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

Martin Koppenhoefer dieterdreist at gmail.com
Sat Jun 9 15:31:26 UTC 2018



sent from a phone

> On 8. Jun 2018, at 10:22, Mateusz Konieczny <matkoniecz at tutanota.com> wrote:
> 
> I would start from easy wins, for example why we have both FIXME and fixme tags?


+1
We have done something similar in the past: unification of yes, 1 and true for example. 



> Why we still have wikipedia:pl, wikipedia:en duplicating wikipedia keys?


they should not duplicate the wikipedia tag. These tags are meant for cases where wikipedia=* and interlinked pages of other languages are not sufficient. There can be several reasons for this, e.g. there are several articles for the same thing in osm, or the article in one language cannot be linked in wp (or simply isn’t yet linked), but should be linked in osm.

I don’t think we should change a significant amount of tags and restructure a lot of things, but there is a tiny fraction of unfortunate tags that could be rethought, in particular some few values of landuse, tourism and natural.

I also believe in principle in the system of tags emerging from mapping the world together, but it would require only or mostly mappers like you, who read the tagging mailing list and show general interest in tags. And we have already lost a lot of potential by imports. Imports generally work very bad together with a system like ours for creating tags (a system to describe the world). Most actual mappers (mapping by distinct user) don’t see tags, it’s all abstracted away from them behind presets and preset names (i.e. typically one word for the key/value pair, usually the value). Want an example? church and church. Both were presets for quite some time.


What ends up in the presets is decided by a group of programmers in Josm (ultimately by Dirk I believe, but he is typically not involved in preset questions AFAIK) and according to my experience just one maintainer in iD (Bryan). The only thing you can do to change a preset, remove it or add a new one is asking politely and explaining your reasoning, but there is typically no public discourse about presets, and while you can create alternative preset rules for josm if your plea was rejected, in iD there is nothing.

I do not intend to blame Bryan or Dirk for this, I can see it is simply because they are in the position where the decisions ultimately become code, and as there is no other process established yet, it is natural that it is like this, but I believe we should create a comunity led process to document and manage what ends up in presets / tag completion and maybe also in preset translations (many people use translations). For example while iD generally uses a sensible system to determine interesting tags (by usage), there are some “obscure” parts in the process which lead to some tags filtered out nonetheless they are used in bigger numbers, or in other occasions have been introducing tags which had no prior usage.
It is also clear that looking only at numbers is biased as well and cannot be the only answer to the problem, because some things occur more often than others, but being relatively rare doesn’t mean always something is not map-worthy. E.g. continents or country capitals are relatively rare, trash cans or traffic lights occur more often. And somehow it must either be determined which way of tagging is preferable or alternative ways should be presented and explained, what is generally not happening now.


Cheers,
Martin 


More information about the Tagging mailing list