AFAICT, it's a confusingly aggrandizing term that does offer some new things, but should at most be considered a new subtype of programming language (rather than something different from programming languages altogether[0]—and I'd bet "new subtype" isn't even necessary). He mostly characterizes 'programming language' (as distinct from 'computational language') as telling the computer what to do, as if all previous programming languages were low-level imperative languages:
> With standard programming languages, we’ve had a way to talk about the low-level operation of computers. But with computational language, we now have a way to apply the computational paradigm directly to almost anything: we have a language and a notation for doing computational X, for basically any field “X” (from archaeology to zoology, and beyond).
From what I can tell, what actually characterizes his language is:
- It's pretty thoroughly declarative
- It offers nice ways of incorporating diverse data inputs/outputs into source code
- It seems to incorporate some very abstract notion of an entity that can be involved in a generic symbolic computation process, and maybe a way of defining collections of related entities (i.e. new 'domains')
- It seems unafraid of 'pluralism,' i.e. walking away from the elegance/mathiness of the previous item, and incorporating an encyclopedia-like collection of custom 'domains' into the language.
----
[0] He pretty clearly contradicts himself on whether it's actually a programming language or not. He knows he's not going to get away with saying that it's not technically a programming language, but then... also wants to insist that the term isn't adequate to describe it:
> Yes, it’s a computer language—a programming language. And it does—in a uniquely productive way, I might add—what standard programming languages do. But that’s only a very small part of the story. And what I’ve finally come to realize is that one should actually think of the Wolfram Language as an entirely different—and new—kind of thing: what one can call a computational language.
I once was outlining notes for a course I would like to give introducing computer programming. The idea was to have exercise and theory sessions interleaved, which left the first session, which has to be theoretical in order to introduce the thing, kind of pointless.
So I wanted to explain kind of what programming is, and what it means, and I went to the etymology and why we call it what it is. And it turns out that the etymology tells you that it's a throwaway word. It comes from the same root as a program at a play -- it is just a schedule of what a computer will do, when. The hard concept is not "programming", the hard concept is "computer". Indeed our failure to so far resolve something like P = NP can be traced to our ambiguity on what exactly a computer is.
So I wondered what I would say to kids who had never programmed before and decided that I would just be honest and say that I don't know exactly how to define computers, but that in our experience we can say that they are machines which are at least good at a few things: Math, Art, Control, Graphs. Maybe I had one or two more, or maybe that was it.
This actually changed the course vision directly. Before I was doing what came natural to me as someone with a heavy mathematics education: let's do fibonaccis and factorials and primes. But that's wrong. If you want to teach programming then you want to hit all of these more evenly, not one of them heavily. And if you want one thing that hits all of them directly, you would ask people to start programming games. After all, what is a game? It is a user-Controlled walk-through-a-Graph whose nodes contain Art that rewards or punishes the user for how they walked through that graph. Possibly math exists in the form of, say, physics engines and drawing and such: but it is more of a means to add some interesting features rather than a purpose in itself.
So I had to go back and rewrite a bunch of the follow-up sessions and now it is not in a 90%-done state but in like a 40%-done state, heh.
But yeah, if computations are just what computers do, then the only obvious division between computation and programming -- which Wolfram may have intended -- is that programming languages typically have to focus hard on the sequence of what is going to happen; technically one might say a computational language is less about what order things happen in and more about what result is wanted. Logic languages might be more "computational" than functional languages which might be more "computational" than imperative languages, on this account. And indeed as one goes down that path one sees less and less symbolic representation of expressions and more concrete focus on data structures and values.
Possibly what one needs after that is a sense of metadata and metalanguage: in the language I specify what my conditions of adequacy are for what I am looking for; with some metalanguage I might be able to help a computer figure out one way to get there.
Its a bit domain specific to describe as a programming language. The blog post didn’t describe and control flow features for example. I suppose computation because it is able to incorporate diverse computational algorithms. The unit aware resolver seems like a spin in type systems.
