> Consider how this sounds when rewritten to talk about natural languages:

Computer languages are formal systems. You should compare to math, not
natural languages.

We need a "standard model" of programming languages. Every day, type
theorists are busy making bits and pieces of it. Once they rid
themselves of the mindset that every language is an island, it's only a
matter of time until we get what I want.

> Of course the reason is that a natural language is "un système oú tout
> se tient", although to varying degrees.  A language using Finnish case
> endings, Bantu noun classes (like grammatical genders, except there are
> 22 of them), and Navajo verb endings applied to English roots isn't a
> system, it's a mess.

Computer languages are not inherently messy. Probably most of the
improvement in the field proceeds directly from cleaning old messes.

>     A language, as we now know it, is an agglomeration of features,
>
> That's just what it isn't, any more than a natural language is a big bag
> of words.

A natural language is an even worse agglomeration of features than a
programming language. In the former case it's excused since a language
serves to protect a culture. Programming languages also protect
cultures, but usually for all the wrong reasons. That's what perennially
keeps the field back.

> I see no "inevitability" here.

I see no angle from which it isn't inevitable, apart from inertia.

For some reason people tend to start thinking from the language and say,
"what kinds of problems can this solve?" Try starting from the problems
and asking "what language features does this problem need?" Most
problems don't need anything exotic; they are widely translatable.

> As the thief said to the magistrate,
> "Mais, monseigneur, il faut vivre"; the magistrate replied scornfully
> "Je ne vois pas la né-ces-si-té."

MessageNotUnderstood