I found around 20.000 places that have both tags set. Spot checks show usually noname=yes was set first, then later a mapper added a name. Those are mutually exclusive, or is my understanding wrong? Can I add an editor definiton somewhere to it gets marked by validators automatically?
Tag:noname=yes - OpenStreetMap Wiki suggests mappers added noname=yes to avoid a warning “silences validator complaint (presented as a pink outline) about a missing name” while in reality the place often had a name, it was just unknown.
No, you are right. Having both noname=yes and name=* on an object means something is wrong.
Probably most of the times someone set noname=yes just because the name was unknown at that point. And someone else added the name later without removing noname. But I am not sure if we can assume that this is always the case.
I defenatly think that this conflicting tagging should get fixed but I am not sure how to do this efficently.
A warning when editing objects tagged this way would be very handy. You can request such a change for the iD-Editor, but I am not sure how exatly that works.
The problem is, what is “correct” can be difficult to determine. As an example, have a look at this random one. I probably wouldn’t have used either noname=yes or name on that, since it has name:left and name:right, but I’d probably want to check with the people who added both those tags (perhaps by referring them here!).
This only makes sense if the user being told of the problem is able to decide which is correct. In the case of noname / name combinations, they would need to (a) be familiar with the area, and not just using aerial imagery and (b) look at the object history (like in my example above, where arguably both tags are wrong) to understand what happened.
It’d be interesting to do an analysis of iD’s “X was fixed” changeset tags to see what proportion are actually genuine fixes, as opposed to just making QA tools shut up.
There is more to many of these cases than meets the eye. SomeoneElse’s first example highlights this well. You should refrain from mass editing this, because we would risk losing track of the unusual naming situations that exist IRL. As Mateusz said it’s better to include a check for this in QA tools.
There’s always an “Ignore this issue” option. The validator warning in the current pull request only provides this option, but I’ve suggested adding two more options, one to keep only the positive tag and the other to keep only the negative tag.
I don’t think it’s necessary or even possible for the editor to know which of the two is more correct. Rather than have the editor guess, it’s better to present a choice to the mapper. In general, we trust mappers to know what they’re doing when tagging something. But if you’re concerned about mappers picking at random, the warning can come with some text directing the mapper to editing aids such as street-level imagery. This text can vary from key to key.
