My current project: Having spotted (with others) a completely incorrect path near Craiglockhart Pond (Edinburgh, UK) I decided to do some investigation. It turned out that there were some relatively big errors in other paths (incorrect path connections etc) in the area. I took initial GPS tracks and then returned a second time armed properly with camera, pen, paper etc. Difficult decisions about what to map... do we include the relatively insignificant worn paths through undergrowth or unofficial worn bits down the steep muddy hillside? I've decided to do this in some places, perhaps where this demonstrates that there's a confused network of paths - sometimes leaving a path incomplete to suggest it's partially there - and to tag paths with 'tracktype' grades to help suitable renderers to make sense of their relative significance.
Previous mapping in a tree-covered glen has helped me to work out some techniques for dealing with inaccurate GPS traces. A rather obvious thing I'd not thought of before has been to map in the winter as the trees presumably block the GPS signal less.
I've completed work on half of Easter Craiglockhart Hill (near the pond) and plan to look at the path network around the top of the hill another time. This may well be more accurate already - although I've already spotted some significant missing tracks. If anyone wants to help with this (or comment/criticise) do say so.