Note: An LLM was not used in writing this article.
In a tweet (before Musk bought it and everyone left), during Covid, I wrote (which was quoted in a talk called “Stop Writing Dead Programs” by Jack Rusher):
Becoming more and more aware that languages in the ML tradition (Haskell, Rust, OCaml, etc) are about a certain kind of linguistic game that the developers have fun playing, and not because it makes any software arrive quicker, behave righter, perform better, or last longer.
I thought I’d expand on that a little bit more.
Techniques, methodologies and technologies can become something that people get attached to, and, as I wrote in 2023, can become a bit of a crutch:
It seems to apply strongly to formal languages (XML, Nix, Terraform, Kubernetes YAML), some architectures (microservices, the cloud, SPAs, MPAs, OO, containers), and type systems (dependent types, effect systems, the borrow checker), and methodologies (devops, scrum, open source). [2026 note: and LLMs] All quite good used in moderation, but tend to be used like violence. [If it’s not solving your problem, you aren’t using enough of it!]
All these things, people build businesses upon writing about, giving talks about, and engineers are enthralled by. Sometimes managers and C-levels are worried about falling behind if you don’t adopt that thing, or aren’t doing it enough.
I did consulting for about 35 different companies, who were all using Haskell. Not one time did I look at a codebase or management structure and say wow, this is perfection. As you can imagine, or not imagine, there was a lot of garbage, a lot of mess. Many of these companies are gone or still around. But no one particular technology, methodology or system would have moved the needle. Haskell/Rust doesn’t save you, microservices/monoliths don’t save you, AI doesn’t save you. Politics, the market, culture, bad luck, not enough money/people, or too much, are the tide that we swim in. Companies are “wicked problems” (if you look it up on Wikipedia, even Harvard Business review has an article on this).
I don’t mean to frame this in a negative light (though, you may interpret it that way). There’s loads of joy to be had in using any of these things to tackle the wicked problem of working in the software industry. This is why I choose Haskell companies to work with, where possible, because despite all its flaws, I simply like it. And I like the quirky and interesting people that also enjoy it. But I don’t pretend it or any other modern piece of technology, methodology or system really moves the needle.