Nikado − now we can move nodes up/down Sometimes you don't capture tasks in the right order… while I don't want the distraction of managing the layout, re-ordering with a keyboard shortcut is really fast & convenient 🙂⚡ Demo: nikado.app#N4Ig5g9ghgNg...

New post. www.fightforthehuman.com/cognitive-he...

In the hope of inspiring me to treat what is currently a 21,000-word blog dump more like a book in progress, I knocked up a cover and wrote a forward.

Confusing output with outcome is still one of the biggest issues in our industry. With "human communication" still in the Top 1, of course 🙃

Drowning teams in system documentation rarely works. @roundcrisis.com talk explored using bytesize sessions to share architecture insights efficiently. Watch or listen:

My imagination makes me human and makes me a fool; it gives me all the world and exiles me from it.

Every gun that is made, every warship launched, every rocket fired signifies, in the final sense, a theft from those who hunger and are not fed.” — President Dwight D. Eisenhower www.washingtonpost.com/national-sec...

I think - since corporate tech may take a while longer to come to its senses - I'll be pivoting to offering training for devs direct, which is how my business got started all those years ago. If your employer won't invest in you, maybe you'll invest in you?

To light a candle is to cast a shadow.

Engineering Leaders: Your AI Adoption Doesn’t Start With AI, Jason Gorman codemanship.wordpress.com/2026/03/17/e...

Pro Tip: Having options is a good thing, as long as they are good options.

In contrast, consider a team that is focused on preventing defects in the first place. They rigorously apply techniques like TDD, ATDD, and BDD to accomplish this. Whenever a defect does slip through, the team pauses all other work and addresses it immediately. This reflects a fundamentally […]

I wrote up a more detailed blog post about that, if you're interested: understandlegacycode.com/blog/visual-...

So there it goes: Nikado, a free visual tool for the Mikado Method You set a goal, break it into smaller tasks, and work from the leaves up The whole graph lives in the URL: No sign-up, no backend, no data stored anywhere. It's just a tool, not a SaaS. nikado.app

I built something 😄 Ever tried to make a big change in a complex codebase and ended up with a mess of half-done things everywhere? Yeah, so there's that technique called "Mikado Method" to prevent that But I couldn't find a satisfying tool to use. And paper has limits…

"All the optimism about using AI for coding, Smiley argues, comes from measuring the wrong things." www.theregister.com/2026/03/17/a...

Terminamos la última edición abierta en C# del curso Code smells & Refactoring codesai.com/cursos/refac... 🚀 ¡Muchísimas gracias a los asistentes! ❤️

Fast automated quality checks are even more essential when we're drinking from the code-generating firehose if we're to avoid testing and code review bottlenecks being overwhelmed. But just how good are those checks? codemanship.wordpress.com/2026/03/18/t...

Many predictions about AI are simply wishful thinking. "Now that AI is here, designers will have more time to focus on quality." Really? Is that what you see happening? Is that the trend, is that where the incentive structure is leading us?

... and a follow-up chapter about Subagents, now a feature of Codex and Claude Code and Gemini CLI and Mistral Vibe and OpenCode and VS Code and Cursor simonwillison.net/guides/agent...

New chapter for Agentic Engineering Patterns: I tried to distill key details of how coding agents work under the hood that are most useful to understand in order to use them effectively simonwillison.net/guides/agent...

Exploring the You Reflect pattern youtu.be/BeUP0Xyb4ik

I still care about the code, Birgitta Böckeler, 09 July 2025 martinfowler.com/articles/exp...

Engineering Leaders: Your AI Adoption Doesn’t Start With AI, Jason Gorman codemanship.wordpress.com/2026/03/17/e...

The whole thing is depressing, anybody can launder OSS through a LLM and spit out a clone they can adapt internally. Why contribute back if you can have claude vendor exactly what you need? Of course it's an open legal question whether LLM generated code can even have copyright at all...

As a not prolific OSS maintainer I’ve found the part of my brain that is like “this is nice I will release it” is now more like “why would I give my competitive advantage away for free” Or something like that The fun bit is gone. Release a gem, people chat about it, star it, contribute

We blew past this milestone without much fanfare, but it bears repeating: building awareness & goodwill by releasing open source no longer makes strategic sense for many companies. Agents increasingly consume & adapt OSS—often without users' knowledge—and cut out the creator.

