DFC.
Once again, I beg to differ. I think in the pre-RNAV days, the tracking of aircraft would've been less accurate (scalloping etc), than with the dead-on multiple GPS gismos today. Perhaps you should have a read of your own post, in that RNP5 separation quoted is less than for VORs.
If you are trying to say that pilot error is the negating factor, then that applies equally to any instruction (such as flying a heading). We are talking about a specific situation; where the instruction has been issued to track to direct to somewhere. If the controller doesn't check the tracking, before basing sep on it, what's the difference to issuing a heading and not checking the a/c is approximating that?
As for all the stuff about route separation....I'll bear that in mind if the radar fails. I'll give you a tip; just because an aircraft is left on it's own nav, on it's route, radar separation doesn't stop applying.