# [Tagging] Verifiability wiki page: "Geometry" section added

Wolfgang Zenker wolfgang at lyxys.ka.sub.org
Sun Apr 28 22:07:17 UTC 2019

```* Tobias Knerr <osm at tobias-knerr.de> [190428 14:31]:
> On 28.04.19 13:51, Mateusz Konieczny wrote:
>> "A place=hamlet often lacks verifiable borders. Hamlets in farming areas
>> often have scattered houses and farms extending outward for several
>> kilometers. In this case the approximate center of the place may be
>> well-know, but the outer limits are not clearly determined,
>> so mapping as an area is not verifiable."

>> is a good description of a common situation. Are you claiming that it
>> never happens?

> I agree that this is a common situation. However, I'm claiming that we
> actually know _more_ than just the hamlet's center in that situation.
> There will often be houses that are well known to be part of that
> hamlet, and other houses (in fact, almost 100% of the world's houses)
> that everyone will agree are not part of that hamlet.

> So the world's houses and farms can be (somewhat simplistically) divided
> into 3 sets:
> A: Verifiably part of the hamlet.
> B: Verifiably not part of the hamlet.
> C: May or may not be part of the hamlet.

> In my opinion, verifiability is not a valid reason to stop people from
> attempting to map the sets A and/or B.

Let me suggest a way to map features with "fuzzy" borders based on A and
B as defined above: We could use a relation that contains one ore more
objects (nodes/ways/relationjs) that are "inside" or "part of" our
feature and zero or more objects that are "near by but outside" our
feature. Features defined in this way could be used for basically two
things only, labelling that feature on a rendered map, and searching for
objects "in" or "near" that feature (where "in" and "near" could return
the same result, as the exact border is "fuzzy" anyway).
Examples: A valley could have a river or stream "inside" and maybe
another river at its mouth or mountains around it as "nearby but
outside". A mountain range could have some peaks "inside" and maybe
valleys outside.
The most difficult thing here is probably finding a good name for that
relation type and for the "inside" and "near by but outside" roles.

Wolfgang

```