Tech Debt and AI the big opportunity javil.substack.com/p/tech-debt-...

🦄 ⏰ ¡Nueva sesión de charlas en Software Crafters Barcelona! El próximo martes Jeniffer Díaz, Igor Jauk y Gabriele Brambilla nos traen dos charlazas y networking en Dow Jones. www.meetup.com/software-cra...

It's a little thing, but when you can't figure out which window you need, you need to set the window title using a`tmux` configuration setting you barely understand. Everyone starts somewhere. https://jb.rainsberger.ca/permalink/byobu-session-name-in-terminal-window-title

Before you attach that code-generating rocket to your dev process, maybe take a look at the vehicle you're strapping it to first. Can it actually take the stresses? codemanship.wordpress.com/2026/03/17/e...

"AI is exhausting workers so much, researchers have dubbed the condition ‘AI brain fry’" www.cnn.com/2026/03/13/b...

Most devs are obsessed with Technical Debt. But there’s a silent killer that’s far more expensive: Knowledge Debt. 🧵 If you don't understand how your app works, you're building a "Black Box." Here is how AI can either save you—or sink you. open.substack.com/pub/javil/p/...

Ya estamos en la distopía del conocimiento: una máquina que escribe y otra que comprime, la reflexión colectiva se convierte en una cadena de destilaciones cada vez más pobres. El debate público acaba reducido a ecos de segunda mano.

In the last year, I have been remarking more and more job ads with messages similar to this one (and worse): it looks like too many companies are feeling that they don't need to pretend that they care for the people anymore (they never did, for sure, but not even pretending is one more step) 🤡

After using the software modernization strategy overviews in workshops and training sessions, I've significantly updated the companion website: Check it out: feststelltaste.github.io/littlestmode... All materials are available under Creative Commons BY-SA 4.0. #SoftwareModernization #LegacySystem

The pivotal session that is helping me see a path forward for AI-augmented test-driven development. (Video is a bit rough 'n' ready, but packed with insights...er, once we got going...) #AIdevelopment #LLMs #agentic #TDD #softwaredevelopment youtu.be/Oz3KS9-Yohg

With examples (including test data and test results in failure feedback), if it hasn't worked on the first attempt, odds are ~50/50 it'll work by the 5th. Without examples, odds are close to nil. Basically, Claude just goes round in circles, never converging.

The completion rates across multiple passes are quite interesting too. Seen this a bunch of times, too.

I think maybe one reason I have a different perspective on LLM-assisted programming is that I don't go on "feels". I actually test it. For example, does including examples (tests, basically) with your requirements (and with feedback) actually help? I'd say that's a "yes".

Me: "Claude change the background colour of this one button" Claude: "Just let me grep all the files in the codebase"... five minutes later... Claude: "I now have a full understanding, what was the task again?"

"You don't get expertise from a book. You get it from failing and learning." If you're an engineering leader, this video is well worth 4 1/2 minutes of your time. www.youtube.com/watch?v=Zxws...

Tomorrow on Code Remix Weekly, guest hosts Kevin and Bryan from Moderne demo Prethink — the structured knowledge layer built to give AI agents the architectural understanding they’ve been missing. 🧠 🔴 Tune in tomorrow at 21:00 CET / 3 PM ET

I hope the #Java #community will do better than this…so it’s also up to 🫵🏻

I love this insight from @icooper.bsky.social - that AI tools can do coding but not programming. ian-cooper.writeas.com/coding-is-de...

If cheaper code generation with LLMs is such a competitive advantage, why haven't you outsourced and parallelized all of your product development teams all along?

We've seen spec-driven development before, in the 1990s, and it didn't catch on. Find out why not, what's changed this time around and what's still the same, in the full video on the Chiron Codex Patreon: www.patreon.com/posts/152374... #softwareengineering #aisoftware #aicoding #specdriven

Good morning, "It's just another layer of abstraction" folks! I'm working on a compiler that randomly interprets your source code wrong 30% of the time. You're gonna love it.

I'm a part of it. Not separate from it. I walk on the ground and the ground's walked on by me, I breathe the air and change it, I am entirely interconnected with the world.