3 ms·
I'll post the same rant I usually do whenever this article comes up: "The Blub Paradox" is mistaken. It's a misunderstanding. To see why, think about Haskell
by AnimalMuppet 4y ago
I'll post the same rant I usually do whenever this article comes up:
"The Blub Paradox" is mistaken. It's a misunderstanding.
To see why, think about Haskell. Haskell users are sure they're at the top of the power curve. When they look at Lisp, they're sure that they're looking down. "How can you get anything done in Lisp? It doesn't even have a decent type system."
But Lisp users, when they look at Haskell, they also are sure that they're looking down. "How can you get anything done in Haskell? It doesn't even have macros."
This situation - two languages, each certain that they're looking down at the other - tells us that there's something wrong. And what's wrong is the idea that languages can be ranked on a one-dimensional axis labeled "power".
So, what is "power"? Or perhaps better put, power to do what? Well, to write programs. What programs? General programs? I've never written a general program in my life. I've written a bunch of specific ones, though.
When I'm choosing a language, I care about its power to write a specific program. (Because why do I care about how good a language is for writing a program that I'm not trying to write? Why would I care about that?)
So, ask what technical issue makes it hard to write the program you're trying to write. Then look at what language goes the furthest in the direction of making that problem easier. (You could go further, to think of the set of technical issues as a vector in a multi-dimensional space, and languages as a tree in that multi-dimensional space. Pick the language that reaches furthest in the direction of the vector of the technical problems.)
But you have to have a somewhat broad idea of "language". It includes libraries and ecosystem. It includes the ability to write code that can be maintained over multiple decades (if that's part of what you need).
TL;DR: The Blub Paradox is mistaken. For it to work, you have to be able to rank languages in one dimension (power). You can't. It's much messier than that.
- sokoloff 4y ago> why do I care about how good a language is for writing a program that I'm not trying to write? Why would I care about that? If you’re planning to pick a language de novo for each program, you don’t care about this. That’s probably not the common case that a dev intends/is willing to learn a new language for each new program.
- AnimalMuppet 4y ago"Language I know" vs "don't know" is one more axis of the vector space. If language X will make it easier to solve the problem, then the trade-off is "learn X and use it to write the program" vs "use Y to write the program". Does X make it enough easier to be worth the effort? And then I have to think about, well, X sounds like a useful tool. Is it useful enough that I want to make the goal be "learn X, and write this program" instead of just "write this program"?