I have noticed this while working on oldest one and not sure about later one, but I glanced over the latest version, and now doubt it had few improvement about this point. So please ignore if it's by false evidence...
Most questions in rmcnew's test are according to existing descriptions of every type, chopped characteristics into partculler for functional comparison. Then, if functional analyze does some right, somebody determined on certain type by its description will find it as his/her type.
Rocky once thought he's ENTp and now ISTp, so the test worked right on functional analysis. Just sure on functional, it can't say if model X is doing something much better than existing one is, nor Rocky is actually ENTp or ISTp.
Of course, I think Rocky can be both ISTp or ENTp (though approachable and recommended to be a lawyer is the rarest thing for ISTps, I think. Yeah, on this issue I'm SOOO narrow-minded on guessing in spite of
on 2nd function
), but the test tells nothing more than existing descriptions, so nothing is new if you've already read through them all.