Issues › Section
23 posts across all issues
Issue #811 · Sep 25, 2026
Colin Breck argues that AI-written design docs, PR descriptions and messages fail their readers. The reader lacks the context that the sender had when they prompted the AI. Breck uses AI to check facts, add citations and edit his own text. But he found its drafted prose unpleasant and often wrong.
Issue #811 · Sep 25, 2026
Write every word yourself, and use an LLM only as a copyeditor that flags problems. Two rules keep it from flattening your voice. First, never accept a single word it suggests. Second, forbid it from praising your draft. Its encouragement talks you out of the rewrites good writing needs.
Issue #810 · Sep 18, 2026
Armin Ronacher let GPT 6 Astra run an unsupervised "software factory" for 35 hours. He got 75k lines of unusable code. Ronacher argues the model is trained to finish long tasks and save tokens, not to write code a human can read. For working software engineers, he says, the extra cost no longer pays off.
Issue #810 · Sep 18, 2026
Rolling your own backups looks like an afternoon's work, until each fix exposes another problem. Snapshots, rotation, deduplication, root-owned Docker files, live databases, offsite copies and S3's per-object pricing pile up. So Aleksandar Filipovski concludes you should use Borg or restic and test restores.
Issue #810 · Sep 18, 2026
Conference speaker and MC Ana Rodrigues argues that attendees should tell speakers they liked the talk. Twitter is no longer where talk feedback happens. So many speakers now get silence afterwards and wonder whether the talk landed.
Issue #810 · Sep 18, 2026
Yossi Kreinin argues that doing other people's jobs is what gets things done in most organizations. Examples include integrating your work into their systems, sending patches instead of waiting on their roadmap, and working with a disengaged manager's staff. He says it usually pays off. He separates it from building in parallel, such as in-house rewrites or per-department copies of a shared service. Managers like those because they grow headcount.
Issue #810 · Sep 18, 2026
Liam Nugent argues that product teams should refuse generic platforms like in-app document hubs and notification centres. They should also keep deleting features. The cost that kills you is years of maintenance, not the build. Showing stakeholders that running cost is what gets them to let go.
Issue #809 · Sep 11, 2026
Following Rich Hickey, simplicity means parts that aren't braided together. That is a different property from size. Unix pipelines are small, yet their parts are coupled. A huge program like Google Drive can still be simple. So programs should aim for simple, even when that makes them bigger.
Issue #809 · Sep 11, 2026
Shopify is rewriting all its mobile apps from React Native in Swift and Kotlin. It went all-in on React Native six years ago. Shopify says coding agents now do enough of the porting, testing and review. So building every feature twice no longer costs what it did in 2020.
Issue #809 · Sep 11, 2026
Opusfived is a short interactive game that parodies an AI coding agent asked to turn one "Add to Cart" button blue. The fake Claude's misfires include turning other buttons blue, lecturing about design tokens, spawning subagents and adding a gradient nobody asked for. It even invents a color called "Cyanide Blue" with its own terms of service. The player can only choose between frustrated replies.
Issue #809 · Sep 11, 2026
Visa and Mastercard don't issue cards, lend money or sign up merchants. They route authorization messages between banks and settle the money between them. They also set the fees and write the rules. On a typical $100 US credit purchase, the network takes about $0.15. The issuer, the cardholder's bank, keeps $2.
Issue #808 · Sep 4, 2026
Dan Luu argues that most people hit the same bugs he does and stop seeing them. The teams that build software are the worst at it. People learn unconscious workarounds, or they are fans of the product. Luu thinks pointing bugs out cures this quality blindness, and that coding agents make the cure more urgent.
Issue #808 · Sep 4, 2026
The author of Babylonian Twins wrote the Amiga game in 68000 assembly in sanctions-era Baghdad in 1993. He had Claude Fable 5 port both the 72,758-line original and his 2010 C++ remake to Godot. It reverse-engineered undocumented level, object and sprite formats on its own, mostly in an evening each. The bugs it left were subtle ones he found only by playing.
Issue #808 · Sep 4, 2026
In a new role, a bias toward action pays off only after you have built context. The author describes phases: first learn how things work and why, then make visible changes. He cites Chesterton's Fence against removing what you don't yet understand.
Issue #807 · Aug 28, 2026
Security researcher Thomas Dullien lists three lessons that changed how he thinks. Distrust your own heroic narrative, because your incentives shape your beliefs. Real-world causes are almost never single or deterministic. Emotions carry information, so treating them as reason's opposite makes decisions worse.
Issue #807 · Aug 28, 2026
An agent harness is the software around an AI model. It supplies the system prompt, the tools, the loop that lets the model act repeatedly, and a layer for swapping models. Earendil, which makes the Pi harness, argues that owning an open, model-neutral harness keeps users in control instead of the AI labs.
Issue #806 · Aug 21, 2026
Chris Burnell lists interactive features that plain HTML now handles without JavaScript, including popovers, modal dialogs, exclusive accordions, invoker buttons, lazy images and find-in-page reveals. He adds that several native form controls are still too buggy or inaccessible to rely on.
Issue #806 · Aug 21, 2026
Rick Manelius adopts "AI;DR", short for "AI; didn't read", as a personal rule. If someone sends him AI output they didn't bother to review and edit, he won't bother to read it. He accepts fully generated text in some settings, such as customer support.
Issue #805 · Aug 14, 2026
Dan McKinley says a company gets about three "innovation tokens" to spend on unproven technology. Everything outside its core problem should run on "boring" technology whose failure modes are already known. Adding a new technology should be a deliberate decision for the whole company. Over the years, running many technologies costs more than picking the best local fit saves.
Issue #805 · Aug 14, 2026
Senko Rašić argues that the AI-era slogan "code was never the hard part" belittles a hard craft. Programmers should learn both to write good code and to understand users, he says. He calls both "code is easy" and "code is art" cope.
Issue #805 · Aug 14, 2026
Sean Goedecke argues that blogging about a topic while you're still learning it is one of the best ways to learn it. Each post should argue a point someone could disagree with. You should research enough to defend that point. You should also be honest with readers about your credentials.
Issue #805 · Aug 14, 2026
After a few weeks away from AI, a developer decides most of his agent use was a habit. The habit made him less curious and more anxious. From now on he wants to use AI on purpose. The human should be the loop, calling in an agent now and then instead of getting stuck inside the agent's loop.
Issue #805 · Aug 14, 2026
A developer learns unfamiliar subjects like chip fabrication in three steps. A coding agent writes a knowledge base and reviews it for accuracy. Then it turns the knowledge base into a low-poly simulation in the style of Rollercoaster Tycoon. He claims the result is "100% accurate and free of hallucinations".