You consider a reorg, but you know it could backfire just as easily as succeed.

The human tendency to over-anticipate makes this even more challenging. We move early to gain an advantage, but acting too soon can trap us in a new, worse terrain, especially when our commitment prompts others to respond.

This is why war games and scenario planning exist. Not to predict, but to simulate: to explore how hard it is to recover from being wrong.

Yet sometimes you must commit despite uncertainty. And if the very possibility of reversing your decision can endanger it, you might take the extreme option of burning the ships, making going back impossible. You turn a local maximum—where you’re stable but stuck—into a local minimum in a new direction. It forces motion. It says, we’re committed now. It is irreversible, but it also redraws the map.

In systems with emotional, political, or resource asymmetry, movement costs are never evenly distributed. Defending home ground is easier in some ways (your troops and supplies are already there), but uncountably more costly psychologically because the damage is personal and retreat feels like collapse.

What these situations share is that the first move isn’t choosing the best path; it’s recognizing the terrain and realizing that sometimes the best you can do may still not be good enough. Even n-ply thinking has its limits.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🇦🇺 Australians will get 3 hours of free electricity a day thanks to solar power

A solar panel boom in Australia has created an energy glut in the middle of the day: wholesale prices are very low or even negative during that period. As a result, the federal government is now requiring utility companies to provide customers with free electricity for at least three hours a day. Households with ‘smart meters’ will be allowed to run appliances such as AC units, washers, and dryers for free during that period. The idea is that power demand will be shifted to this cheap period, thereby reducing demand in the peak evening hours, when power would be more expensive and potentially less renewably generated.

🚏🤑 The Coinbase CEO trolled prediction markets by saying random words in an earnings call

A hot new offering on ‘prediction market’ sites like Kalshi is “mention markets,” where people can bet on which words someone will use in a public forum, such as an earnings call. Coinbase’s CEO apparently saw one of these mention markets and decided to mess with them, saying the following at the very end of his company’s latest earnings call: “I just want to add here the words Bitcoin, Ethereum, blockchain, staking, and Web3 to make sure we get those in before the end of the call.” (People seem to have been betting on whether he’d mention specific crypto keywords in the call, so he threw these non-sequiturs in at the end to scramble the markets.) Armstrong insisted it was just for fun, but Bloomberg warned that it showed how easily these markets can be manipulated, potentially opening the door to insider trading.

🚏🏢 October layoffs in the US hit their highest level in 22 years

American companies cut over 150,000 jobs in October, a striking 175% increase from last year and the most October layoffs since 2003. And overall, this year has been the worst for layoffs since 2009. The technology, retail, and service industries were the hardest-hit in the private sector; major reasons include AI, cost-cutting, and “belt-tightening” for both consumers and firms. In the public sector, the federal workforce has been getting slashed by layoffs.

🚏🎼 Musicians are suing Spotify over “billions” of fake plays from bots

A new class-action lawsuit is accusing Spotify of allowing “billions of fraudulent streams” generated by bots. Because Spotify pays artists out of a fixed pot of funds, these allegedly fraudulent listens would give the benefiting musicians a greater share of the money while depriving all other artists of royalties. Alleged examples of botting activity cited in the lawsuit include 250,000 streams of a Drake song that came from Turkey but were tunneled through British IPs “to obscure their origins;” some accounts listening to Drake’s music “23 hours a day;” and large amounts of streaming coming from places with “zero residential addresses” but presumably a lot of data centers.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• Seven Data-Driven Lessons From the 2025 Elections (Strength in Numbers) — Independent political statistician G. Elliott Morris derives some insights from the US’s off-cycle elections, which were held this Tuesday. He finds that a focus on affordability mattered more than ideology (a similar pattern to 2024), and that a lot of the “shifts” we saw last November, such as the voting patterns of Latinos and Gen Z, appear to have snapped back to long-term averages.

