Innovation: Weekly Summary (November 03-9, 2025)

The week’s blog pile about “innovation” felt like a messy toolbox hauled into the living room. Some tools were shiny. Some were rusty. Some folks argued about who gets to boss the job site. I would describe them as part-state, part-startup, part-garage tinkering — all talking past each other a little, and sometimes agreeing by accident.

Capital, coordination, and who pays for big bets

There were a few posts that kept circling the same question: who should write the checks for big, risky tech? Lawrence Lundy-Bryan sketches it like an old movie. After World War II, the state ran point. Then, from about 1980 to 2020, venture capital took over. Now the state is coming back. I’d say Lawrence makes the mood of the week clear: capital is not just money. It’s coordination, timing, and sometimes a big country deciding it wants an industry.

That theme bounces off of the Europe/China comparison in the policy brief from Naked Capitalism. Their piece asks a plain question: if China is sprinting on chips and AI, can Europe even lace up its shoes? They argue Europe is fragmented and slow, while China is focused and deliberate. To me, it feels like watching two football teams: one with a clear playbook and a coach who yells, and another where every club shows up with its own ball and a different rulebook. Read their piece if you like policy-level eye-rolling and real policy checklists.

There’s also a more playful take from Robert Yaman who basically says: give the DARPA model to farms. “DARPA for Chickens” is short, concrete, and a bit joyful. It’s the idea that a small, weird agency that funds high-risk, high-reward projects can fix real bottlenecks. This ties back to the capital question. Sometimes money needs a structure to make it useful. The DARPA model is one such structure. If money is the gasoline, DARPA is the ignition system.

A little further down the stack, Jonny Evans notes a quieter kind of coordination. Corning and Ensurge teamed up on microbatteries. That’s not headline politics. It’s not sexy like semiconductors. But it’s the kind of cooperation that changes product timelines — and that matters. It’s the kind of small bet that only looks boring until a wearable device suddenly gets useful battery life. That’s practical capital allocation. It’s patient, slow, a bit like putting up drywall in the rain. Not glamorous, but needed.

If you want the broad strokes, Lawrence and Naked Capitalism give them. If you want the micro, Corning’s deal is a good read.

State action vs market hustle: the old fight, new clothes

Several posts examine the tug-of-war between state-led coordination and market chaos. Jamie Lord writes about China’s J-36 stealth fighter being redesigned in ten months. It’s presented as a muscle-flex — rapid iteration versus Western programslike the F-22 and F-35 that took decades. I’d say it reads like a parable about two ways to build things: iterate fast in the real world, or precompute everything and hope the simulation is right.

That parable shows up again in the piece about US sanctions on China’s AI supply chain. Dave Friedman argues the sanctions backfired. Instead of slowing China, the pressure made labs more efficient and creative. There’s a pattern here: when one route is blocked, another route opens. To me, it feels like highway construction. Close one lane and the drivers find the next street and sometimes end up discovering a shortcut.

This is the same knot that Lawrence ties to megafunds and government. Big, strategic sectors like AI and chips are not markets that sort themselves out neatly. They need coordination — policy nudges, large capital pools, special institutions. The posts do not all agree on the right mix, but they do agree on the problem. There’s too much hand-wringing to be comfortable and not enough clear recipe to be smug.

The machinery of making things: chips, light, glass, batteries

Hardware posts this week were quiet but deep. Vikram Sekar dug into Substrate’s X-ray lithography claim. There’s a lot to like about the post if one enjoys a technical spine. Substrate says X-ray lithography could be cheaper and faster than EUV. Vikram says: cool idea, but here’s the skepticism checklist. Lithography is not one trick. It’s thousands of tiny tricks.

This piece pairs well with the Corning/Ensurge short from Jonny Evans. One is about printing features at atomic scales. The other is about squeezing more energy into a tiny package. Both are about overcoming physics and manufacturing inertia. It’s like two cousins in a family who both want the same old heater fixed. One wants a new nozzle; the other wants the thermostat replaced. Both matters.

