[Tagging] Facts and opinions

Tobias Knerr osm at tobias-knerr.de
Wed Jan 9 23:13:53 UTC 2019

On 07.01.19 16:12, Bryan Housel wrote:
> we can’t use the same key `service=*` to contain both things like `tyres` (a few thousands) and `driveway` (a few millions).  Sorry, but the `service=tyres` has to go.

These two different meanings of 'service=*' would not need to coexist on
the same element, so it's not impossible to use the same key. I happen
to agree that it's not a good idea, but it's not a foregone conclusion.

Therefore, it makes me a little uncomfortable that this kind of change
would be promoted through iD, instead of convincing mappers to make an
active, informed decision to switch to a new tagging scheme. I've been
silent on this recurring issue so far because, well, most of your
decisions have actually been quite reasonable, and it felt a bit silly
to object based on purely hypothetical concerns. Also, I don't feel
strongly about vehicle services in particular.

But with great power comes great responsibility, if you forgive the
stale quote. And while I'm not opposed to doing some sanity checking
(i.e. not automatically supporting poorly thought out tags just because
they are common), I do feel that the default editor on osm.org should
generally only promote clearly established tagging styles.

> I encourage everyone to just disregard everything that’s on the wiki and go by what taginfo says as far as how the tags are used and what the accepted values are.

The wiki is an invaluable source for understanding OSM tagging, and I
use it all the time during mapping and when coding software that works
with OSM data.

Taginfo is an awesome resource as well, and I use it almost daily, but
it cannot fully replace the wiki. It tells you that foo=bar has been
used thousands of times, but it doesn't tell you what that tag means¹.
It also doesn't tell you about the conventions for its use (default
values, directionality, lots of other essential details). Ultimately,
Taginfo isn't documentation – the wiki is.

Besides documenting current tagging practice, the wiki is also a useful
tool for coordinating and spreading new ideas (even though the specifics
of the process can be controversial at times). If you're not a software
developer or one of a few highly respected community members,
discussions on community channels and wiki proposals are pretty much
your only good options to make your genius tagging idea known to the
world. Without this first step, that idea is unlikely to get enough
traction to even show up in Taginfo to a meaningful extent: Using the
tag yourself only gets you so far.

For all these reasons, I consider the wiki a key asset to our project.
As a result, I spend a lot of time improving it, as do many other
community members. It hurts to see that some developers of core OSM
infrastructure seemingly value these contributions so little. To me,
people discussing and documenting our data model are a vital part of our
community. So are software developers, of course! It's my belief that
the project can only thrive if there's mutual respect between these groups.


¹ Taginfo actually does provide a definition, but that's because it
extracts them from wiki pages.

More information about the Tagging mailing list