G'day
keithl mate!
Why did I know, waaay back in 2003, that this topic would come back to haunt me? And, given that I was aware of it, why didn't I resurrect all the work I did on it, many
MANY years ago, to come up with a pilot tolerance of 1 NM? I guess the answer is that I put it all on a file, allllllll those years ago!
I'll try to find time to recreate the calculations that I used. However, in the meantime, I would just say that people should not automatically assume that Pans Ops is a "standard". It is not. It is a document, not an Annex, so there is no requirement for States to blindly follow it, for the simple reason that it provides guidance for procedure design, not how to fly the resultant procedures.
Sure, it provides information on the assumptions that are used to develop the design criteria for each type of approach. But the fact is that everything in it is the result of an RSS (route-sum-square) of the "assumed" values. It is not possible to specifically identify the contribution of each assumed value in such an equation. This is why there is such wide variation in the interpretation of the pilot tracking tolerance between States.
All we, as designers, can do is to take the basic data and interpret it in a way that works in our own environment. I agree that pilots should be able to fly an arc within + or - 1/2 NM of the nominal distance and, indeed, I've never had any problems with that. As others on this site can undoubtedly attest, I'm not the swishest IFR pilot around so, if I can fly within 1/2 NM on an arc, then everyone else should be able to achieve far better results!
It might take a while to reconstruct my calculations, so don't hold your breath waiting for it!
Even then, it won't be authoritative across the world - just the way that resulted in the value that is used here.