On 14/07/2016 16:19, Éric Gillet wrote:
So if the changeset correct 300 restaurants but 2 are "damaged" by the automated edit, would the edit be bad enough to be reverted or not be done in the first place ?
I'd revert it. It's essentially the same as the "trees" example upthread (where the mechanical editor thought incorrectly that deciduous implied broadleaved, and vice versa). It's easy for people processing OSM data to say "obviously these are mistaggings; I'll assume that people have just got it wrong". However, going to Burkina Faso (in the Mac Donald example) is inherantly much harder; we need to respect someone who's actually been there.
Where there a small number of potential mistaggings the correct approach would be to _ask the previous mapper_ or if that doesn't work _ask someone else in the area_. OSM provides tools that makes it really easy to do exactly that.
You might argue "but surely if more data is corrected than damaged the overall quality is improved?" but you'd be wrong. It's important to leave as much original data there as possible for downstream processing. I spent a good few years in the 80s and 90s arguing the superiority of statistical approaches to data interpretation over rule-based ones. To cut a long story short, there's a reason why e.g. the anti-spam measures used with email today are Bayesian (statistical) - it works. Don't second-guess what data consumers might need if you've not been one.
That doesn't mean that if you see that someone has mapped an obvious primary highway as "highway=pirmary" that you shouldn't change it - but do always ask yourself if by "tidying up" you're actually removing information from OSM, even if that information is "there is some doubt as to whether the original mapper knew what they were doing".
Best Regards, Andy
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
