Ordinary engineers, not heroic inventors ⊗ Umyazu

Like many people, I’m getting pretty tired of AI over here, as you might have noticed by the real header photos over the last few weeks. I’m not having enough fun with the image-making vs other issues and backlash. Still using for other things though, but I’m already tired of the relentless pace of “innovation” and bs. However, tech is still a core component of my interests, and this newsletter. So yes, there will still be articles about “new forms of hyper-quantitative data” (yes, sometimes newsletter issues and blog posts compost deep in a bucket, but I finally got to this one from a year ago) but I’m trying to reduce them a bit and focus on pieces about the deeper future, are more conceptual, critical, contrarian, or which paint a different set of possibilities. Last week’s AI is not conscious, but it is becoming our unconscious was one such take, and Tim O’Reilly’s piece below also fits the bill.

Ordinary engineers, not heroic inventors

Tim O’Reilly builds his framework on Jeff Ding’s book Technology and the Rise of Great Powers, which sets out two competing theories of how new technologies reshape economic power. He leaned in more towards application in companies than geopolitics and, what I would have liked, comparison with China, but still a good read. I angled my summary to the bits that are less “corporate tips.” The default assumption, in AI and in the industrial revolutions before it, is the “leading sector” theory: whoever invents and dominates the fast-growing new industry captures the profits and wins the era. England did it with steam, the US with electricity, and then Japan tested this directly. It led the world in semiconductors and consumer electronics through the 1980s and still lost the information revolution to the United States, which pulled ahead not through hardware but by diffusing computing into every sector of the economy. Ding’s alternative, “diffusion theory,” thus argues that national leadership in a general-purpose technology comes from embedding it broadly across ordinary economic activity, over decades, through what he calls skill infrastructure—the education and training systems that widen the pool of people who can use it. As Ding puts it, “it is ordinary engineers, not heroic inventors, who matter.” O’Reilly reads today’s AI industry, fixated on frontier labs and flagship models, as running on leading sector logic, and argues the same diffusion dynamic will decide which companies win.

That diffusion is slow because it’s an organisational problem, not a technical one. O’Reilly invokes Paul David’s history of electrification: factories that bolted a giant electric motor where the steam engine used to be, driving the same belts and shafts as before, saw almost no productivity gain. The payoff came decades later, once engineers redesigned plants around many small motors, each driving its own machine, on a floor laid out for the actual flow of work. He draws the same lesson for AI, sketching a future built less around ever-larger centralised models and more around a decentralised network of smaller, often open source, models sized to specific tasks—with the real bottleneck being accumulated, person-by-person know-how rather than access to the technology itself. From there he pulls in Ethan Mollick and Trail of Bits’ Dan Guido for a set of concrete moves companies can make: standardising the toolchain, writing down clear rules instead of vague warnings, building a capability ladder, running adoption sprints, and fixing the incentives so that sharing a workflow raises someone’s status rather than making them look replaceable.

Swinging back to the geopolitical thread Ding’s book opens, O’Reilly reframes “sovereign AI,” usually read as a race for national frontier capability, as instead a natural consequence of diffusion: a technology that spreads widely gets adapted by different societies and institutions to their own needs, and that adaptation is the point, not a defect to be engineered away. He warns against the arms-race framing of AI as a scarce strategic weapon, and against the instinct of some labs to homogenise the technology into one sanitised, universal system. The electrification analogy cuts both ways here too: motors decentralised, but power generation consolidated onto a central grid—a pattern now straining under the load of AI data centres, and one O’Reilly thinks the AI world should avoid repeating. His proposed middle path is standards and interoperability, on the model of railroad gauge, where incompatible systems produce decades of friction rather than healthy diversity. Open source matters less for its licensing, in this reading, than for the protocols and shared interfaces that let different systems, proprietary and open alike, interoperate without becoming identical.

