[Tagging] The endless debate about "landcover" as a top-level tag

Jeroen Hoek mail at jeroenhoek.nl
Thu Jun 7 07:59:51 UTC 2018

On 07-06-18 00:51, Warin wrote:
> On 07/06/18 00:57, Martin Koppenhoefer wrote:
>> 2018-06-06 16:39 GMT+02:00 Christoph Hormann <osm at imagico.de
>> <mailto:osm at imagico.de>>:
>>     There probably is a strong majority among OSM mappers that
>>     (rightfully)
>>     think key semantics are irrelevant for the actual meaning of tags.
>>     What a tag in OSM means depends on what it is actually used for, not
>>     what someone says the key used requires it to mean.
> The meanings of the key and value should make logical sense .. other
> wise it is a convoluted thing that is being thrust on all mappers and
> data consumers.
> The meaning should not migrate over time by the misuse of the tag by a
> number of mappers.

Not only that, but it hampers mappers in learning how to correctly use
them. If landuse consistently means "the primary use of the land", and
you learn about landuse=residential and landuse=railway through the
wiki, then a pattern emerges and you know what is going on when you
encounter landuse=commercial and landuse=retail.

But then you get to the local park and see that all large areas with
trees are tagged with landuse=forest (despite there not being an actual
forest), which you learn is how we tag an area covered in trees, and now
the system apparently has a bunch of exceptions you have to keep in mind.

At some point we end up with a database that requires our own private
arcana to understand — I fear that would create a gap between occasional
new mappers contributing through ID, and experienced mappers who can map
major infrastructural projects without breaking a single bus route, but
nothing in between.

Semantics matter a lot in a database that consists of tagged
geographical features.

Kind regards,

Jeroen Hoek

More information about the Tagging mailing list