Comment by JSR_FDED
18 hours ago
It’s funny, the article proceeds to answer the question in great detail: familiarity, level of abstraction, documentation, etc.
I agree with his point that learning to do it yourself can be fun and lead to a flywheel of improvement where the next time you’re faster and the next time faster still.
In the past that was the recipe for success as a developer, nowadays with AI many feel that’s no longer the case.
Extreme apathy towards "learn to do it yourself" is exactly why "AI" is so popular right in programming right now.
It's not apathy towards "learn to do it yourself", it's apathy towards dealing with the recursive fractal of bullshit that is modern software. It's not 2000s anymore, we have orders of magnitude more of nonsense to learn (ironically, most of it are attempts at repackaging and hiding nonsense from lower layers), and learning that is just a waste of life. LLMs finally give us the option not to.
Web dev is absolutely a shitshow, but it wasn't necessary to learn all of it to have high paying job. There were plenty of jobs where you could specialize in frontend, backend, infra, data... even subsections of those. And yet a lot of those people still learned the bare minimum they needed. They wouldn't bother to learn the languages they were using every day deeply, and they wouldn't learn their tools deeply (editor, git, etc). I knew plenty of those people. If you were doing true fullstack then you sure, you have a bit of a point, but many people did not, at least at the places I worked.
Nothing prevents you from just using the lower layers of abstraction. I have developed a few games using Raylib and it was really fun.
1 reply →