Innovation: Weekly Summary (October 13-19, 2025)

This week’s scatter of posts on innovation reads like a busy market. Stalls with shiny gadgets. A couple of serious think pieces hawking old recipes. A handful of people who are elbow-deep in metal, resin, or code and can’t help but tinker. I would describe them as restless. To me, it feels like the same conversation happening at different volumes — from a polite lecture hall to the noisy back bench of a pub. I’d say there’s a clear through-line: context matters, whether you’re copying a playbook, funding a grant, or shipping a weird little product from a Kickstarter page.

Old playbooks and the cost of copying

Phil McKinney’s tale of Kroger trying to ape HP’s innovation method is a good place to start. Phil McKinney writes about a team that spent 18 months trying to transplant a successful formula into a grocery chain and hit a wall. Stores cared about efficiency and margins. They cared about shelf space, not experimental theater. The team learned the obvious but often-forgotten lesson: you can’t just lift a model and expect it to behave in a new home. Culture, ops, incentives — they all matter.

I would describe that mistake as like moving grandma’s favourite recipe to a new oven and wondering why the cake burns. Same ingredients, different heat. The fix wasn’t grand theory. It was smaller and messier: stop telling, start showing. Demonstrations beat presentations. The Advantage Checkout scanning tunnel — a practical, store-level solution — only happened after the team shifted from theory to practice.

That thread shows up again in a different tone in Yue Zhao. Yue argues that leadership needs to evolve from skill-showing to systems-making. Early on, leaders may prove they’re the smartest on the team. Later, they need to build systems so others make good calls locally. That meshes with the Kroger story. It’s not just about having bright ideas. It’s about creating the context so a store clerk or a line worker can use the idea without having a PhD in innovation.

Science, institutions, and the long game

There were a lot of posts this week pushing the same point from different angles: science and institutions are the slow, heavy legs that carry a lot of fast-moving innovation.

The Nobel news — unpacked in pieces by Maia Mindel and Anton Howes — celebrated Joel Mokyr along with Philippe Aghion and Peter Howitt. The takeaway was not trophies for cleverness. It was a reminder that growth models that focus only on capital miss most of the story. Knowledge, ideas, and culture matter. Mokyr’s work, in particular, nudges us to respect history as a source of insight into how ideas travel and stick.

Read that and then check out Steve Blank yelling (in a practical, maybe gruff way) about the danger of switching off the science engine. He breaks down the roles — scientists, engineers, entrepreneurs, VCs — and says they’re not interchangeable. If you stop funding the science bench, there won’t be much left for the startup bench to turn into products. It’s obvious when you say it, but the point keeps getting overlooked when budgets tighten.

There’s a small, powerful echo in the ACX Grants piece from Scott Alexander. He lists 42 funded projects out of 654 applicants. Some of those awards went to things that feel very much like long-term science bets — genetically engineered corn, data-fraud detection tools, organ donation projects. The grant panel rejected many applications for valid reasons: conflicts, misalignment, or just not fitting the philosophy. That’s the hard guardrail that keeps grantmaking from becoming a charity bazaar. The combination of prizes and patient capital keeps the engine running.

To me, it feels like innovation here is a garden you can’t rush. You can plant a dozen different seeds, but you still have to water the right patch, know the soil, and sometimes wait years before you harvest.

Hardware is back — big machines and the small makers

There’s a surprising amount of hardware and materials innovation in these posts. It’s less shiny-app and more sweat-and-chemical-reaction.

Ashlee Vance visited Perseus Materials and came home excited about a machine that prints very large composite parts on-site. Think bridge segments or wind blades cast and cured in place instead of shipped across states. It’s not an incremental widget. It’s a re-think of how size and logistics work. I’d say it’s like turning a caravan into a small factory. You need a different checklist: chemistry, curing, weather, and, yes, messy human decisions.

Then there’s the energy stuff. Incautious Optimism writes about molten carbonate fuel cells (MCFCs) as a way to start squeezing more efficiency and less pollution out of coal-era infrastructure. The promise is big: better thermal efficiency, less waste. The hitch is practical. Ash-free feedstock. Durability. Money and regulatory systems. Still, I’ll admit, a cleaner coal plant sounds like a very 20th-century dream, but in an honest way. It’s pragmatic, not utopian.

