Recently in the course of thinking about training sets for machine learning I’ve found it useful to make a two-way distinction between different kinds of propositions. It would be helpful to have a bit of jargon to summarize this distinction so I’ve coined some terms.
There are two kinds of propositions, which I’ll call conventional and objective. Conventional propositions are true solely as a matter of convention. If enough people who know something about conventional proposition X agree that it is true, then it is true ipso facto. Statements about language are good examples of conventional propositions. If one hundred literate users of a language with a Roman alphabet agree that a particular handwritten symbol is an “A”, then it’s silly to claim otherwise. It is a conventional truth that the word “glory” means “praise, honor, or distinction” and not “a knock-down-argument.” Everything that’s not a conventional proposition is an objective proposition.
Though this is a structural cousin of the standard a priori/a posteori or analytic/synthetic dichotomies, it is not the same thing. (In particular, I think it only applies to synthetic and/or a posteori propositions.) There is some fuzziness around the question of whose opinion counts, and an extreme relativist position could be expressed as “all propositions are conventional.” In practice, though, it’s often clear who gets to weigh in, and in these situations the conventional/objective distinction can become sharp. If one hundred literate English speakers agree that a particular handwritten character is an “A”, it’s an “A”. But if one hundred ancient Greek astronomers agree that the sun revolves around the Earth they’d still all be wrong.
I’m surprised that I don’t already have jargon for this distinction at my fingertips, but I don’t so “conventional” and “objective” will have to do until I can dig up the more broadly accepted terms of art.