"But while the paper points out some interesting facts, those facts, one expert says, are neither revelatory nor particularly useful for GPS users." The expert is saying we already knew what the paper proves, and it means nothing to a person who owns a GPS watch.
The paper is analyzing a theoretical situation in which there is no interpolation error (when you go around a turn the GPS takes samples at a high enough frequency that its straight-line distance calculation doesn't shortcut the actual distance), and in which the overall distance is calculated by simply summing up the straight-line segments between points without doing any filtering of the data. Neither of these is true for real GPS devices. In fact the first one is often used in articles like this to argue that GPS distance measurements will always be short!
Anyone who has tested actual GPS devices knows they sometimes slightly overestimate the distance, and sometimes slightly underestimate the distance. The idea that they always overestimate is simply untrue.