I think that discretes and questions relating to passages will be experimental, but I do not think that an entire passage will be chalked as experimental (just my $.02). I imagine that once a suitable passage is located there is a q-bank written for it. This q-bank is eventually narrowed down via experimental questions until they're satisfied with the curve.
Discretes are slightly more straightforward. First they need to decide whether or not they're going to use the question. Once that's decided, they'll then have to figure out "good" wrong answers. I feel like a question we had on 4/27 was serving this purpose. There were simply too many inconsistencies and errors within the question for it to be a legit question i.e. structures given did not match names and atoms were missing.
I keep thinking back to the exam date and I've realize that, as far as I can remember, there are two PS questions and two BS questions that I got incorrect. I need to get this god-damned test out of my head. I felt amazing after the exam and now I'm allowing myself to get defeated by it when there's nothing I can do. I just hope that one or two of those incorrect answers are experimental questions.