[Tagging] Lyft and nameless sectioning in OSM

[Minh Nguyen]

Vào lúc 14:06 2022-10-11, Evan Carroll đã viết:
> This is also really well said, and we should not overlook that I'm new 
> to OSM and don't know of the time when buildings were not mapped. I see 
> all buildings mapped, and wonder why I need a container to tell me that 
> all things in it are that which it contains to some arbitrary subjective 
> precision. I can't imagine making a technology that would use this 
> as-is. I can imagine answering the problem differently, with a technical 
> solution. It sounds like your opinion is _they're less useful now then 
> ever_. If so, is the community hostile to deprecation? Is there no 
> precedent for saying "this is no longer useful" let's ditch it, or 
> automate it.

If you find manually mapped, unnamed landuse=* areas to be superfluous, 
you'll just love abutters=*, a key that as of 2008 was just as common as 
landuse=*. It remains for when there isn't even enough available 
information to map a proper unnamed landuse=* area manually, let alone 
algorithmically.

Even where buildings have been thoroughly classified, lots of urban 
neighborhoods are dominated by apartment blocks with ground-floor retail 
along major thoroughfares. As the multi-landuse proposal was rejected 
[1], I suspect people are making different judgment calls based on what 
they perceive to be the more prominent of the two uses.

None of this is particularly relevant to Houston, but I don't think 
there's any precedent or mechanism for formally deprecating a broadly 
defined tag in only the places that satisfy certain criteria. Some local 
communities try to dissuade mappers from using certain tags or mapping 
addresses in a certain manner, but it doesn't involve proscribing a 
practice as well-known as unnamed landuse. The closest thing I can think 
of was the situation around population=* in China, now resolved. [2][3]

If, like me, you want to see fewer unnamed landuse areas in your 
backyard, map more named landuse areas corresponding to retail and 
residential developments. These areas not only reduce the pressure to 
"fill in" the map visually but also add information about the shape of 
these developments that's often difficult to obtain from other map services.

[1] https://wiki.openstreetmap.org/wiki/Proposed_features/Multiple_landuse
[2] https://lists.openstreetmap.org/pipermail/tagging/2021-May/061449.html
[3] https://wiki.openstreetmap.org/wiki/Proposed_features/china_population

-- 
minh at nguyen.cincinnati.oh.us