[xt00]

I wonder if at some point Apple reduced the sampling frequency of the gps position to save power? Since this sounds pretty important to the guy, I would buy a Garmin watch or similar to compare to and run with both. I also suspect one of the challenges here is that taking a lot of corners like in an urban environment might result in rounded paths — like if you ran around a rectangular city block it ends up being logged as a square with rounded corners. Whereas if the sampling frequency was higher it would capture those corners much better, or if the authors runs were more of a straight line.

[loufe]

It would explain his shortened distances, as low sampling would round out corners, "shortening" the total distance.