I think it is pertinent to note that GPS distance will always be less than actual when there are switchbacks involved. I've never had my garmin reflect the actual distance traveled unless it was a long, straight point to point course (like the ADT Marathon). It always comes up short when there are curves or turns involved.
The two spots you mentioned are loaded with lots of short, sharp switchbacks. GPS systems tend to cut the tips of them off, often losing 20ft for each one even with a 1 second sampling rate (sometimes a good bit more if the sampling rate is set to "smart' or "every 3-5 seconds). The accuracy of a modern commercial gps unit is +/- 15ft. Data smoothing algorithms take the jitter out of the map to make the tract look better than the raw data, but in the process also shortens the distance a bit as the trail swaggers from side to side.
You also have the question if if the system is recording surface distance or the flattened distance. That difference alone is good for about a tenth of a mile on the Ascent. The extreme case is the Incline at 0.9 miles by gps but 1 mile if you were to drag a rope along it.
A while back, Matt Carpenter took a wheel to the trail for the BTMR. Up and down the entire course twice. I'd love for somebody with a lot of time on their hands to do the same for the BC -> Summit segment. But I don't know anybody who is that bored. :-)