I think you're seeing a couple different effects here.
1. The data is reported from a many varieties of GPS devices and phones. So the data is noisy. Some singles tracks will be very noisy, to the point of being fiction.
2. Not all of these tracks are on the road. If you click between bike/run/both you'll see the bike rides are more correlated with roads and the runs seem to have more "off roading."
3. some of those paths are sidewalks and pedestrian bridges and underpasses.
1. The data is reported from a many varieties of GPS devices and phones. So the data is noisy. Some singles tracks will be very noisy, to the point of being fiction.
2. Not all of these tracks are on the road. If you click between bike/run/both you'll see the bike rides are more correlated with roads and the runs seem to have more "off roading."
3. some of those paths are sidewalks and pedestrian bridges and underpasses.