I started to play around with crosspicking now a week ago.
I’m now a TWPS, which isn’t second nature yet, but I’m quite confident that I’ll stick with that for the future.
Now I starting noodling with 4 string rolls and with some time to find parameters I can blaze through that (this is pretty subjective 
) with almost no effort.
So i thought I’d give an old enemy a try - 1 NPS inside on 2 strings … no way.
I tried 5 string rolls … no way.
Now after looking close on what’s happening in my case on crosspicking, I think there’s a lot more than just one general motion, I use the string impulse to get over in one place and have the wrist flexion in another (and probably more I didn’t notice so far).
Now if I change the picking pattern all those variables inside the general motion fall apart.
Now after rethinking the Andy Wood interview, to me comes this question up:
Should the crosspicking approach be wider and include specific patterns?
I mean would it make sense to build maybe 5 exercises with dfferent patterns and connection points and ‘allow’ totally different motions instead of trying to find the one holy grail motion?
This is really a question, in the sense of I don’t know.
Troy’s version looks as if he could play basically want he wants without changing anything (eventhough it looks as if there’s slant tendency that changes with descending or ascending).
On the other hand he wrote the main problem is string tracking, which might be just another version of my experience.