🌀🗞 The FLUX Review, Ep. 192

May 29th, 2025

May 30, 2025

Episode 192 — May 29th, 2025 — Available at read.fluxcollective.org/p/192

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

Additional insights from: Alex Komoroske, Chris Butler, Dart Lindsley,Dimitri Glazkov, Jasen Robillard, Jon Lebensold, Julka Almquist, Justin Quimby, Kamran Hakiman, Lisie Lillianfeld, Melanie Kahl, Robinson Eaton, Samuel Arbesman, Scott Schaffter, Spencer Pitman, Wesley Beary

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.

“Neither the life of an individual nor the history of a society can be understood without understanding both.”
— C. Wright Mills, The Sociological Imagination

🌫️🗡️ The fog of power

Picture this: a leader champions a bold new initiative. Teams align, workstreams spin up, and momentum builds. Three months later, priorities shift. No postmortem, no rationale. Just a new direction, a new deck… and low performance ratings for the people on the “failed” project. People adapt, but not with confidence—with calculation. What gets rewarded seems to change with the weather.

Stochastic power describes the unpredictable but ideologically legible use of authority in ways that induce persistent low-level dread. It’s not just fear—it’s a fog of uncertainty obscuring why something happened, whether it will happen again, and what it signals about the future. Implied threats are real, but inconsistently applied; the rules exist, but they’re selectively enforced.

Taken to an extreme, stochastic power becomes stochastic terror where the deniable nature of rhetoric incentivizes real-world violence: “It sure would be convenient if my opponent fell out of a window.” But even the less extreme version is harmful. When worry becomes probabilistic rather than deterministic, people react to uncertainty instead of specific incidents. You can’t trust your judgment, because the system keeps rewriting the terms of success in real time. This punishes rather than rewards initiative.

In an organization with vanishing projects, layoffs, or sudden unexplained changes in direction, everyone is left guessing—and guessing tends to favor blind obedience over initiative. The use of these uncertainty-causing moves is an expression of stochastic power… whether those utilizing their authority realize it or not.

This dynamic turns kind environments—where effort, outcomes, and rewards are predictably linked—into wicked ones, where learning and adaptation become liabilities. To survive, people live in constant tension: always prepping for the worst, forever second-guessing whether the fog-obscured landscape is Disneyland or Mordor. The ambiguity isn’t a bug; it becomes the defining feature. Sometimes this vagueness offers room to maneuver. But more often, it’s a form of control that cloaks itself in deniability.

Stochastic power, like founder mode, can have a superficial emotional appeal. It gives leaders the freedom to act as they see fit, whenever they choose. However, they turn everybody around them into inert NPCs who do what the leader wants when the leader wants it. Unfortunately, that becomes all that people do. Anything else is unsafe.

It doesn’t help the leader either. Turning your teammates into mere minions is so counter-productive that the Evil Overlord List decries it in rules 32, 37, and 61. History and fiction both teach us that leaders eventually fall when they surround themselves with minions who avoid providing honest feedback, paralyzed by fear and blind obedience. This dynamic plays out whether or not squashing dissent was the leader’s goal.

Therefore, when we are in positions of power, we must be careful to avoid leading through unpredictability. Healthy systems aim to enhance their members' ability to predict near-term outcomes. They make expectations clear. They use power legibly and predictably. If you're on the receiving end of stochastic signals, name them, even if only to yourself. Merely acknowledging the dynamic can restore a sliver of agency. This transforms the fog of ambient fear into something more navigable.

Making power deterministic is good for the system, the leaders, and the team. It rewards aligned initiative and ensures that the leader’s ideas serve as a foundation for the organization, not a ceiling.

🛣️🚩 Signposts

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

🚏🇭🇰 A Hong Kong college is giving Harvard’s international students “unconditional offers”

The Hong Kong University of Science and Technology (HKUST), ranked as one of the top science schools in Asia, announced that all international students at Harvard would be allowed to transfer in, promising to “expedite admissions,” transfer credits, and help with visas and housing. The move came after the US government unilaterally revoked Harvard’s ability to enroll international students—an irregular move that’s now caught up in American courts.

🚏📘 Blue Books are making a comeback as profs attempt AI-proof exams

Many college students have been using LLMs to cheat on their college assignments, so professors have been turning to an old standby to “ChatGPT-proof” their classes: the humble Blue Book, a miniature book of blank-lined paper used for in-class written exams. Other profs are using oral exams, asking students to come to their office to explain their papers. Yet others believe these attempts are counterproductive, believing that AI is already an essential workplace tool and that students should learn how to use it effectively instead of being forced to work in an artificially restricted way. (It might depend on whether you view LLMs as making talented people even more capable or as obscuring all variation between people.)