Ding’s argument is that the breakthrough sector itself is not where the long-term value for national power lives. And I believe that the same applies to corporate success. The value is in how widely and how well the technology gets embedded into the work of the people you already employ. […]

In “The End of Programming as We Know It,” I made the case that AI expands who can build rather than replacing the people who build today. This means that a company’s best source of applied R&D is the everyday experimentation of the people it already has. The job is to make that experimentation visible, shareable, and rewarded. […]

The constraint on enterprise AI adoption is not just the raw skill of the people. It is whether the organization has built incentives under which sharing what you learn raises your status rather than lowering it. Get that right and diffusion follows on its own. Get it wrong and you can have a small kernel of great people leveraging every frontier model on the market while adoption stalls out at a small fraction of your workforce. […]

Open source AI depends on far more than open models. It depends on the architecture of participation built into the systems above and around them: the protocols, servers, interfaces, and shared technical conventions that let many different actors build on common foundations.

Umyazu

Mandy Brown opens with a claim: the attention economy is misnamed. What gets harvested isn’t attention, she writes, but its opposite. Doomscrolling and feed-flicking as are kind of self-abandonment, the deadening of a capacity that reading, by contrast, awakens. Reading, for her, is an act of vulnerability: it opens a person to being changed, to shedding an old self and trying on a new one. This stands against the vigilance the stream demands, where we go in braced and wary, worn down by the very alertness that’s supposed to protect us. The etymology of “paying” attention matters here too, tracing back to an older sense of appeasement: the endless refreshing and reacting reads as tribute paid to what Brown calls the angry gods of capitalism, an exchange framed as a political project rather than a personal failing. An attention-starved public can’t intercede in the world—it’s left to receive somebody else’s version of it instead.

Against this, she reaches for Ursula K. Le Guin’s novel The Telling, where a Terran observer named Sutty arrives on a planet mid-industrialization, its literary culture banned and its people reduced to “producer-consumers.” In a mountain village she finds the maz, teachers who keep alive an oral practice called the Telling, paid for directly and honestly, coin for story, with no shame or manipulation in the transaction. Brown holds this up as a model of a “cash, not credit” economy of attention, one where payment and reward are immediate and unexploitative, against which the platforms of the current attention economy, that pay twice: once in time, once in the erosion of the self that gave it. She closes by arguing that something like the umyazu—the places where the Telling happened—already exists, dispersed through worker-owned presses, zines, libraries, (newsletters, wink, wink) and every reader who takes up a book instead of a feed. Every dollar and every gaze redirected there, she argues, isn’t just a transaction outside the attention economy, but an active refusal of it.

Nor do we read when we slip through the stream or flick through the feed. Reading is an awakening of attention, not a deadening of it. We read to come alive to ourselves, not to forget who we are or what we are doing, or what is being done to us without our consent. […]

This narrative completely obscures the fact that annihilating attention is a political project with clear benefits for the billionaire class: if we cannot attend the world, neither can we intercede in it. We become passive recipients of their worldbuilding, disenfranchised from our own responsibility to make sense of—and therefore to remake—the world around us. […]

It is the skill of reading that hones that curiosity, sharpens our ability to notice what is before us, what is real and what is not, which bargains are fair and which are usurious. Reading is how we attend the world, which is also how we change it. […]

Every contribution to the reading economy is two less for its nemesis: once in cash, a second time in the resurrection of attention, in the art of reading, in the gift paid in return for the gift. A fair bargain, and payment on the spot

§ Stacking spaces. Another one from a little while ago (see intro). Love it! “ tl;dr: Hidden behind an unmarked doorway in Barcelona, two coffee businesses share the same small space — one roasting and selling, the other teaching and tasting — but not at the same time. This smart setup, which I called a stacked space, cuts costs, spreads downside-exposure, and promotes urban experimentation and innovation. It’s an emerging pattern you can now spot in Singapore, Tokyo, Manhattan, and Paris … and a kind of urban innovation planners and developers haven’t caught onto yet, but should.”