• A Tool That Crushes Creativity (The Atlantic) — Charlie Warzel observes that the “SlopTok clones” like OpenAI’s Sora and Meta’s Vibes—populated by an endless feed of AI-generated short videos—are the logical conclusion of social media’s endless desire to minimize friction. But this frictionless mainlining of content into our brains leads to “a world of creativity without craft.” More broadly, “slop emerges when we eliminate not just toil (the burdensome aspects of work) but labor itself (the meaningful human engagement with creation).”

• Why the World’s Biggest Film Bombed in America—and What That Says About Us (Steven Mintz) — Argues that Ne Zha 2’s massive success in China ($2.2 billion worldwide) but failure in America exemplifies the clash between Chinese epic storytelling and American narrative expectations of the linear individual triumph of a hero’s journey. The film’s rejection arguably reflects Hollywood’s declining ability to absorb and transform foreign cinematic traditions, a strength it once wielded to incorporate German Expressionism, Italian Neorealism, and Japanese samurai epics into its own idiom.

• The Goodhart Trap in Politics (Dustin Moskovitz) — The Asana founder argues that many political candidates optimize their positions too much to match ‘what the data says,’ thus losing their authenticity—a sort of Goodhart’s Law trap. He proposes a 2x2 matrix of candidates: boring vs. passionate, and data-informed vs. conviction-driven; those who maintain conviction and authenticity tend to succeed, even if their résumés are thin. That holds true across the political spectrum, from Zohran Mamdani to Trump.

📚🧸 A reading list for young systems thinkers

Long before we learn the language of systems thinking, many of us encounter its ideas in stories. The best children’s and young adult books entertain while expanding our minds. Here are some books that—in hindsight—helped us on our journey.

If You Give a Mouse a Cookie by Laura Numeroff
A classic, step-by-step causal loop diagram disguised as a bedtime story. Each action generates the next in an accelerating feedback loop until the system returns to its point of origin. It’s a perfect sketch of how a system can be driven by its own outputs—and it all comes back to milk and cookies.

A Wrinkle in Time by Madeleine L’Engle
Through the cold, synchronized sameness of Camazotz, the lush diversity of Uriel, and the emotional richness of the Happy Medium, we glimpse the danger of excessive legibility. When order is total it erases life’s joyful, organic variability.

The Giver by Lois Lowry
Jonas’s community is a closed system that achieves order by ruthlessly minimizing variety and suppressing memory, showing the trade-off between stability and resilience, and the high price of legibility and ‘optimization.’

Lord of the Flies by William Golding
A vivid portrait of social feedback loops: the breakdown of order starts small. A lack of structure and a failure of communication reinforces fear and irrationality. This, in turn, leads to further erosion of rules and increased violence, resulting in savagery.

The Neverending Story by Michael Ende
A recursive tale where the in-book reader becomes a key player in the story. The boundary between observer and system dissolves, evoking the dichotomy of “quantum vs. classical” observers without ever mentioning either.

The Thirteen Clocks by James Thurber
After an evil duke slays time itself, life and love re-emerge only when the prince restarts the clocks. The tale contrasts the rigidity of mechanical order with the resilience of living systems, where love and nonsense are critical.

Subscribe now

© 2025 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 213 — October 30th, 2025 — Available at read.fluxcollective.org/p/213

Contributors to this issue: Ade Oshineye, Wesley Beary, Erika Rice Scherpelz, Ben Mathes, Dart Lindsley, Stefano Mazzocchi, Justin Quimby, Neel Mehta, Boris Smus, MK

Additional insights from: Anthea Roberts, Jasen Robillard, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“I like big systems and I cannot lie.”

— Sir Mix a System

🛑🚪 The sticky door

If you take a wrong turn in an urban grid of two-way streets, recovery is cheap and easy. You circle the block and try again. But if you accidentally merge onto the freeway, your options quickly narrow — no U-turns, no off-ramps for miles, and a long recovery before you’re back on track.

That’s the difference between shallow and steep recovery gradients. In both cases, the move is technically reversible. But one costs you 30 easy seconds while the other might cost you 15 stressful minutes.

