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

Christoph Hormann osm at imagico.de
Wed Jun 6 14:34:27 UTC 2018

On Wednesday 06 June 2018, Andy Townsend wrote:
> > This issue has a long history... seems to me tagging awaits
> > rendering, and rendering awaits tagging. In such cases, you need a
> > commitment from both sides, with enough support to fuel trust. Then
> > things can get rolling.
> Speaking as someone who both adds things to the map and maintains a
> map style, only the second of those is really true - rendering does
> (thankfully) follow tagging, rather than whatever someone happens to
> have changed a wiki page to recently.  Tagging absolutely _does not_
> follow rendering - as I've found out many times when looking through
> taginfo.

I have heard this claim many times in the past but never have seen this 
backed up by evidence.  OTOH i see plenty examples where it can be 
demonstrated that rendering of tags has massively affected their use - 
not necessarily in absolute numbers where other effects (like editor 
presets and imports) are often much stronger but very often how they 
are used.

The clearest examples of the effects of rendering on tag use are in case 
the standard style stops rendering certain tags.  This is extremely 
rare.  One good example is landuse=conservation which was removed in 
September 2015 which according to http://taghistory.raifer.tech/ marked 
the begin of the decline of use of this tag.  But you always have to 
keep in mind that the effects on overall numbers will usually be 
subtle.  Mappers will not commonly start mapping certain features in 
huge numbers just because they start being rendered.

Another positive example was the differentiated rendering of wetlands 
(natural=wetland + wetland=*) added in May 2015 which lead from 
previously only about 1/4 of the wetlands having a wetland=* tag to by 
mid 2015 more than 1/3 of the wetlands having a wetland=* tag.

Anyway what i am absolutely certain of is that rendering different tags 
identically in a map has never encouraged mappers to consistently 
differentiate between them. :-)

> [...] The wiki page
> https://wiki.openstreetmap.org/wiki/Tag:landuse%3Dgrass says "It is
> typical that landuse
> <https://wiki.openstreetmap.org/wiki/Key:landuse>=grass is misused
> and should be changed to landcover
> <https://wiki.openstreetmap.org/wiki/Key:landcover>=grass
> <https://wiki.openstreetmap.org/wiki/Tag:landcover%3Dgrass> ".  I'd
> argue that that's wrong; it's the majority usage of a tag by
> thousands of people that determines what is "correct", not a wiki
> page edited by 1 or 2.

The problem here is that it is not just one or two, there is a 
significant group of people, at least a dozen overall i suppose, who on 
the wiki consider it their mission to educate mappers on correct use of 
tags (based on certain ideas regarding key semantics or data model 
ideas in general) rather than documenting their actual use.  The 
uselessness of many tag pages on the wiki - 
https://wiki.openstreetmap.org/wiki/Tag:landcover%3Dgrass and 
https://wiki.openstreetmap.org/wiki/Tag:landcover%3Dtrees are good 
examples here - is largely due to that.  They are filled with talking 
points from the fight for 'correct' key semantics that leave the mapper 
looking for substantial information on tag use with nothing but 

> The tagging list does occasionally fall into the wiki-hole of trying
> to tell people how to map rather than communally deciding the best
> way to map something (including by looking at how people already
> do).  In any situation where you're trying to suggest that "everybody
> else is wrong" you need to get over it, and OSM in particular has
> thrived where other similar projects failed simply because people can
> always find a way of expressing a particular concept - they can
> create a way of representing it themselves without a "domain expert"
> creating it for them first. There may well be a concept out there
> waiting to be mapped that needs a "landcover" tag (and it might be
> "municipal greenery"), but it's not grass or trees.

Very well put.

Christoph Hormann

More information about the Tagging mailing list