§ Why Leonardo was a saboteur, Gutenberg went broke, and Florence was weird. I wouldn’t call this a great interview, as Patel was, for my taste anyway, trying too hard to make connections with today’s entrepreneur vibes. However, Ada Palmer was brilliant, erudite, absolutely on top of all her topics, and you can almost watch this as a lecture more than an interview. Also, why do so many tech guys talk at 1.5x speed all the time? You’re not a zip file, no need to compress your words! I’ve only just started, but there’s also a follow-up: Machiavelli is the most misunderstood thinker of all time.

Futures, Fictions & Fabulations

• Futuring through discomfort. “One Good Ancestor example was Anab Jain’s talk from Superflux at Future Days. She recounted a Gulf urban project where decision-makers were resisting cutting car traffic. So her team reproduced the exact air their children would breathe in that future, made it in a lab, and let them sit in it until they changed their minds. And they did!”
• Designing the Future: Lessons from the Symbiocene at Melbourne Design Week. “The Symbiocene proposes a transition from the human-dominated Anthropocene to a future where culture, technology, and the economy operate in reciprocal, sustainable relationships with the living world.”

Algorithms, Automations & Augmentations

• ArtIA: imagining a new digital commons for creative AI. “While American screenwriters were resisting AI capture, a vibrant community of digital creatives and arts organizations across Québec and Canada formed the alliance ArtIA. ArtIA advances commons-based alternatives to cultural AI, in response to the observation that while AI is rapidly transforming creative tools and workflows, the cultural sector lacks the critical resources, reference frameworks, support systems, and training needed to navigate this shift.”
• Human-centered Large Language Models. “Large language models (LLM) have moved from research laboratories into the infrastructure of everyday life. They power everything from developer tools to educational tutors, healthcare assistants to enterprise agents. Yet the frameworks guiding LLM development remain anchored in technical performance metrics that tell us little about whether these systems actually benefit the people who use them.”
• How AI is changing language. “As allegations of LLM use rock the literary and media worlds, linguists explain what really distinguishes human and machine writing, while novelists including Jennifer Egan and Jeanette Winterson reflect on the future of fiction in an age of ChatGPT”

Built, Biosphere & Breakthroughs

• Project Circle (post by Dan Hill). A new research project, following last years’s Australian Reduction Roadmap, proposing a new framework for making better choices about materials, oriented towards a regenerative built environment sector that works well within planetary boundaries” and #108 Project Circle Design Phase 1 (the project’s first publication).
• Air quality is improving across Europe as report finds “steady decrease” in major pollutants. “The EU’s Copernicus Atmosphere Monitoring Service (CAMS) says in its latest report that while some parts of the continent are still experiencing issues with local air pollution the overall outlook ‘remains encouraging’.”
• China’s gravity battery tower stores wind power. “Surplus wind power hoists 35-ton blocks cast from recycled concrete and industrial aggregate toward the top. When the grid needs electricity again, the blocks descend and spin the generators. The tower stores 100 megawatt-hours and can deliver 25 megawatts for about four hours, at a targeted round-trip efficiency above 80 percent, with a 35-year lifespan and no chemical degradation.”

Asides

• Umberto Eco library opens in Bologna, 10 years after his death. “Before the move, the library was surveyed shelf by shelf in Milan, with the position of every volume, thematic groupings and connections between authors and disciplines carefully documented, so that the new premises in Bologna could reproduce Eco’s original arrangement exactly, down to which books he kept lying flat and which stood upright.”
• Dune: Part Three, official trailer. “Set nearly two decades after Paul Atreides seized control of the Imperium. Now a ruthless Emperor, Paul must face the consequences of his reign as old allies return, terrifying new threats emerge, and betrayal lurks in every shadow.”
• “Most famous tree in the world”: Sherwood Forest’s 1,000-year-old Major oak dies. “The huge tree, which has grown in Sherwood Forest in Nottinghamshire, England, for at least 1,000 years, failed to produce any leaves this year, after becoming stressed by a series of hot, dry summers.”