A valuable model for decision-making is one-way vs two-way doors. However, it’s worth adding some nuance. Many decisions are technically reversible, but the path back is long, uncertain, and full of organizational or reputational cost.

This is where considering the recovery gradient comes in handy. We ask not just “Can we reverse this?” but “How hard or expensive will it be to reverse this — and will we actually do it?”

Take a product launch. You could walk back a rushed feature, but if it damages trust, confuses customers, or exhausts your team, that recovery is uphill. Or a reorg could be theoretically reversible, but if it triggers attrition or saps morale, the reversal cost may be too high to attempt.

Subjective perception plays a huge role here. Some might see a slight slope where others see a cliff. This difference in judgment — not just about the decision, but about how it feels to recover from it — explains why teams can debate endlessly without clarity.

The first step is understanding the recovery gradient. But even better is if we go beyond classifying the decision. We can design the terrain. When the gradient is too steep, can you flatten the slope? Delay commitment? Add side paths? For instance:

• Instead of a full pivot, start a special projects team.

• Instead of reorging the entire company, make small edits to the most critical teams.

• Instead of changing your core pricing, test a new model in a limited segment.

• If exploring new markets, use market research and MVPs to test the waters.

None of these ideas are new. Instead, it’s interesting to think of them through the lens of increasing reversibility by making gradients shallower.

But beware that some decisions come with inherently steep gradients. After a layoff, you can rehire but you’ll never get the same team back. Shutting down a product can destroy customer trust and momentum. Announcing a major strategy shift can permanently change your perception in the market.

Reversibility is helpful. But recoverability can matter more. In a dynamic system, decisions should be shaped not just by possible outcomes, but by how hard/expensive it is to backtrack when the terrain surprises you.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🏪 Japanese stores are using robots remote-controlled by workers in the Philippines

Over 300 convenience stores in Tokyo have started using AI-powered robots to restock their shelves. But if the bot drops an item, a remote worker in the Philippines—armed with a VR headset—takes control and moves the robots’ limbs to pick it up. Given Japan’s aging, shrinking, and (therefore) expensive workforce, outsourcing work to Filipinos (who earn about $300 a month) can cut operating costs substantially. The robotics company is collecting data from the workers’ interventions and selling it to a startup that’s trying to build AI models for robots, with the hopes of ultimately replacing the human teleoperators with fully autonomous robots.

🚏🧾 AI-generated receipts are driving “expense fraud”

Employees at many companies can submit receipts for professional expenses to an online portal and get them reimbursed, but the makers of these expense management tools are warning that some workers are submitting AI-generated receipts in an attempt to defraud their bosses. AppZen said that 14% of “fraudulent documents” submitted for reimbursement were fake AI receipts, compared with none last year; Ramp said it had flagged over $1 million in “fraudulent invoices” in the span of 90 days. These AI-generated images are apparently quite convincing, including wrinkled paper, signatures, and items that match real restaurant menus.

🚏🥷 Ransomware profits are dropping as victims refuse to pay

A new report found that just 23% of companies who fall victim to ransomware attacks are paying the attackers, down nearly fourfold from 85% at the start of 2019. During this same period, though, the median ransom payment has grown from just a few thousand dollars to well over $100,000, indicating that attackers have shifted to targeting a few large “whale” companies rather than a broad “mid-market.” Still, the industry’s profits are decreasing as victims are wisening up and trust between the various actors (attackers, software vendors, affiliates, infrastructure providers, etc.) continues to fracture.