There was also a different kind of hardware nostalgia from ObsoleteSony. The Sony XEL-1 OLED TV from 2007 is a classic case of being ahead of the market and then waiting for the market to catch up. The XEL-1 was small, outrageously priced, and brilliant in tech show terms, but not a mass product. That’s the slow burn of innovation. Sometimes being first just means you learn faster. Sony’s later work shows the payoff — but only if you’re patient or have deep pockets.

Reading these together, the theme is clear. Hardware is expensive and unforgiving. New manufacturing tech needs patient capital and coordination. That’s why the capital and state conversations matter.

AI: hype, limits, and market shaping

AI shows up in many flavors. Some posts are optimistic. Some are skeptical. Some are annoyed.

Nominal News writes about the Nobel in Economics given to Aghion, Howitt, and Mokyr. This isn’t party-line AI hype. It’s a reminder that innovation is often a process of creative destruction. Firms race to invent. Sometimes they win monopoly rents. Sometimes they don’t. The model helps explain today’s genAI rush. Investment is pouring in because the payoff could be massive. The post is a tidy bridge between old-school growth theory and new AI money.

Then there’s the old-you’re-doing-it-wrong tone from loganssite@loganmarek.com (listed as Logan’s Site in the dataset). The post titled “If You Don't Use AI At All, I Will Poke Fun at You” is cheeky. It’s the kind of nudge that’s half friendly banter and half peer pressure. The argument is simple: try small, useful tools or get left doing boring tasks. It’s practical. It’s the “use a wrench, don’t look for a magic hammer” school of thought.

On the flip side, Pawel Brodzinski wrote “AI Won't Generate a Good Product Idea.” That’s a darker counterpoint. The claim is that language models spit out statistically likely phrases. They’re great for drafts and scaffolding. They’re not Oracles of Genius. They don’t sit in a cramped workshop and come up with the one bonkers tweak that changes everything. Pawel’s point is blunt and useful: AI helps you iterate, not ideate from scratch. The two posts together are like siblings arguing at the dinner table: one saying eat your greens, the other saying don’t expect vegetables to become steak.

There’s also a short, sharp bit from Simon Willison quoting Nathan Lambert about the sudden rise of Chinese AI labs like DeepSeek, Qwen, and Kimi. The tone is more observational. It says: talent and focus can move things quickly. Pair that with the piece by Dave Friedman about sanctions, and a picture forms: constraints can concentrate effort and sometimes produce nimble, specialized innovators.

A different take comes from Ben's Blog on “AI Market Makers.” This thread imagines startups that remove friction. Think of someone who builds a bot that makes it dead simple to sell old furniture or to get government forms filled. These are not headline-grabbing moonshots. They’re useful. They could win markets quietly. I would describe them as the kind of businesses that make life less grumpy.

Finally, Conrad Gray gives a sort of industry snapshot. OpenAI is focused on growth. Anthropic is aiming for revenue. Big cloud partners are deepening ties. It reads like a trade magazine where everyone’s building the scaffolding for the next phase. Lots of cash. Lots of infrastructure. Lots of runway but not much guarantee of a graceful landing.

Put these together and the trend is obvious: AI is both a hammer and a mortar. It’s changing workflows. It’s not replacing deep domain creativity. It’s shifting market models. And geopolitics is pushing labs into strange, sometimes smarter corners.

Speed, iteration, and the human cost of fast

Some posts asked what speed actually means. Jamie Lord shows China’s fighter program as a case study. They changed major components in ten months. That's dizzying. It looks like innovation as a sprint of prototypes. Western programs look like marathons with planning committees and safety harnesses.

There’s an obvious tradeoff here. Fast iteration can mean learning in the air. But it can also mean cutting corners. The post doesn’t fully side with either. It just observes. It leaves a sense that different political economies support different rhythms of innovation. One is iterative shock therapy. The other prefers long vetting.

Hilarius Bookbinder wrote something a bit different. It’s a reflection on measurement and perspective. The main idea is that short-term thinking hides big cultural changes. He calls the current scene a “Cambrian Explosion” of culture. It’s a nice phrase because it captures the messy, rapid branching of ideas and tools that don’t register when you measure everything with a too-small ruler. It’s a call to widen the lens and stop obsessing about the quarter.

These posts together remind that speed isn’t a moral good. It’s a trade. Fast can be ingenious or wasteful. Slow can be careful or paralyzing. The challenge is finding the right tempo for the right problem.

