Hacker News Reader

Comment by bgirard | Hacker News Reader

> We did not type-check our codebase

For me that's the most surprising. Personally type-checking has such an immediate ROI on productivity. Even if adopting it incrementally. It's certainly not required to build a business but it's not something I'd skip.

I still like the overall point of the article.

View on HN

This is list is not a list of thing not to do, simply some popular things we did not do. I'm absolutely not saying type-checking is bad.

Yup, I understood that. That's just the one I personally found the most surprising in your list.

I love type checking but I do think the point at which type checking becomes a necessary business decision is very far in the future. Companies such as Stripe, GitHub, Facebook, etc. are examples of this. And with some typed languages, e.g. Rust, Swift, OCaml, there's an additional factor of a weaker ecosystem. Yes, even Rust's ecosystem is worse than say, JavaScript's for some areas such as desktop applications.

I use typed languages not for a quantitative ROI. I use them to make programming not a huge pain in the ass.

I worked on Facebook Web Infra while we were adopting Flow. And typing was a huge productivity and safety boost. Sure, you can build without out. There's no denying that. But now we're in 2022 and we have mature type checkers for JS so it's not something I would skip on today.

I would love to know if there's studies out there that demonstrates any relationship between programming languages and business metrics.

The closest I found was a blog post [0] that implies that Ruby and Python were the most prominent languages among the most valuable startups from YC. Those are distinctly not languages that lend themselves to type-checking.

[0] https://charliereese.ca/y-combinator-top-50-software-startup...

I'm unsure that ROI is necessarily there for a team of 4.

Reply on news.ycombinator.com