Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
Anthropic’s Scion is heavily inspired by Mononoke
A former Mononoke engineer compares Anthropic's description of 'derived data' in its new Scion version control system with how Meta's open-source Mononoke source control server used the same idea years earlier, and argues that Anthropic should have credited Mononoke as prior art.
4 min · 855 words
On the Death of Amazon’s Writing Culture
Former Amazon principal engineer Steve Huynh argues that AI-generated six-pagers are hollowing out the writing culture that once forced clear thinking at Amazon, describes 'AI;DR' and teams that stopped writing entirely, and proposes keeping human writing and reading for one-way-door decisions.
8 min · 1,826 words
Scott Helme responds to the viral 'Passkeys are a trap' post, explaining what passkey attestation actually does, why it rarely restricts users in practice, and why passkeys remain the best protection available for accounts.
4 min · 880 words
Stop Telling AI What It Can't Do
Cal Evans argues people only hand AI tasks they already know how to do — and that describing the problem and letting the conversation explore solutions unlocks more than pre-limiting what AI “can’t” do.
2 min · 426 words
Why is every rationalist blog post way too long?
Alex Wennerberg argues that rationalist bloggers such as Scott Alexander and Eliezer Yudkowsky write posts that are far too long, spending effort on layers of detail and nuance instead of questioning their premises, and says what the verbose style lacks is wisdom: the discretion to know which ideas are worth pursuing and what needs saying.
2 min · 371 words
Ben Recht looks at the result in OpenAI's math repository claiming to compute Fourier transforms faster than the FFT, with tutorial material on the FFT and computational complexity, and argues it is less a breakthrough than an exploit of theoretical computer science's habit of proving galactic algorithms that never run on a computer.
8 min · 1,732 words
AI Is a Tool. Using It Well Takes Skill.
A response to Rich Hickey's "'AI' is Not a Tool": automation and tool use can coexist, an LLM can be wielded by a human even by Hickey's own definition, and using it well requires skill, feedback, and judgment like any other tool.
2 min · 555 words
Gentoo developer Michał Górny asks Python package maintainers not to adopt uv-build as their build backend, arguing that its size and Rust toolchain requirement break builds on less common architectures and downstream distributions, while its benefits over backends like flit-core or hatchling are minimal.
3 min · 579 words
Can Ease of Creation Lead to Decrease in Quality & Value?
Gustavo Pezzi of Pikuma reflects on whether AI tools that make software and games trivially easy to produce will flood the market with cheap, generic work, drawing parallels with past shifts in other crafts and asking what will make people care about what you make.
11 min · 2,431 words
Where is all the new, good software?
Nate Meyvis lists candidate answers to the skeptic’s question—if AI coding is so powerful, why isn’t more great new software obvious—covering quieter quality gains, limited adoption, political bottlenecks, ephemeral and private tools, leisure capture, and AI sometimes reducing software use.
2 min · 511 words
Living in the Weimar era of Ruby
Julik Tarkhanov responds to DHH's RailsWorld vibe-coding keynote, arguing that celebrity disinterest in Rails does not obligate the community to abandon it, and framing a freer, post-celebrity chapter for Ruby.
13 min · 2,998 words
Just say no to OpenAI and Anthropic
Security engineer Micah Lee argues against paying OpenAI and Anthropic, citing the frontier labs' AGI ideology, and explains why open-weight models (e.g. GLM) are now good enough for complex work, with private inference options like Tinfoil and cheap ones like OpenCode Go.
7 min · 1,529 words
I'm sorry, but you still have to think
Piotr Sarnacki examines DHH's AI-agent rewrites of Campfire Once from Rails into Rust, Elixir and Go, showing how unspecified prompts turn design trade-offs like backwards compatibility, latency versus throughput and crash behaviour into coin flips, and why benchmarks and reliability still require human critical thinking.
8 min · 1,742 words
Paul Anthony Webb reacts to Deno's team moving to Cloudflare and winding down runtime development within a year, the second VC-backed tool he depends on to be handed to the community. He announces Ende, his own Deno fork that drops Node compatibility and extras like Deploy, KV and Jupyter, aiming for a December soft launch.
3 min · 735 words
Alin Tomescu surveys concerning attacks on trusted execution environments such as Intel SGX and AMD SEV, and argues that teams should reconsider how, and whether, they deploy TEE-based applications given how hard hardware security claims are to verify.
5 min · 1,057 words
Ray Myers argues mathematicians are not Luddites: AI-assisted math, built on tools like the Lean prover and the Lean FRO, had natural synergy that researchers embraced, but in his view OpenAI's hype and conduct broke the community's trust. He walks through a timeline from 2014 to October 2026 and draws parallels to how AI companies treat software engineers.
20 min · 4,573 words
Cendyne recounts being the last engineer at a company that bet on marketed LLM intelligence to replace engineering headcount, reflecting on confidence theater, M&A incentives, and what "Fable/Mythos tier" AI meant for the job.
4 min · 912 words
A short personal essay from 0xFF on leaving the corporate rat race, arguing that companies use perks to keep employees on a 'happiness treadmill' while treating them as replaceable, and pointing to remote-first, asynchronous companies like Todoist, 37signals and Linear as proof that flexible work functions fine.
2 min · 474 words
Ollie Green argues that after AI companies got away with large-scale IP use for training, AI's ability to decompile, understand and recompose software and other knowledge work is now driving the value of prior work toward zero, collapsing the scarcity that intellectual property was built to protect.
2 min · 442 words
Rich Whitehouse argues that "artificial intelligence" is a poor label for frontier generative models, criticizes large language models and the companies building them on grounds including copyright, labor and societal harm, and urges readers to stay aware of what they are participating in.
14 min · 3,304 words