[Tagging] How to overcome lack of consensus

Peter Wendorff wendorff at uni-paderborn.de
Mon Sep 16 18:13:24 UTC 2013


Hi,
let's tackle the "problems" you mention one by one:

Am 16.09.2013 16:41, schrieb Matthijs Melissen:
> Dear all,
> [...] The lack
> of consensus does cause problems for the Openstreetmap community, though.
Most often it only cause problems for consumers of our data, not for the
mappers. Nevertheless you're right, it would be good to have clearly
defined rules in some cases.

> Therefore, it would be good to have ideas or procedures on how to create
> consensus.
> 
> There are currently quite a lot of OpenStreetMap features for which there
> is no consensus on how to tag them. Some examples (but I'm sure there are
> many more):
> - What is the difference between highway=footway and highway=path?
IMHO:
A footway is a footway, mainly or only dedicated to people using it by
foot. Any exception should be stated as such (e.g. bicycle=yes if bikes
are allowed).
A path is a path - which might be used by pedestrians/walkers, but not
only and not necessarily - you may guess what's allowed and what's not,
but you should not rely on anything without taking additional tags into
account.

That's for you as a mapper.
For the data consumer that might be different:
- a path may be estimated to be accessible by foot if not stated
otherwise, because (!) there's not enough detailled data to be able to
rely only on existent data.

> - What is the right scheme for tagging public transport?
(I'll skip this)
> - Is an unsurfaced residential road a track?
At first: unsurfaced? there's no surface? should be impossible because
if there's no surface then there's no road ;)
If you mean unpaved: it depends how you map it.
I personally would add a residential as highway=residential, independent
on the surface. In many development countries residentials are nearly
never paved, so that's only a guess (depending on the region) if not
stated otherwise.

A residential therefore IMHO should be highway=residential, and if you
want or if you think it's useful as it's not what anyone would estimate
there, add surface=* to it, like surface=unpaved, surface=mud,
surface=sand or whatever matches.

Again: On the consumer side a residential e.g. in most parts of Europe
would usually be guessed as being paved with paving_stones, asphalt or
concrete, but that's only an estimation again, so to make it clear, map it.

> - Should we use shop=betting or shop=bookmaker?
> - Should we use shop=fishmonger or shop=seafood?
> - Should we use office=estate_agent or shop=estate_agent?
no idea - never used it.
> - Should we use shop=tailor or craft=tailor?
Depends...
Is there a tailor inside? or are they only selling what a tailor produces?
Is the shop directly connected to/containing the workshop? or is there a
shop selling the stuff and another place where it's crafted?

The craft tag is there to get exactly that difference, I think.
A better example might be confectionary.
A shop=confectionary sells confectionary, but it does not necessarily
produce them, there does not need be someone making them - instead they
could by it.
A craft=confectionary on the other side may sell their own products - or
not.
That leads to three possible combinations, all valid and all telling
different things:
- shop=confectionary + craft=confectionary: selling and making confectionary
- craft=confectionary: making confectionary, but not necessarily selling
them via a shop (they may sell them via internet, or only to other
shops, or they may sell it in their shop somewhere else)
- shop=confectionary: they buy their stuff somewhere and sell it to you.

> The lack of consensus becomes clear by the fact that there are
> discrepancies between documentation on the wiki, the outcome of a voting,
> actual use (as documented on Taginfo, for example), and what editors and
> renderers support.
> 
> The lack of consensus creates several problems. These problems include the
> following.
> - Multiple parallel tagging schemes and unclear documentation creates
> confusion for newcomers.
That's true, but how do you want to tackle that without limiting the
freedom to invent new and better tags?
The multiplicity of parallel tagging schemes comes in where there's no
best variant - or where there's a lack of documentation.
Lack of documentation: go and document ;)
Confusion to newcomers: Explain it. Explain why there are several
possibilities, explain that they themself may invent own tags, and give
them taginfo and the wiki to search for reasons to use one or the other
variant.

> - Users are often advised not to follow the documentation on the wiki, and
> to look at Taginfo instead. This makes the wiki useless. It also leads to
> the fact that hardly anybody bothers to edit the wiki anymore.
True: Do you have any idea how to solve the problem of documentation not
reflected by practice or the other way around?
In general both has to be seen. A documentation in the wiki might be
outdated - then mappers should avoid following it, it might be missing,
and it might be plainly wrong. On the other side sometimes there is good
documentation but the documented features are very rare or the tagging
is new, so that there's not yet "current usage" you could find on taginfo.
Therefore both is necessary, and yes, it's not easy to see both and to
decide what to do, but how to solve it?

If you would put the rule to follow documentation "only", what if Mapper
Max invents a new tagging scheme, documents that but nobody uses it,
because it's a bad idea?
If you would put the rule to follow actual usage, what if a group of
mappers get a really great new tagging scheme, documents it, but it's
not used yet? Nobody would (and could, following your rule) use that new
scheme as it IS not used yet.

Both is necessary, and the connection between wiki and taginfo (taginfo
linking the documentation and the wiki showing stats from taginfo) is
one step to the right direction.

The next step which has to be worked on continuously is to extend and
polish documentation in the wiki.

regards
Peter




More information about the Tagging mailing list