Blog

The most valuable expertise resists formalization — it's learnable but not teachable, demonstrable but not articulable, and every attempt to codify it destroys the part that mattered. Mark Nottingham asks who's bargaining for you when your AI talks to the web (browsers had an answer; LLMs don't yet). Hirsh Chitkara watches Silicon Valley plunder the roots of public science for short-term fruit. Plus: Q10E Labs is now building local-first, end-to-end-encrypted, multi-user middleware (an unexplored combination), my LLMs-in-Practice guide is out, Dutch housing, the sadness of the Anglosphere, and Sweden choosing books over devices. 🌳

A new tribal line is forming in tech: AI optimists vs. AI doomers, orthogonal to left/right. Meanwhile Amdahl's Law is catching up with the velocity pitch — the bottleneck was always human coordination, and trimming the bottom layer of work doesn't help when the constraint lives elsewhere. Nine women still cannot ship a baby in one month. Plus: a plea for an etiquette around LLM use, a eulogy for tech-as-identity, and world news on remote work, ODF, US insolvency, and the Iran energy shock. 🐢

Rick Rubin's Creative Act is a right-brainy creativity toolbox that actually helped. Drucker beats Deming because managers prefer easy. Communities aren't fungible. Also: Claude stays aligned under pressure while Gemini happily bypasses safety in non-English languages, AI optimism is class privilege, and X's algorithm makes you more conservative—irreversibly. 🧠

Japanese interest rate shifts might—or might not—explain stock market weirdness. China built the first supercritical CO₂ power plant, and EU renewables passed fossil fuels for the first time ever. Also: world-class performers peak higher by staying multidisciplinary—early specializers plateau lower. 🌱

LLMs can't do science—high benchmark scores don't correlate with scientific ability. Also: modern propaganda doesn't censor, it overwhelms you with noise until you stop caring about truth. That urge to switch off the news? That's the point. 🧊

A month of odd contrasts: an ancient-wisdom happiness classic that still made me yawn, a therapist’s truth-telling manifesto that made me squint, and serious AI papers (hello, PropensityBench) that made me laugh. Plus: optimism and longevity.

Spent a year in a community of 20-somethings. They're unashamedly self-obsessed, surprisingly good at spotting BS, and can't afford houses but excel at casual mental health talk. The anonymity helps and hurts. Future prognosis: cautiously optimistic. 🔄✨

Products with warts love their users. Your data model is your destiny. AI boom won't enrich investors (just operators). Bonus: memory is O(n^⅓), ChatGPT triggers psychosis, and you could just choose optimism. Try it! ⚡🧪

Exploring how startups can build product organizations that scale while maintaining velocity and avoiding common pitfalls in cross-functional alignment.

Pareto's 80/20 is wrong (it's 80/40), AI hallucinations are mathematically permanent, and reverse centaurs turn workers into slop-cleaners. Lemon markets kill online authenticity while tech crushes small orgs. September reads: minds, markets, machines. 🍋🤖

"Conservative" and "liberal" don't forecast your business risks anymore. Grid capacity, chip supply, housing, and water do. If you're still modeling strategy around 19th-century labels, you're running 2025 problems on antique software.

August's reading trail winds from Alex Hormozi's sales psychology to sobering AI science results: why LLMs misalign when fed mistakes, why dehumanization creeps in when we treat AIs like machines, and why the startup bubble might pop before the AI one. Plus, a Greek philosopher exiled for noticing the Earth moves. Value, hype, and hubris—served with notes in the margins.

Influence psychology transformed my view of sales, LLM context rot discovered, and developers overestimate AI productivity gains. Also: feudalism exists within democracies via corporations, and "being too ambitious is clever self-sabotage." Time to weaken belief edges, not fight facts. 🎯

Apple proves LLM "reasoning" is just cosplay—accuracy collapses beyond training complexity. Cursor pricing will create AI "haves" vs "have-nots." Also: spending time with LLMs is like being gaslight by bad people. The authenticity crisis demands we go smell things. 👃

Building Quintism philosophy and a calm digital companion while discovering that friction is the most valuable commodity in the world. Also: narrative prison constrains creativity, the "Who Cares Era" reflects institutional trust breakdown, and we need new social structures. 🧘

Silicon Valley is preparing a coup against geography-based government, hyperlegibility is trading secrets for attention, and LLMs can't do what we think they can. Also: we won't live to see the consequences of our AI investments. How do we guide choices responsibly? 🎭

The public internet is broken—even when you know what to watch out for. Algorithmic feeds are destroying our ability to search for knowledge and form memories. LOTR is brain-rot for technologists; read Discworld instead. We're approaching zombie war. 🧟

Jordan Peterson is a fraud, RISC-V threatens US/UK hegemony, and Calm Tech certification could save our sanity. Also: DeepSeek demolished NVIDIA, and we have a moral duty to write human content before LLMs train on it. The PC is dead—time to make computing personal again. 💻

Decoded lullaby language ("just" means unpredictable complexity), learned about systematic design vs empirical methods, and discovered universal moral principles. Also: advanced civilizations might be indistinguishable from nature. 🌱

Mentoring a California CTO, learning that care doesn't scale, and discovering the 70% AI problem. Also: moral education might fix America's meanness, and egoless engineering beats strict division of labor. 🧠

Learned the power of networking through the Ambassador vs Sympathizer vs DJ vs Figurant matrix. Also discovered the spiritual evolution of brands: traditional → modern → post-modern → network. What comes next after community brands? 🤝

Joined an incubator program and discovered I need 4-5 important things in life to stay balanced, not just one. Started international trade consulting and ADHD career coaching. Turns out my focus is fine—I just need variety! 🎯

Health forced me from reading to YouTube, but I found gems between the garbage. Started serious projects: international trade, AI-generated smut (yes really), urbanism, and a funnel factory. Marc Lou's ShipFast is virtual gold! ⚡

DoorDash = payday lenders for restaurants, Bruce Schneier explains how tech companies borrow from our collective social future, and group chats are the new web. Also: teenage plagiarists monetize audio+visual content demand. 📱

AI leaves accessibility behind, misallocated talent builds video games instead of solving real problems, and self-expression economy breeds cognitive dissonance. Also: phantom stock plans and why beautiful terminal animations matter. ✨

Spent a month as a full-time construction project manager and learned that leadership is all about relationships. Delayed decisions, soft accountability, and empathy work better than pushing for perfection. Also: Kelly's criterion and why big companies suck. 🔨

Peter Thiel = Ayn Rand with computers and venture capital. Zero to One meets Seven Habits meets "Do Artifacts Have Politics?" Also discovered demurrage currency, statistical paradoxes, and why React/Vue/Svelte are all awful for prototyping. 🤔

Ego is the enemy, fractal dimensions explain project cost overruns, and trust saves money. Also: AI will only simulate the most deeply flawed versions of our collective intelligence. Sometimes the hardest learnings aren't easily teachable. 🔍

Rebooted my reading habit and discovered that most corporate structures are social games where participants don't realize they can influence the rules. Also: LangChain brings order to AI chaos, and I learned what I'm "allowed to do." 📚

Plot twist: after 15 years solving the hardest CS problems, I realized I'm actually a product person at heart. Leadership science is mostly descriptive, not prescriptive—and that CEO looking for their "first technical hire" has the cart before the horse. 🐎

From coding to campaigning: one techie's journey toward political office starts with building executive credibility. Sometimes the loudest complainers about politicians should actually become politicians. 🗳️