I don't believe any s/w based routing sytem will ever be 100% applicable to Audax purposes. How many show the 'hole in the hedge' shortcut a.k.a., the Maidenhead Manouvre on the Upper Thames/Willy Warmer, for example? Don't look, its just an example.
Now, as a Garmin user, I do all my base mapping in Mapsource. This is because apart from the fact that the S/W is on my PC it enables me to have several versions of a route in the same file, so I can play around with options. I usually end up with two routes, 'on the road' and 'controls only'.
For route validation purposes I used to load the control only route into Autoroute and submit the .axe file and a printed report as pdf which includes a summary report detailing the distance between controls, which lends the route application a certain authority. Ususally this goes through without an issue.
I do find that Mapsource and AR generally disagree on the shortest distance. It's generally not significant but can be a royal pain if AR shows mapsource underdistance by a couple of KM as I then have to go hunting for them, playing around with the controls, whatever. Its generally best all round to do this rather than raise an issue with the Org. I try and use the same version of AR as my local DIY org and as Martin has Mapsource he can view .GDB files natively. I agree that Mapsource throws in an occasional 'long cut' but these are immediately obvious from a cursory scrutiny and can be corrected by including an addition waypoint (note, not control) on the 'Controls' route, so I really do not see that as a problem.
Historically a small amount of flexibility was offered to allow for this imprecision - well a couple of km - as DIY riders would generally clock up a few extra KM finding controls, riding to/from the start point (pre-GPS). That seems to have been lost in the search for the holy grail of the perfect solution.
To restate my opening comment I simply do not believe that any s/w routing system can ever be 100%. One does get the impression that the effort expended in searching for it is counter productive as stress testing reveals problems/inconsistencies and the route planning process more complicated, as evidenced by the endless questioning on the matter.
So what to do? I'd like to see clearer guidelines on route planning and (regardless of which tools are used) the Orgs empowered to apply some discretion. That seems to inline with the spirit of Audax.
[FWIW I'd have turned down FBs route. It's inappropriate to assume trunk roads are de-facto motorways - many scheduled routes have long distances of trunk roads albeit on overnight/sunday morning sections. Best to simply plot Controls that keep you away from them. GPS users can place controls where they chose so that isn't a problem, and the more controls one has the fewer 'extra' km one has to ride. This must be traded against flexibility for the 'on the road' route. YPYPATYC.]