On the other end of scale is the maker community. Matt Webb describes joining a 3D printer club and the little joys of printing foot adaptors to cut vibration. He riffs on RepRap and the weird idea of self-replicating machines. It’s charming and a bit uncanny — like watching a bonsai become a forest in slow motion.

There’s also a small crowd of craft and boutique innovators. Stefan Karlsson writes about the Woking Boutique Guitar Show: pedal makers, custom luthiers, folks who treat sound like a lab. It’s community-driven, in-person innovation. Like a town fair where everyone is selling a better mousetrap, but in this case the mousetrap sings.

And yes, the double-barreled laser flashlight from Political Calculations reads like a cool gadget — laser focus, extreme range, waterproof. It’s a Kickstarter darling. I’d say it’s the sort of thing that makes you feel like a kid again, pointing a torch at the sky, even if you’d only use it once a year.

Small and big hardware. The thread is that physical constraints still force creativity. You can’t fake materials science with a powerpoint.

AI: discovery, ethics, and the new toolset

AI kept showing up, but not only as chatbots. There were posts about AI actually making discoveries and about what people want to teach AI.

Nate pulls together a week of AI news where one of the juiciest bits is that Google used AI to generate two novel cancer-therapy hypotheses that got experimental validation. That’s not a demo of a more persuasive ad copy. That’s a machine nudging biological research in new directions. It’s the future-of-science angle that people have been squinting at for years.

There’s a practical conversation about where training data should come from. The Trichordist shares Ed Newton-Rex’s point: you don’t need to feed a model novels and pop songs to get scientific usefulness. Papers, manuals, lab notebooks — that’s already plenty. The tension here is cultural and legal. Artists don’t want their work cannibalized. Scientists want tools that can accelerate discovery. You can feel both sides tugging.

Then there’s Dave Friedman musing on Dario Amodei’s ‘synthetic nation’ idea — a country of geniuses in a data center. Friedman suggests we should be explicit about assumptions: what does success look like, and how do you test it? To me, it reads like a sober check on big visions. It’s useful like a lawyer reading a manifesto and drawing up a checklist.

On the macro side, Otakar G. Hubschmann touches on GDP implications, young tech magnates, and cultural effects of AI. There’s unease about value concentration, but also excitement about democratization of training. The hard bit is balancing the power shift — who controls the chips, who builds the models, and who benefits.

I’d say the theme is: AI is becoming a tool for science, not just a marketing trick. That changes the stakes.

Why people start companies, and what they actually want

There’s an emotional strand in the week about why people found startups. Mert Deveci colorfully calls some firms ‘fuck you’ companies. The idea is that a lot of founders want autonomy, the ability to say no to bosses, to carve their own path. Mert argues that successful startups aren’t merely better versions of old things. They are differently shaped — the Uber/Airbnb story, not Quibi.

That’s echoed in an interview with Jacob O'Bryant, who says he values invention more than the grind of scaling a business. He didn’t fail so much as reorient. I’d describe those choices as the difference between being a luthier and running a guitar factory. One makes art and tweaks. The other hires a thousand people and deals with health plans.

Robert Ambrogi reported on [Relativity Fest 2025] where CEO Phil Saunders urged legal tech to embrace discomfort and keep transforming. The message to lawyers was blunt: what you built isn’t enough. Invest in generative AI, get used to being uncomfortable. Safer bets are comfortable. Innovation rarely is.

Put these together and you get a picture of founders who want control and meaning, funders who want rigor, and customers who want real fixes, not shiny differences.

Fear, history, and the politics of technology

A couple of pieces asked why society sometimes resists tech. Jovono used the old fight over artificial ice as a mirror for modern tech fears — autonomous cars, lab-grown meat. Campaigns of fear recur. They’re often emotional, sometimes useful, sometimes flat-out anti-progress.

Nathan Knopp tied technologies of communication — the codex, printing press, the internet — to political shifts. The codex helped early Christianity escape the empire’s messaging control. The printing press broke down old gatekeepers. The internet did similar work for the modern era. These technologies aren’t neutral. They reconfigure power. That’s obvious in retrospect, but messy in the middle.

