bun's rust rewrite is a marketing stunt

yesterday Bun merged it's million lines long PR that rewrites the entire project from Zig to Rust. you might be able to see said PR here, if Github feels like cooperating.

this has, expectedly, caused a big commotion. a lot of people are talking about it, and lot's of people have opinions on the subject. the HN post has over 700 comments, the Lobsters one over 100.

and everyone seems to be discussing this change on its technical merits. on if it makes any sense to switch from Zig to Rust, if there's any need, if a million llm generated locs are maintainable, if the new code is going to be safer, or more buggy, or whatever.

but like. i find it hard to belive this is anything more than a marketing stunt. i mean, i believe it probably did start as an experiment. bun's creator Jarred said so himself on HN when the branch got noticed originally. but i also know for a fact that anthropic took note of all the buzz it got. and we also know Anthropic has done big projects like these before (like their c compiler). if you were anthropic, would you not immediately call Jarred and ask him to drop everything and focus on getting that PR merged?

if there's one thing that most engineers can agree on, is that whole-codebase rewrites are costly, expensive, and hard. at least, it's an Engineering Blogpost worthy endeavour. it's not something you even consider unless there's very clear benefits. what better showcase of LLM's capabilities than doing the unthinkable in a week?

and now Anthropic get to say that they solve this problem! it's an incredible selling point for Claude Code. at $work we use PHP, and i some times get asked if i've considered changing to any other language. the answer has always been an incredibly easy No. the cost would be way too high, and the pros/cons are never too good or concrete to justify spending all that time and money. but now that cost has been slashed by a couple orders of magnitude. maybe not enough to make it worth it for $work, but certainly for a lot more companies.

and next time you propose a rewrite to an Exec, do you think they'll say: "yes, let's spend 18 engineer-months on this rewrite" or "spend 20K on claude tokens to do it in a week and get it over with"? this Bun rewrite creates new market opportunities and captures a big chunk of the existing ones. if i had an MBA i would probably make up a big fancy number for how much money they'll maybe make from this.

so why are we analysing the technical merits of this migration from Anthropic/Bun's perspective? i don't think they care about the technical case when the business one is so strong.

if i worked at Anthropic i don't know if i would care too much about Bun being unmaintainable after this rewrite. fundamentally, software is now Cheap. we are entering an era of software industrialization. as it happened with clothes manufacturing, the price to produce software has gone down incredibly. like. it doesn't matter if Bun is unmaintainable in this new version. the point is that it took them 6 days (SIX!!!) to generate this entire rust codebase. if it's shit, they'll just make it from scratch again with what they know.

if your tshirt is broken or you've outgrown it, you don't go to the tailor to get it fixed. you buy a new one, because it's cheap as hell. so of course i don't think Anthropic cares about whether their new Bun code is maintainable. they have proven they have the tech to make a new Bun from scratch whenever they want.