One of the memes on Twitter is that “programming is solved now.” Familiarizing yourself with a language no longer matters, and some friction for humans does not matter for agents. As a result, LLMs make language choice much less consequential. You can rewrite code in another language and pick one you are unfamiliar with. This means people choose based on marketing much more. As a long-term programmer, I found it fascinating to see people now ship Rust code who previously might not have chosen it.

I attribute this to two recent vibe shifts: there is a lot more talk about wanting fast software, and about LLMs being exceptional at optimizing code without regressing behavior. Folks like [PERSON_NAME] and others always carried an obsession with fast software and are receptive to agents writing code. With autoresearch, you don’t necessarily need to know all the tricks; you just need to put an agent on it.

Projects increasingly pick hard languages. Even Zig benefits, despite creators being negative on AI. For instance, Cloudflare’s new Artifacts service uses a pure protocol engine compiled to a roughly 100 KB Web Assembly module, and fx, a coding agent advertised to be small and fast. People are also tackling much harder technologies like DWARF files, e BPF, and custom crypto. Maybe the world will have more slop, but it might also have more developers wanting things fast and small.