🚏🧑‍💼 Consulting companies’ stocks are way down due to AI and economic uncertainty

The S&P 500 index is up over 50% in the last two years, but the stocks of several major management consulting companies (including Accenture, Capgemini, and Tata Consulting Services) are down by about 30% in the same time. These firms’ bread-and-butter is offering IT and other services for far less money than it’d cost clients to do it themselves, but AI can do these same tasks for even less money, thus pushing down the prices that consultants can charge—or even removing the need for them entirely. Plus, inflation and “economic uncertainty” is leading clients to tighten their budgets, and the US government (a huge client) has cancelled “multiple billion-dollar contracts” in an attempt to save cash.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• The Useful Idiots of AI Doomsaying (The Atlantic) — Argues that AI doomers like Yudkowsky project human-like desires and motivations onto what are essentially bullshit-generating machines incapable of distinguishing truth from falsehood. The real threat isn’t hypothetical superintelligent AI but rather the acceleration of existing Moloch-like dynamics, where tech oligarchs and runaway competitive systems already pursue resource maximization without regard for human wellbeing.

• The Hidden Linguistic Messages in Brand Names (PBS Otherwords) — Examines how certain linguistic sounds and patterns have consistent interpretations across languages, from the famous bouba–kiki test to the associations people have with b/d/g sounds (“potent and luxurious”), nasals (tasty food), back vowels (large and heavy objects), and more. Then explores how brands use these patterns when naming products to position them in the market.

• Long Degeneracy: Hypergambling Eats the World (Oldcoin Bad, Newcoin Good) — Argues that wealth inequality and hyperinflation in asset prices (most notably in homes) has led young generations to believe that the “linear” path (get a steady job and save up) will no longer let them achieve their goals in in any reasonable amount of time. As a result, people are resorting to asymmetric bets so they can get a shot at wealth; think crypto, sports betting, options or buy-now-pay-later. But this is all effectively gambling, and gambling just loses you money in the long run, meaning that money will flow even more to the winners—thus perpetuating the cycle.

• First Shape Found That Can’t Pass Through Itself (Quanta Magazine) — Describes a classic geometry question (can a copy of a 3D shape pass through the same shape?) first explored by England’s Prince Rupert in the 1600s. Most objects have this “Rupert property,” so many mathematicians believed they all did—until two friends recently proved that a certain 152-faced polyhedron (which they dubbed the “Noperthedron”) can’t be passed through itself.

🔮📬 Postcard from the future

A ‘what if’ piece of speculative fiction about a possible future that could result from the systemic forces changing our world.

// 20XX. A suburban living room, before dinner.

“How was your day, kids? Did you have a good time at school?”

“Ugh... it was awful. We are doing this ‘recent history’ section and we had to watch a bunch of videos that television news crews recorded as filler that they would show as intros or filler during broadcasts. A bunch of it got left in some building after the ‘teevee’ station shut down, and they found it a few years ago.”

“Yeah, the teacher kept going on and on about how this ‘found footage’ was culturally relevant and highlighted ‘shifting cultural norms’. Mostly, it was just creepy!”

“Yeah, everyone was standing around in public doing stuff that we know is just bad for you! “

“I can’t believe how many people were just staring at their phones!”

Subscribe now

© 2025 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.

Episode 212 — October 23rd, 2025 — Available at read.fluxcollective.org/p/212

Contributors to this issue: Ade Oshineye, Erika Rice Scherpelz, Neel Mehta, Boris Smus, MK

Additional insights from: Ben Mathes, Dart Lindsley, Jasen Robillard, Justin Quimby, Lisie Lillianfeld, Robinson Eaton, Spencer Pitman, Stefano Mazzocchi, Wesley Beary, and the rest of the FLUX Collective

We’re a ragtag band of systems thinkers who have been dedicating our early mornings to finding new lenses to help you make sense of the complex world we live in. This newsletter is a collection of patterns we’ve noticed in recent weeks.

“If men use their liberty in such a way as to surrender their liberty, are they thereafter any the less slaves? If people by a plebiscite elect a man despot over them, do they remain free because the despotism was of their own making?”

— Herbert Spencer

📣💂Good definitions gatekeep

Back in February, Andrej Karpathy coined the term “vibe coding” for using LLMs to produce working code without caring about the code itself. The only requirement was that it should function, even if it was janky or mystifying under the hood. The term caught on, and soon it started getting used for any AI-assisted coding, from exploratory prototyping to meticulous software engineering.

