That makes the concept much less useful though. Might as well just have two different words that are unrelated. The point of having the inverse idea is to be able to guess words right?
I’d say the main benefit it provides is making learning easier—instead of learning “foo” means ‘good’ and “bar” means ‘bad’, one only needs to learn “foo” = good, and inverse(“foo”) = bad, which halves the total number of tokens needed to learn a lexicon. One still needs to learn the association between concepts and their canonical inverses, but that information is more easily compressible
That makes the concept much less useful though. Might as well just have two different words that are unrelated. The point of having the inverse idea is to be able to guess words right?
I’d say the main benefit it provides is making learning easier—instead of learning “foo” means ‘good’ and “bar” means ‘bad’, one only needs to learn “foo” = good, and inverse(“foo”) = bad, which halves the total number of tokens needed to learn a lexicon. One still needs to learn the association between concepts and their canonical inverses, but that information is more easily compressible