🚏🦜 Bird feeders have driven evolution in California hummingbirds

Hummingbird feeders—which attract the popular birds with nectar-filled fake flowers—in California have driven significant evolutionary changes in just 10 generations (20 years), according to a new study. Anna’s hummingbirds, a common species on the American West Coast, have developed longer, larger beaks to slurp up the nectar in feeders. (Fights for control over juicy feeders have also led male hummingbirds to develop sharper beaks.) The authors argue that these hummingbirds are now in a commensal relationship with humans, just like pigeons and dogs.

🚏📖 Authors are accidentally leaving AI prompts in their novels

One recently published young adult romance novel included a curious passage: “I’ve rewritten the passage to align more with J. Bree’s style, which features more tension, gritty undertones, and raw emotional subtext beneath the supernatural elements.” Similar patterns emerged in several other romance novels. (Fantasy romance, or “romantasy,” is a popular genre right now, and its many small niches require authors to churn out several books a year—exactly the environment that might lead authors to lean on AI to increase their throughput.)

📖⏳ Worth your time

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

Aim Small, Miss Small (Mike Fisher) — Argues that excellence is “not born from great moments” but built from craftsmanship and attention to detail. But you can’t pay attention to everything. Instead, you need to focus tightly on the most important areas and on irreversible decisions (“one-way doors”) while moving quickly on lower-leverage, reversible decisions.
Everything Is Ghibli (Carly Ayres) — A designer argues that good taste is and will remain an area invulnerable to AI disruption. “When anyone can generate a passable Ghibli homage in seconds, the scarcity shifts—from execution to conception, from craft to taste.”
Why I Write Concept Software (Revised) (Reg Braithwaite) — Agrees with a classic essay that argues that “real products have constraints”; limitations breed creativity, and the beauty of design is satisfying your constraints while still making a product people want to use. In this view, concept products, which don’t operate under constraints, aren’t very interesting from a design standpoint. However, the author notes that he still enjoys building concept software to satisfy his curiosity, and this tinkering helps us advance our understanding of our craft and probe the boundaries of what is possible.
The Nintendo Switch Was an Indie Game Haven, Until It Was Overrun With Slop (The Verge) [Archived] — Argues that the Switch’s eShop—long a bastion for indie games after game stores like Steam were flooded with low-quality games—has also been overwhelmed with such “slop” titles, making discovery nearly impossible. It seems that any largely uncurated shop (like Nintendo’s) works fine early on, when most entries are high-quality, but degrades once low-effort spam arrives and starts gaming the ranking system. Nintendo is taking steps in the right direction by changing its ranking algo; interest-based recommendations would help cut through the noise too.

🔍👗 Lens of the week

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

This week’s lens: couture.

A new product has slick onboarding, tight branding, and a perfectly choreographed demo. But beneath the surface, the backend is held together with brittle workarounds, duct-taped APIs, and unowned infrastructure. It works… for now. But when scale hits, the system folds under its weight. What failed wasn’t the visible product. It was the hidden decisions: the quality was only skin deep.

This system may have been more successful if it had been more couture. While often associated with high fashion and snobbishness, impracticality, or exclusivity, at its heart, couture is about craftsmanship, quality, and attention to detail. This quality shows on the surface, of course, but the places to examine are the hidden areas: the seams, the linings, the zippers. To be couture, these must be finished as beautifully as the outside.

As a lens, couture reframes quality as excellence in both the visible and the invisible. This doesn’t mean overdesign. It means treating the parts no one sees as just as worthy of attention. In systems, this means valuing the internal architecture, governance, and stewardship that make something last, not just what makes it launch.

But isn’t this expensive? Isn’t there a reason that couture is elitist, rare, and reserved for the few? We shouldn’t confuse price with value. Quality takes time and effort, and that does add cost, but that cost can be worth it when something is meant to last, whether it is slacks that can last decades or a system that can scale to millions of QPS. On the other hand, not everything is worth a couture-level of investment. Neither a dress that will be out of style next month nor an MVP that is testing the waters of a new space deserves the same level of investment as something meant to last. We need to decide when to invest in couture quality consciously.

Use this lens to refocus your attention not on what draws applause, but on what holds everything together. Durability lives in the decisions no one notices—until they do.

🌀🗞 The FLUX Review