I find that I have further modified the technique of carving the inside of the plates first. Rather than allowing the inside shape to totally control the outside shape, I have been going back and forth between inside and outside. I graduate 2mm thicker than final finish, then correct the long arch on the outside. Then do the outline and purfling, then the channel and recurve, and then the final graduations from the inside. This method is a little slower and defies Van Zandt's theorem (don't revisit) but it seems to produce a nice plate. The inside layout controls the recurve and is therefore easier to quantify than 'it looks good'. The long arch is also initially driven by the inside shape but small variations that crop up can be fixed before the final thickness is reached. Another bonus is the fact that there's no necessity for a hollowing form because the bulk of  the work, where you would want maximum support of the plate, is done while the billet is thick and very strong.