For cycling, I now use an Edge 305 and have very consistent numbers with a Polar and Ciclo computers so I feel I'm getting stats somewhere in the ballpark as long as I use the right software for processing.
Then, because of all the talk about motionbased and sheer curiosity I wanted to run stats on a file where I didn't have any preconceived notions of terrain, in other words, someone else's MB file from across the continent. Therefore, I took
Mr. Silver's Shawnee Forest motionbased file and loaded it various ways:
Stats at MB (can't tell what corrections he applied): 4,389 ft gain / 4,617 ft loss
GTC: 3,732 ft gain / 3,962 ft loss
SportTracks w/o correction but w/data smoothing: 3,742 ft gain / 3,969 ft loss
SportTracks w/ correction and datasmoothing: 3,869 ft gain / 4,065 ft loss