I find it frustrating that many arguments about LLMs break down into a debate about whether LLMs are useless for writing code or whether they're only useful for churning out 10,000 lines of code that nobody will read. That's why I was inspired by @nolan's Using AI To Write Better Code More Slowly, leading me to write my own post about Being A 1.5-10x Developer. https://build.ms/2026/5/26/being-a-1-5-10x-developer/
Replies:
2
@mergesort Yep this is almost exactly how I operate. Slop-cannoning is fine for PoCs or demos, but for everything else I prefer to go a lot slower. Honestly sometimes I'm a 0.8x developer because I'm spending so much time analyzing prod data, weighing pros and cons, researching alternatives… 😆
@nolan I think my process is probably a bit different, but in the same ballpark. I seem to do enough work upfront to slop cannon a working version that’s better than a proof of concept, and then mold that like lump of clay. But on net my work tends to go faster than it was before because the initial boost is so high, even with phases where I’m a 0.8x developer. And then my ambitions tend to grow where I’m doing a lot more than I would have if the code production depended on my hand written code.
