I agree with the author we're in a weird moment where the Best Practices seemingly flipped.
Five years ago it was common knowledge on HN that moving away from the code (like taking a management position) atrophied your programming skills. The routine suggestion was to get your hands dirty again.
AI-first folks will tell me I should see my new responsibility as being just such a manager over little agent kiddos running around. However I can turn a flesh-and-blood Junior/Intern developer into full independent developer I can trust with more complex tasks. I can't do that with AI.
I previously made the observation [0] that the AI talk mirrors the Outsourcing talk from the early 2000s. I have a terrible feeling we are going to learn these bitter lessons all over again.
> I can turn a flesh-and-blood Junior/Intern developer into full independent developer I can trust with more complex tasks. I can't do that with AI.
To some extent, you can: better doc and better support scripts (linters, tests, infra as code, etc) do not replace a human being but it goes a long way toward improving your coding agent. Especially since many of those ad-hoc scripts wouldn't even have been written by hand in the first place. The cost/benefit ratio changes.
> I agree with the author we're in a weird moment
We're in the twilight zone, AI ain't there yet, neither are developers.
> where the Best Practices seemingly flipped
There will be new BPs when AI becomes real - for now it's enough to make sure open source AI gets there first.
> the AI talk mirrors the Outsourcing talk from the early 2000s. I have a terrible feeling we are going to learn these bitter lessons all over again.
Because of this and a lot more.
It is a weird state.
- You can't write code as fast or "good enough" as AI.
- Not writing code quickly atrophies your ability to do anything code related.
It seems to me like a giant bet on AI being able to replace software devs (and soon - other professions). And you're forced to participate in that bet if you want to stay a software dev.
I love coding and I love teaching. I greatly enjoy working with novices because I can teach them; but I’m not a manager. Agentic stuff sounds like my nightmare: little of the coding, none of the teaching, all of the managing. Ugh.
>> However I can turn a flesh-and-blood Junior/Intern developer into full independent developer I can trust with more complex tasks. I can't do that with AI.
Why not? Over the past 20 years I've worked with some incredibly smart and accomplished engineers, some of whom have single-handedly built complex products and services used by millions. And today none of them would come anywhere close to what frontier models are capable of.
I agree with your comment but I also think an important factor to consider is that AI is also capable of producing good quality documentation, diagrams and explanations of the code that it produces.
This has the potential to mitigate some of the atrophy that is well documented in management positions and perhaps for someone in such a position (be it managing people or AI) to review code better than they otherwise would.
> Five years ago it was common knowledge on HN that moving away from the code (like taking a management position) atrophied your programming skills. The routine suggestion was to get your hands dirty again.
Yes, you're right. Moving away from the code would atrophy your programming skills, sort of, but if you're still involved with the code on a day-to-day basis, then you're not moving away from the code.
> AI-first folks will tell me I should see my new responsibility as being just such a manager over little agent kiddos running around
Yeah, I've never heard someone talk about it like that, but I'll assume that you have and just say that whoever you're listening to is nobody that knows what they're talking about. I've never seen anyone describe it like that, and you can safely not listen to them.
Which means if you're getting your information from people like that, you probably are very much misinformed, because I wouldn't trust anyone who said something like that to properly inform me.
The important thing to realize is: is somebody moving away from code and moving to non-code things, like a management position, or are they still involving themselves day to day with the code and just working with the code differently? And the disinformation that you've been fed seems to have polluted your way of thinking. I would go back and review where you got this information because, again, it's not a trustworthy source if they're talking about kiddos.
I found that keeping my prompts small and incremental and reviewing the output at each step helps a lot both with keeping my skills sharp and keeping knowledge of the code. It is interesting that I actually learn new tricks from LLM generated code and keep it in check when it goes overengineering.
But that makes LLM speedup like 1.5x the manual-typing coding speed, vs 50-100x you can get if you vibe all the way.
I still use LLMs for troubleshooting, searching, codebase knowledge gathering, solution exploration, writing tests and documentation and whole-codebase changes and there the speedup is dramatic. However I don't feel I become a better developer for doing these kind of tasks without LLMs.
For documentation and tests though I feel I am slipping on reviewing the output because it is just so boring that I often "surrender" the task to the LLM which generates bloated tests and bloated docs.