Single file social media - like twitter but by writing markwhen. Learn more.
Thesis: interesting work (i.e., work worth doing) is less amenable to the use of AI than ... non-interesting work.
I have to admit that I get a little FOMO when I see people on HN laude the use of AI and how much more productive it's making them. Meanwhile, I feel like if I tried to offload my work to an LLM, I would both lose context and be violating the do-one-thing-and-do-it-well principle I half-heartedly try to live by.
I don't mean to disparage anyone working on boilerplate, it's something we all have to do, but at the same time that's just not what I'm doing. Or at least, that's not what I'm spending most of my time on. Why would someone be writing boilerplate all the time? How is that not, even without LLMs, a failure of efficiency? Why would you think your boilerplate-writing job would be secure?
Have I been misunderstanding what a lot of software engineers do, or think they do? I thought we solved problems. I didn't think we just matched boilerplate to feature requests, merely the last worker in the assembly line of idea manifestation.
