[OSM-talk-be] rise of the voetwegen, part 2

[Glenn Plas]

On 28-01-16 08:16, joost schouppe wrote:
> Marc,
> If by "they" you mean "trage wegen vzw", I think they have a pretty good
> understanding of what does and does not belong in Openstreetmap. They
> actively promote OSM use amongst their public, which is a good thing.

That is true but I'm missing a general mapping strategy, there is some
widespread tag misuse IMHO.  I wonder if they actually consulted an
experienced mapper to determine the keys/tags that should have been
used.  I'm not saying everyone does a bad job but depending on the
mapper, keys differ.

I was planning to analyse and review all that work using overpass API
and do some gardening here locally.

> But if you mean mappers in general, I would say that we still could do
> more to guide people. TheTrage Wegen blog post mentioned above was
> originally meant to include a link to the wiki page I want to make, but
> because I'm so slow, that wasn't ready yet. I think it would be very
> useful to have a wikified version of our discussions here.
> Edge cases of what does and does not belong in OSM is one of the
> subjects in that wiki.

It's really a binary thing, either it exists and verifiable in the
field, or either it's not.  We don't record historic buildings that have
disappeared, same applies to those roads that are gone and merely exist
on paper.

> As to names and numbers, I really like Ben Laenen's proposal to put
> those in vicinal_ref=*. "Voetweg XX" should then only be put in the
> name=* if it is marked as such in the field.

The fact that voetwegen are getting a very general name also gives
problems when geocoding.  Chances are when you reverse geocode something
close to a voetweg, that the result will not be the real street we want
to know about, but the voetweg.  The word problem might be a bit harsh,
but noone really knows about them, people are interested in real
streets.   I don't think the term 'voetweg' belongs in the name tag.

> I don't see a problem with vicinal_ref being filled with data that
> cannot be verified in the field (but verifiable with open data). All
> data should be verifiable, but not all of that will ever be verifiable
> in the field.

That is actually a base requirement of OSM, it has to be physically
there.  I think Marc is very correct in saying ways that pass through a
house are a big nono (he means: they do not exist in real life).  So it
has no value for OSM.  Besides the fact that crossing a road with a
building is an error.

Glenn