Manufacturing myths and startup bravado

Startup claims are a recurring motif. Vikram Sekar is cautious about Substrate. Substrate says X-ray lithography will beat EUV. Vikram says: homey, hold your horses. The post lays out the hard parts — resist chemistry, overlay accuracy, throughput. It’s a good, skeptical read for anyone who likes the tension between booster optimism and practical friction.

That skepticism shows up elsewhere too. The Sony story is one of early fame that slowly turns into real influence. The XEL-1 did not make Sony rich right away. But it left engineering lessons that mattered later. It’s the “loser at the trade fair who later trains champions” story.

And then there’s the policy side where some writers gently accuse Western programs of being too risk-averse and others warn against the dangers of copycat industrial policy. There’s no single villain. Just a messy ecosystem where startups promise the moon and veterans whisper about supply chains.

Small things that matter: microbets, public goods, and weird niches

A number of posts emphasized small but vital bets. Corning + Ensurge is one. DARPA-style agriculture funding is another. Ben's Blog mentioning market makers suggests many useful businesses remain unbuilt. These are the bets that don’t show up in headlines. But they change how people use technology.

Daniel Debow appears in a transcript published by Tara Henley. He talks about Canada needing “bold adventurism.” That’s a nice phrase. It’s not the same as reckless spending. It’s about trying more things and having a tolerance for failure. It connects to DARPA and the microbatteries: small, bold experiments can create options.

There’s also an odd little story in the list: an author describing culture as a Cambrian explosion. That’s not policy. It’s a reminder that lots of seemingly small cultural experiments can add up.

Points of agreement and the bits people keep arguing about

A few things almost everyone agrees on, even if they squabble over the solution:

• Big strategic technologies need more than one kind of funding. Whether that’s DARPA-style prizes, megafunds, or active industrial policy, the consensus is that the market won’t always sort it out by itself. See Lawrence Lundy-Bryan and Naked Capitalism for the clearest takes.

• Constraints breed creativity. This came up again and again. Whether it’s chip sanctions pushing Chinese engineers to write leaner code (Dave Friedman) or startups iterating on hardware prototypes, pressure often leads to surprisingly useful work.

• AI is useful but not magical. Pawel Brodzinski and Logan are not enemies. One says don’t expect AI to invent your product for you; the other says don’t ignore AI when it can do the boring stuff. Together they make a neat practical maxim: use AI where it helps, and don’t outsource your soul to it.

The big arguments are over method. People disagree about whether state-led pushes are the right fix for Western slowness. Some want big industrial policy. Others fear copying tactics that might work in China but clash with liberal market norms. There’s no silver bullet. The debate is messy and a bit like watching relatives fight over how to get a stubborn lawnmower started.

Little tangents and odd takeaways

• There’s a subplot where nostalgia meets reality. Sony’s XEL-1 is a reminder that sometimes being first is a learning exercise. The market catches up later. Feels like an old vinyl record found in the attic that becomes the new sampling gold.

• The tone of some posts is cheeky. The “I will poke fun at you” piece is one of those friendly shoves. It’s annoying in a useful way. It’s like a coworker who insists you use a calendar app and then gives a high-five when you finally do.

• A recurring metaphor: innovation as cooking. Some kitchens are Michelin-perfect labs. Others are food-truck chaos where good food happens fast. Both feed people. Both have different rules. This keeps cropping up. Maybe because making things is literally like following recipes that sometimes need improvisation.

If any single idea wanders through most of the week’s posts, it is this: innovation is social. It’s not just the clever engineer in a basement writing code. It’s funds, rules, supply chains, government decisions, tacit knowledge, and sometimes dumb luck.

If one wants to dig further, the original posts are worth a read. They’re the kind of thing that rewards slow reading — the more time spent, the more little details pop out.

A last little note: the mix of optimism and caution in these pieces feels familiar. It’s like watching a city slowly build a bridge. Some folks clap for the cranes. Some folks question whether the foundation was poured right. Both sets of voices matter. Both keep the project honest. And on the ground, building goes on, sometimes noisy, sometimes quietly. Read the posts if the noise and the quiet both interest you.