[Tagging] Development of Tag Replacements (was: Feature Proposal - RFC - landuse bush)

[Stefan Tauner]

On Wed, 3 Feb 2021 23:12:05 -0500
"Brian M. Sperlongano" <zelonewolf at gmail.com> wrote:

> landcover=trees which overlaps with the overwhelmingly more popular
> natural=wood and landuse=forest
> landcover=grass which overlaps with the overwhelmingly more popular
> landuse=grass

> I personally like the idea of landcover, and I think it was a good idea
> when Martin introduced it 11 years ago (well before my time on the list)
> but it hasn't gone to a vote or really caught on in actual tagging. 

> In an alternate universe where landcover was a widely accepted/used tagging
> scheme, you might imagine that landcover=landscaping could be a valid
> top-level tag, with a collection of landscaping=* tags to further define
> the type of landscaping.

I think OSM falls for this hen and egg problem since its infancy. In
the beginning this was not too bad bad and superior tags had a realistic
opportunity to get adopted even if the existing ones were widely
used. I don't think there has been any chance to do that for a while.
How could that possibly work today?

A proposal to deprecate or substantially redefine such essential tags
like natural=water could be perfectly written, making complete sense
for the future of OSM but would still have a very hard time getting
passed IMHO.

But let's suppose that we discuss this enough so that we get enough
votes together. What then? Getting everybody outside this relatively
small party of tagging contributors in the same boat including all the
software devs, companies and sometimes stubborn private mappers is a
major effort that basically requires a completely separate tagging
scheme to remain backward compatible. This alone might make it already
very hard because there are limited decent top-level key names. More
importantly, this reduces the incentive for everybody to move:
 - mappers might not even know about the new tagging and happen to know
   the existing scheme
 - (smaller) data users won't adopt such new schemes if there is no
   usage anyway
 - editor devs might add them sooner or later but there are various
   true reasons and excuses to not do that, or only optionally/non-
   intrusively - reinforcing #1 again.

From the top of my head the contact: namespace comes to my mind but I
presume there are better examples where a newly introduced scheme is
superior but (still) hasn't caught up.
More interesting would be examples where there were big changes with
deprecations of previously ubiquitous tags to learn what factors are
important for succeeding. Can you remember some and know the
development behind them?

-- 
Kind regards/Mit freundlichen Grüßen, Stefan Tauner