There’s also a soft UK-versus-US note. Incautious Optimism profiles the Advanced Manufacturing Research Centre (AMRC) in Rotherham as Britain’s secret Silicon Valley. The AMRC isn’t glam tech. It’s precision machining, welding, and collaborative R&D. It’s an example of institutional design that actually moves thingy from bench to plant floor. If you’re British, it reads like a local success story you’d tell over a pint.

These pieces remind you that technology is political because it rearranges who gets power. That’s always messy.

Little inventions that make life oddly better

Not everything is a grand manifesto. Some posts are tiny and human.

There’s a coffee cup made from coffee grounds (Tony). It’s biodegradable and even claims to help plants. I’d say it smells like clever upcycling: a tiny thing that nudges you to consider waste differently.

There’s also that double-barrel flashlight. And the boutique guitar show. And a bunch of Kickstarter-y hardware. These are small experiments in taste and utility. Like a new bench in the garden. You won’t solve climate change with one cup or one pedal. But these things rewire expectations. They teach people that different is possible.

Where the disagreements sit

Reading across posts, the disagreements aren’t always loud. They’re often about emphasis.

• Some pieces want more patient public investment and institutional muscle. Think Steve Blank and the Nobel write-ups. They want the long arc — labs, basic science, knowledge infrastructure.

• Others are impatient for direct, fast demos. That’s the Kroger story and the maker show. They want to tinker, to build, to ship a visible thing so people can touch it and say whether it’s useful.

• On AI, the split is practical vs ethical. Nate and Otakar push the practical payoff view: discovery, GDP, trunks of compute. The Trichordist presses the ethical angle: don’t cannibalize artists.

These aren’t irreconcilable. They’re complementary tensions you see in most innovation ecosystems. You need the lab and the demo day. You need the grantmaker and the scrappy night-shift builder. You need a few people willing to risk failure and a few institutions that can absorb long timelines.

Small patterns I kept coming back to

A few riffs kept reappearing across different posts. They’re simple, but I kept noticing them.

• Demonstration beats argument. Show me a working checkout tunnel or a cured bridge segment. People get it faster than they buy a theory.

• Institutions matter. Prizes, grants, research centres, and reputable labs change the game. The Nobel pieces and the AMRC tour both underline this.

• Hardware forces clarity. When your invention has mass and heat and ash, you quickly find the real tradeoffs. That’s not a complaint. It’s a quality control.

• People start companies for reasons that sound personal. Autonomy, identity, the desire to build something different. That personal edge matters when you measure success.

• AI is shifting roles. It’s not just an assistant. It’s a tool for discovery when paired with lab work and experiments.

A few asides and tangents (because you’ll forgive one)

I kept thinking about how these debates sound like family dinners. One cousin wants to invest in a greenhouse. Another cousin says, ‘Nah, just grow herbs on the windowsill’. A third cousin shows up with a new chest of drawers that’s actually a secret lab bench. We argue, we try things, we eat leftovers. The point is that every idea meets a real table and a real set of hands.

Also, there’s a little cultural detail: the Woking guitar show and AMRC pieces feel like two sides of the same British coin. One is the creative, grassroots community. The other is the industrial muscle. Both matter. Both make noise. Both will be in the background of the same regional economy, like tea and biscuits in the break room.

Where to go if you want to dig deeper

If any of these threads snag you, the original posts are worth the read. The Kroger story from Phil McKinney is a good, practical case study. The Nobel commentary by Maia Mindel and Anton Howes gives historical depth. The Perseus Materials report by Ashlee Vance scratches a tinker’s itch for big hardware. For AI-and-science vibes, Nate and The Trichordist give you different, useful angles. And if you like a little human colour, the guitar show from Stefan Karlsson or the 3D-print club story from Matt Webb will keep you smiling.

There’s a lot more in each post than I’ve teased here. The pieces are full of small details — experimental setups, grant criteria, machine specs, and snatches of conversation — that reward a slower read. If you take anything away, it’s this: innovation isn’t a single thing. It’s an ecosystem of labs, grants, demos, makers, leaders, and stubborn people who keep showing up. Some ideas will hatch fast. Others will need patient gardens.

If you want, wander through those posts and follow the threads that feel most alive to you. The stories will tangle and untangle in interesting ways. Some will irritate you. Some will delight you. That’s the point. The week felt like a festival of small experiments and big claims. Plenty to poke at.