I’ve used this idea in conversations for years, and can’t find a post on it, which I find surprising but there you go. There are, broadly speaking, two different ways to look at programming languages. And I think that these mean two different ways to select programming languages, which are asymmetric. However, they can lead to the same choice, but viewed in different ways: if you use one of those languages then you need to understand that these different views exist to understand decisions made by programmers working in those languages.
On the one hand, the Language Lawyer seeks out languages with plenty of esoterica and specifics, in order to become proficient at recalling those esoterica and demonstrating their correct application. The Language Lawyer loves that reference and value types in swift are actually called class and struct, or that the difference between class and struct in C++ is the default member visibility, and that neither is actually related to the words class or struct. They are delighted both to know and to explain that in Perl 5, @var, \var and $var refer to the same variable but are accessed in different contexts and what they evaluate to in those contexts. They are excited to explain that technically the admonition that you must always return a value from a non-void function in C is not true, because since C99 if you have a function called main that returns control without returning a value, the compiler will implicitly return 0.
The Language Lawyer seeks out a language with such esoterica. But learning it is complex and time-consuming, so they probably don’t change particularly often. In fact, changing language may be deleterious, because then other people will know the esoterica that they get caught out on. Imagine knowing how the function overloading rules in C++ work, then trying something similar in Java and finding that you’re wrong. And that somebody else knows that. The shame! Once their language is chosen, the Language Lawyer will form heuristic rules of which language feature to use when solving which problem.
Standing in the other corner, the Lazy Evaluator wants to avoid those edge cases, because it makes them think about the programming language, not the software problem. They’d much prefer to have an environment in which there’s as much as one way to express their solution, then worry about how to express their solution using that one tool. The Lazy Evaluator loves Ruby because Everything Is An Object (and they love Io more because in Ruby there are also classes). The Lazy Evaluator is delighted to know that in Lisp, Everything Is A List. They are excited to be programming in Haskell, where Everything Is A Function.
Both of these people can be happy with the same programming language, though clearly this will be for different reasons, and their different thought processes will bring them into conflict. The Lazy Evaluator will be happy enough using the third-most common C-family language, C++– (full name: “C++98 but never use exceptions or templates, avoid multiple inheritance, and [never/always] mark methods as virtual”). The Language Lawyer will enjoy demonstrating const-correctness and the difference between l-, r-, pr-, gl- and x-values, but the two will come to blows over whether the Lazy Evaluator is denying the full elegant expressiveness of the language by proscribing exceptions.
Similarly, the Lazy Evaluator can look at JavaScript, see the simple classless object model with dynamic dispatch remembered from Self or Io, even notice that Functions Are Objects Too and work in that Everything Is An Object paradigm. The Language Lawyer can be confident that at some point, all of the weird coercion behaviour, the Which this Is That this Anyway question, and the You Need That Flag With This Polyfill To Get That Language Feature issues will come up. But the Lazy Evaluator will notice that That Flag also enables class syntax and arrow functions, we already have prototypes and function functions, and disagreement will ensue.
So in what way are those ways of thinking asymmetric? Invoking Hickey’s Corollary, you can never be recomplecting. It’s easier to compromise on C++ But Without Exceptions than it is to compromise on Lisp But We’ll Add Other Atom Types. You can choose JavaScript But Never Use this, you can’t choose Haskell But With Type-Escaping Side Effects.
This isn’t really about programming languages, it’s about any form of model or abstraction. Languages are a great example though. I think it’s a mindset thing. Your task is to solve this problem, using this tool. Are you focusing on the bit that’s about the problem, or the bit that’s about the tool? Both are needed.