It’s a classic case of conceptual drift. It’s natural… and often annoying to those who preferred the original scope. That’s because naming something isn’t just about communication. It’s about creating a boundary.

Good definitions gatekeep.

Not in the institutional sense — we’re not talking about standards committees. A good definition delimits. It draws a clear enough boundary around a concept so people understand what belongs and what doesn’t.

Sometimes a term is delineated by being clear. “Test-driven development” can mean many things, but much is clearly outside the boundary. Other times, it is delineated by being opaque, forcing you to verify the definition before you engage. Think of the phrase “the lethal trifecta” or “yak shaving.” You don’t know what it means until it’s explained to you.

Contrast either of these with something like “agile.” The term is immediately evocative and broadly appealing… which is also why it’s so vulnerable to drift. The word promises speed and responsiveness (and once upon a time, cachet). But ask ten teams what “agile” means and you’ll get ten different answers.

Given this, how do you choose the right kind of name for a new idea? It depends on what you’re optimizing for.

If your goal is to capture mindshare, broad and memorable can be assets. Terms like “agile,” “growth hacking,” or “AI-native” are compelling because they’re loose. People can project their own meanings onto them. Ambiguity invites participation.

If your goal is to communicate a specific mechanism, specificity wins. If you can find an unambiguous enough term, opt for clarity. “Test-driven development” is harder to misunderstand than “agile.” If not, go for narrowly scoped opacity. A term like “yak shaving” or “the lethal trifecta” may not be clear at first, but it points directly at the concept. Specificity creates tight mappings between terms and concepts.

That’s the tradeoff: drift versus adoption. The more legible and evocative your term, the more people will use it, but the less control you have over how. The more bounded your definition, the more consistent the interpretation, but the harder it is to spread.

Every definition draws a line between what the concept is and what it is not. That’s the gate. You can’t control language, but if you know your intent, you can try to point it in the right direction.

🛣️🚩 Signposts

Clues that point to where our changing world might lead us.

🚏🇮🇪 A deepfake video showed an Irish presidential dropping out of the race

An AI-generated video posted to Facebook and YouTube purported to be a clip from the popular Irish news channel RTÉ, showing a “clip” of Irish presidential candidate Catherine Connolly announcing her withdrawal from the race. The video also featured a studio scene in which a deepfaked anchor introduced the clip, and an outdoor scene in which an AI version of a well-known correspondent discussed the impact of her withdrawal, noting that the election would be “canceled” as a result. The fake clip racked up 30,000 views on Facebook before it was removed; it was also taken down on YouTube after the real RTÉ flagged it. (In the Republic of Ireland, the presidency is a primarily ceremonial office and holds little real power, but it’s still a nationally elected position.)

🚏🧺 AI startups are paying people to record themselves folding laundry

Robotics companies are keen to build AI-powered robots that can do humanlike tasks like making coffee or loading dishwashers, but they need a lot of training data to do so. So, some startups have started offering people money to record themselves doing mundane tasks like cooking dinner or folding laundry. They post ads on sites like Craigslist, saying they’ll pay $10–$25 an hour for videos of chores, but up to $150 an hour for more complex tasks like operating surgical equipment.

🚏🇨🇦 British Columbia will ban new crypto mining projects from the grid

Canada’s westernmost province has unveiled a plan to “permanently ban new cryptocurrency mining operations” from connecting to its electricity grid to “avoid overburdening” the system. (Off-grid mining setups would presumably still be allowed, though their high power demands would likely make that impractical.) BC will also limit the amount of electricity available to AI and data centers starting in January 2026.

