The men who wouldn't document, lint and format their code so they themselves and their coworkers could maintain it 5 years ago, even to save their lives would totally do it nowadays to "help the AI".
@alda 🎯🎯🎯 there's now documentation/tutorials in AGENTS.MD and CLAUDE.MD to help the AI which could have been documentation 5 years ago to help contributors...
@benjaoming The 2026 version of "this could have been an email all along" is "this could have been in the readme all along".
@benjaoming @alda I have seen one of those but it was used to "forbid" agents. It was great, no idea if it does anything in practice but still neat
@alda It's because they recognize that AI is not intelligent, unlike their fellow co-workers, and therefore need baby-sitting.
@alda literally yesterday I almost lashed out at a colleague who discovered that with a complete documentation copilot works better and calls it "AI working culture" where for the past few years my begging for any docs was ignored :(
@hiiamfrompoland Yup. This cognitive dissonance seems to be an issue in the whole industry.

@alda @hiiamfrompoland I'm maintaining hand-written, zero-AI documentation for my #UE5Coro plugin. Anything I add or change gets documented in the exact same commit.

Someone tried to vibe code with it, the LLM invented a type that never existed, it obviously didn't work, and this was somehow the fault of my shitty documentation. 😮‍💨

@landelare @alda

hang in there

10-15yrs and we are going to be the last people capable to lace a shoe

@hiiamfrompoland @alda We were promised self-lacing shoes by 2015. 😢
@alda I also wonder if I am the first to fire, because having documenting habit it seems AI can manage my projects better than of others who made a bunch of undocumented spaghetti code - so they are necessary because AI clearly can't replace them
@hiiamfrompoland Both documentation and UX seem to be considered to be "fluff" by the "I want results" types. Until it started facilitating the LLMs.
@alda @hiiamfrompoland it also irks me that I used to love to nicely format docs, use emojis to make it prettier (not excessively, or a replacement for text of course) and now that is seen as an AI signal. yuck
@spinnyspinlock @hiiamfrompoland As an avid user of em and en dashes and Oxford commas, I feel you.
@alda @spinnyspinlock @hiiamfrompoland
Wait, oxford commas are an AI signal now? I am becoming more radicalized every day
@trampinheavy @spinnyspinlock @hiiamfrompoland It's a *perceived* AI signal just like the dashes, as a lot of Anglophone adults don't really read books anymore — which on the other hand get fed into the LLMs.
@alda @trampinheavy @hiiamfrompoland spent quite a lot of time trying to learn to write "correct" English, learning about style guides so I could write clearly, at least my poor grammar is now an anti-AI signal 
@alda Maybe they're simply supplying the same sloppy code of 5 years ago.
@alda also the managers who wouldn't write a ticket
@byroon But they'll write promts.
@alda exactly
@byroon For my next consulting gig, I'm going to LARP as a Slack-resident AI agent.
@alda
AI doesn't need any of that.
@[email protected] That's a very good idea. I should try that one day.
@alda more than one developer who has told me that AI is “like having a junior you can assign tasks to” were the same developers who refused to have junior devs on their teams because “juniors slow me down with all the hand-holding they need”
@alda @maxleibman Yes, because "Helping the AI" is really "Outsourcing more work to something that isn't me."
@simon @maxleibman It's also about creating administrative busywork when actual work could have taken place.
@alda @ced I think this is of the things that pisses me off the most about the whole gen AI crap because it exemplifies the complete disregard of humans that these bros show…