[OSM-talk] Metrics

Justin Tracey j3tracey at gmail.com
Sat Oct 16 22:41:53 UTC 2021


On 2021-10-16 5:48 p.m., Niels Elgaard Larsen wrote:
> Mike Thompson:
>  >
>  >  Another common one is adding noexit=yes on the last node of a way 
> that ends
>  > near another way when there is no evidence as to whether it really 
> connects.
>  > Typically when one comments on changesets involving such changes, one 
> never gets a reply.
>  >
>  > I suspect that some people are motivated to simply "run up the score."
> 
> Sometimes when you add new ways that ends close to other ways, you get 
> comments from people that are annoyed that they get many new warnings in 
> their validators.
> 
> I try to avoid noexit=yes because even if there was evidence on the 
> ground at some time, it could have changed and it might not be clear why 
> the noexit was there.
> 
> I prefer to tag the reason that the ways are not connected. I.e, add a 
> fence, wall, building, hedge, or ditch. Validators are smart enough to 
> not give warning if there are objects between ways ending close to each 
> others. And it gives other mappers a change to check if that hedge is 
> still there or if there actually is a door in that fence. Or to reassess 
> whether a treeline or a kerb really is preventing people from walking to 
> the way.
> 
> A lot of noexit=yes can be removed by adding a footway that connect to 
> the end of a highway for cars.

I agree, the number of instances where noexit=yes is the best way to map 
something is negligible, if they exist at all. However, it also doesn't 
bother me so much, because as someone who believes this, I can run a 
pretty simple overpass query in an area and find all the instances of 
it, then improve them. The edits from score-seekers who tag things like 
fords that should be culverts, or who blindly connect two ways that 
actually do have some kind of barrier between them, are much more 
problematic, because there's almost no way to find them without a 
contributor happening to run across the problems they cause in person. 
It would be nice if the recommended solutions given by these validators 
were similar to noexit=yes, where they could be readily found after the 
fact and distinguished from more invested mapping.

Another similar (albeit less common) phenomenon I've run into is 
contributors who like to keep an area completely clear of notes, fixing 
the ones they can as soon as they arise, but also just closing notes 
that would require more information (from e.g. a ground survey) without 
comment.

FWIW though, I've personally found that politely reaching out to 
score-seeker contributors, even if they don't reply, is usually (though 
not always) enough to get them to at least greatly reduce the 
problematic behavior in question.

  - Justin



More information about the talk mailing list