🚏🚕 Uber will pay drivers up to $4,000 to switch to electric cars

In 2020, Uber pledged that all trips in Europe, the US, and Canada would be electric by 2030 as part of a push to go “completely carbon neutral” in those areas. But now that the US’s $7,500 federal EV tax credit—something that helped Uber drivers make the switch to electric—is expiring, Uber is stepping in to sweeten the pot. It’s offering a $4,000 grant to drivers in New York City, California, Colorado, and Massachusetts who get a new or used electric car. The choice of locations seems strategic, since some of these spots still offer state-level EV credits; Colorado and Massachusetts each offer $3,500–$6,000 rebates, and Massachusetts also offers an extra rebate for Uber, Lyft, and taxi drivers who go electric.

📖⏳ Worth your time

Some especially insightful pieces we’ve read, watched, and listened to recently.

• “Workslop” Was the Logical Outcome of Productivity Maxxing (Product Picnic) — Argues that many leaders’ simplistic desire to maximize outputs will incentivize employees to submit “workslop” (low-effort AI outputs passed off as human work). Instead of reducing the overall workload, this just pushes effort downstream to people who have to review the work. Their job is made harder by the “humanwashing” of the AI-generated slop; the reviewer can’t tell which parts of the output were generated by AI, so they have to review the whole thing painstakingly, hunting for hallucinations.

• A Post-Literate Society Is a Too-Literal Society (Paul Musgrave) — Argues that our modern society is (as many have observed) seeing a decline in literacy, but not in the “can’t physically read words” sense; rather, people are losing the ability to focus and engage deeply with a piece of media. In a world where every bit of content is designed to be consumed by audiences who are distracted or between other tasks, people expect media to be direct and repetitive; consider how many TV shows do mid-episode recaps or have actors literally explain what they’re doing out loud. In this world, subtlety, misdirection, and subtext are seen as hostile, confusing, and unnecessary.

• How AI and Wikipedia Have Sent Vulnerable Languages Into a Doom Spiral (MIT Technology Review) — Observes a vicious cycle that’s harming endangered and minority languages: AIs are trained on the limited, often error-filled online content in that language, and then AI is used to write new content in that language (such as Wikipedia articles), further poisoning the training data for that language. The only way out of the “garbage in, garbage out” trap is manual translation and a human community dedicated to preserving and growing the language.

• Silicon Valley’s Best Kept Secret: Founder Liquidity (Stefan Theard) — Examines the little-discussed practice where startup founders sell a portion of their shares during a new funding round. This enables founders to reduce the risk they take on, casting doubt on the attractive story that they are “all-in.” What this article doesn’t mention, though, are liquidity events for well-vested employees, which reduce risk for early employees as well.

🔍📆 Lens of the week

Introducing new ways to see the world and new tools to add to your mental arsenal.

This week’s lens: atomic tools.

A writer’s key tools often look like an incomprehensible mess to others. Perhaps it’s a folder with a mess of half-finished notes, interview templates, generic article outlines, and publishing checklists. No writing app could replicate it, but every piece they publish starts there.

The world is full of these sorts of highly personalized systems. Sometimes, they’re informal, like the folder of useful starting points. Sometimes they’re discrete tools, like a bash script you’ve extended over the years to handle a common workflow.

Following a term we discovered in Ep. 658 of the Changelog Interviews podcast, we call these tools atomic tools. Atom tools are tools we build for ourselves, for our own workflows. The quality standard of an atom tool is not polish or robustness. It’s fit. This tool exists to do precisely what you need, and nothing more. Atom tools tend to look rough from the outside, but they feel like extensions of mind.

By contrast, a world tool is built for others (perhaps including yourself). A world tool has to generalize. It may be used by millions or by specialists, but it is outward-facing. A world tool replaces fit with polish and generality. Our lives are saturated with world tools such as platforms, apps, and shared environments. Their prevalence teaches us to prioritize universality over personal resonance.

Atom tools remind us that we can reclaim the small, private layer of our systems. So take a look at something you have created in your life and consider how you can make it better. And look for opportunities where something general isn’t really fitting your needs and could use a little atomizing. Let yourself zoom in.

Subscribe now

© 2025 The FLUX Collective. All rights reserved. Questions? Contact flux-collective@googlegroups.com.