

Thanks for sharing such a personal story, Repair Cafés are a great idea, especially nowadays when “old tech” is still perfectly usable.
Curious: what were the biggest challenges you faced when setting one up?
Engineer, 20+ years of project management, now tangled up in change management and AI. I share the mistakes behind the lessons: what I did, what broke, and the exact Ctrl+Z moment.
Blog: https://www.theundostack.org/ · Mastodon: @ctrlzee@mastodon.social


Thanks for sharing such a personal story, Repair Cafés are a great idea, especially nowadays when “old tech” is still perfectly usable.
Curious: what were the biggest challenges you faced when setting one up?


Same here. Vibecoding my own tools is fine, because I’m the one who pays for the bugs. A kernel patch is different: thousands of people pay for them, so understanding every line is the minimum.
Anyone who sends code to public software should follow a simple code of ethics. A low barrier to entry is great, it’s how most of us started. But the bar for responsibility should rise with how many people run your code, and right now it doesn’t.


It’s an unfortunate news, burnout in open source was real even before AI. Now it has just exploded.
Reviewing code has always been a bottleneck. AI made sending a patch nearly free, but reviewing one costs the same as ever, and it lands on a few volunteers.
For code that millions run, someone has to understand every line, so the sender should do that work, not the maintainer. Otherwise you’re just moving the cost onto the people who can least afford it.
There should be a simple rule: if you can’t explain your patch line by line, don’t send it. And maintainers should feel free to reject any patch by default when the sender can’t explain it.


Totally agree with the conflict-avoidance answer. A small reframing: making mistakes is perfectly normal. A lie is usually an attempt to skip owning one.
What makes you a mature person is:
A lie turns one mistake into two, and the second one is harder to undo.
Excellent advice, you don’t need anything complex, just install git and you’re good to go.
Think of a commit as a save point before the boss fight. One rule is enough to start: commit every time something works, and always before trying something risky. Write the message as if for yourself in six months.
The server can wait until you have something worth losing. Just remember git on one disk isn’t a backup: copy the folder somewhere else now and then.
Practice :) A tiny deliverable after each chapter or tutorial, something simple you can finish in a few days. Accept that it will “fail” sometimes, that’s part of it.
The best bit: the doubts you hit while practicing tell you exactly what to learn next.
That’s how you get out of tutorial hell: you stop following the course and start following your own questions.
I wish I’d had an AI assistant when I was teaching myself to code at 14, in a small village with no developers anywhere near me. My first stack overflow came from a recursive call, before I even knew what recursion was. It took me months to figure out. A patient teacher next to me would have found it in minutes.
That’s how I’d use AI to learn: not “solve this for me” but “where’s the problem? Why? Teach me.” Used like that, it’s the knowledgeable, patient mentor most of us never had. Used the other way, it does your learning for you, and you learn nothing.