Show HN: Semantic Versioning for Natural Language

github.com

8 points·computerlab··3 comments

This ties into a vague idea I have bouncing around my head about viewing natural language works "structurally", especially in terms of premises, arguments, conclusions, etc.

Analytic philosophers are pretty good at this already, as well as good science writers and thesis authors.

They can present the high level "type signature" of their work: something like, I assume the premises X, Y, and Z, (see books 1, 2, and 3 which argue for these premises) and then deploy arguments of kind A and B in order to strengthen the conclusion C.

Interpreted with a programmer's vocational damage, that text would be a function from (X, Y, Z) to C, using A and B as implementation patterns or algorithms, and recommending books 1, 2, 3 as concrete providers of the types X, Y, Z.

(Some kind of hand-wavey metaphorical Curry-Howard equivalence extended to rhetoric?)

I like the phrase "programmer's vocational damage", I think there's often a tendency to attempt to over-extend programming concepts to other realms (for example, natural language). The irony is that a lot of the concepts used in programming are abstractions of patterns from other fields.

:) Here it makes sense to me partially because rhetoric and logic are closely connected already—but I still want to avoid the trap of, like, using "Gödel's incompleteness theorem" to prove that "you can't really know anything, man".

Reply on news.ycombinator.com