[Tagging] Feature Proposal - RFC - Discouraging the use of deprecated schemes

Martin Koppenhoefer dieterdreist at gmail.com
Wed Mar 24 09:04:45 UTC 2021

Am Mi., 24. März 2021 um 01:04 Uhr schrieb <mail at marcos-martinez.net>:

> Frederik wrote:
>    - There will not be one consistent tagging system that works for the
>       whole
>       planet. I think people will agree on some things and chose to go
>       their
>       own paths in others, and that's ok.
> I disgree and I think many people misunderstand the point. I'd never
> impose ONE single tagging scheme as such for the whole world. The world is
> indeed too complex. But we can agree to apply different tagging schemes in
> different contexts.

I agree with Marcos here, a consistent tagging scheme does not mean we will
use the same set of tags globally  but that usage of the tags is
consistently applied (the same tag is used for the same thing as opposed to
using the same tag for different things / with different intended meaning),
one actual question is in the detail: how long is a thing to be considered
"the same", what are its essential characteristics, or in other words: when
do things start to be different and should be tagged differently? We do not
need "consistence" in the sense that only one tag is used for the same
property (no real problem, or a very minor inconvenience, if we use phone=*
and contact:phone in parallel), but we should strive to intend the same
meaning when we use the same tags (when we don't, things break apart).
While we could have different regional standards ("foo"-tag means bar in
Europe and foobaz in Oceania), and it might work (to some extent already
done, e.g. typical tags for bicycle details, ...) we should try to limit
these cases to a minimum because it makes life harder for everyone, better
use different tags all together if different concepts should be expressed.
There is a lot of difference (in the detail) due to cultural differences.
For example if you see a German petrol station in an urban area, you can
probably assume it has a 24/7 shop where you can buy some food (typically
including fresh milk), beer/alcohol and tobacco (because the German law
says they are exempt from Ladenschlussgesetz and can sell "travel
supplies", as everybody knows alcohol makes a perfect travel supply). In
Italy, petrol stations in the city almost never have such a shop, they are
often completely automatic and you will not find any staff. They will never
sell tobacco unless they are also a tobacco shop (only in the countryside).

> Which means we need more complex rules - not the absence of rules. If the
> Italian community decides they handle address tagging very different from
> other countries (for those who don't know: "Housenumbers are usually
> assigned to the access points of a property (e.g. a gate on the perimeter",
> even if it is hundreds of meters away from the actual building) that is
> fine. But this needs to be documented then and regarded as standard.

It is documented in this detailed way, but it is also completely consistent
with a definition like "add addresses to the object where they apply to".
In Germany, addresses are added to building plots (even plots without a
building on it), while in Italy, addresses are associated with doors, gates
and entrances. Actually if you look at the details from this point of view,
it is the Germans that deviate from the definition, because they mostly do
not add the addresses to building plots (unless it is a polygon POI) but
inherit it to a building on the site. Our tagging is consistent, it's the
way addresses are associated in the real world that is different.

> It doesn't make sense to have Italians applying this rule in Germany and
> Germans tagging in Italy as they are used at home!

they're doing it anyway, sigh. One reason are tools. A lot of tools insist
you add addresses tags to a building, nudging you to do it. Tools are very
powerful in OSM, the hints from bug linters are often not questioned,
editor presets are assumed to represent "consensus", with maproulette you
can perform distributed mass retagging, etc. This said, some Germans are
actually mapping like this (in Germany), adding address information on
nodes (often entrances).

> So, yes, I vote for giving more "power" to the outcome of votes. During
> mailing lists debates and the voting process is where consensus should
> happen. Visible, countable, open to everybody. BTW, we really should
> consider a proper community communication tool such as e.g. Loomio for the
> whole community. These mailing lists are a real dread and probably one of
> the reason why there are so few who participate.

Email has the advantage that everybody already knows the system, many have
a client already setup (or use a web client), it's seemlessly usable on
mobile and desktop. Any other tool has a higher barrier of entry, as you
would have to learn it.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20210324/12496020/attachment.htm>

More information about the Tagging mailing list