Procedures are not difficult to design, but procedures that actually work are.
That is why there are such the issues with ATC, RNP with multi-variants was never envisioned in the criteria, as it is built around a 737NG with a Smiths box.
As noted by OK, the boxes act differently, so will the procedures. If you try to put an A320 with a Thales box, on the same procedure that was for a 737NG with a SMiths box, it may not, and most times, does not work. That is why one cannot use the criteria, or the spreadsheets with anything but limited success..hell, they have few turns because TARGETS cant design a turn correctly, the wrong great circle algorithm was used, at least they can use SAAR-Pro, or RNAV-Pro, then dump the results into TARGETS for the database.
Then you assume that you are the only ac in the sky, how do multiple ac of different types, use the procedure, and provide ATC the ability to control the ac?
When you design an multi-variant RNP approach procedure, with a queue ranging from 737, 320, E90, 744, and A380, with Smith, HW, and Thales FMC, and keep them at radar or wake sep, with ATC guidance...
then you will understand what difficult is.