[Tagging] Fuzzy areas again: should we have them or not?

[Anders Torger]

Thanks, very good blog post.

Let's discuss examples. As far as I can understand your blog post states 
that naming a wetland whose borders that can be clearly identified 
through surrounding forest is compatible with verifiability, just put a 
name on that wetland polygon.

However naming a wetland that is so large that it has one name in one 
end, and another name in the other (like in Muddus national park) is not 
compatible as I understand your description of verifiability *if* we put 
names on polygons, but if we place the names in points it is okay, 
assuming the points are not too close to a border. The reason for this 
is that the polygons would be "non-verifiable geometry" as examplified 
by the bay polygon in Norway in your blog post. Not to the full extent, 
but the split is certainly non-verifiable as there is no such border, 
just as where bays transitions into open sea. In the same section you 
say that it is okay to name these things using a point, as it's 
verifiable that it exists, but the borders are not verifiable.

The same with subsections of forest. Putting a named point there is 
okay, making a polygon cutout and name that like I have done is not 
okay.

The point must of course be placed such that it can verifiably be said 
to be inside the area it represents, not too close to those fuzzy 
borders, then it becomes unverifiable again.

I'm not against this concept, it has logic. But in the end, how do we 
get from there to useful maps and queries for these features? I care 
about that, maybe I shouldn't but I do. I think that if we allow in 
addition to set a feature size (like hamlet/village etc) points would 
work from a cartography perspective, but not for queries, and we add a 
layer of verifiability complexity, that maybe makes it close enough to 
make it worth stepping all the way to fuzzy polygons.

I also think one can make fuzzy polygons "verifiable" just through 
definition. It's verifiable that they are fuzzy. The rough extent is 
verifiable. Exact borders are not, but that can be seen as a non-issue 
as they are just defined to be interpreted as inexact.

/Anders

On 2020-12-22 13:29, Christoph Hormann wrote:
>> Anders Torger <anders at torger.se> hat am 22.12.2020 11:28 geschrieben:
>> 
>> OSM is a geodatabase, with a design that makes some geodata suitable 
>> for it, others less so. [...]
> 
> To make this clear, in particular also for those who might be idly
> reading this and could get the wrong impression:
> 
> While OSM is also a geo-database it is foremost a project of
> cooperation between people from very different parts of the world with
> different cultural backgrounds and different views and perception.
> 
> The reason why we have the principle of verifiability and many of us
> put strong emphasis on sticking to it is not technical but social in
> nature.  OSM lacks the social mechanisms to create and maintain
> non-verifiable information in a globally uniform database without
> abandoning the principle of egalitarian cooperation across language
> and culture barriers in favor of a social and cultural hierarchy like
> in Wikipedia.
> 
> See also what i wrote in
> 
> https://lists.openstreetmap.org/pipermail/tagging/2020-December/056829.html
> 
> as well as what i wrote more than two years ago meanwhile in:
> 
> http://blog.imagico.de/verifiability-and-the-wikipediarization-of-openstreetmap/
> 
> --
> Christoph Hormann
> https://www.imagico.de/
> 
> _______________________________________________
> Tagging mailing list
> Tagging at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging