[Tagging] The showstoppers for mapping Scandinavian nature.
Anders Torger
anders at torger.se
Sat Dec 26 10:24:26 UTC 2020
After careful consideration I've decided to stop mapping nature, and go
back to casual road mapping like I did before. It's been intense couple
of months, made 600000 edits, reached place 24 in the world board, 1st
in Sweden (except a cleanup bot that fixed some vandalism we've had in
Sweden lately, which I was also involved in tracking down). This is the
most intense OSM mapping period I've ever had.
In Sweden/Scandinavia we are since 60+ years used to having very good
maps available to the public, first in paper form and now also digital.
We have a rich outdoor culture, we are out hiking, hunting, fishing,
photographing, mountaineering, snowmobiling etc. I have several paper
maps of the areas I use to visit, and also an app with the corresponding
maps.
However, while our local app and services use the geodata provided by
the government, any international service or app typically use OSM data.
That got me thinking, wouldn't it be cool if one could raise the level
of OSM data quality in Scandinavia so that it can work as a drop in
replacement for the maps we are used to?
I didn't really expect that OSM or current renderers would be ready out
of the box, so there would be some gaps to fill in. Indeed there were
some gaps which lead to some interesting discussions on this list and
elsewhere.
Unfortunately it turns out that it's impossible to fill these gaps, and
thus impossible to fulfill the goal I had. OSM-based maps will never be
able to replace our traditional maps, which in practice means that when
I use an OSM-based map I will still need to have an extra app on the
side with the real map to get complete coverage. That kills my personal
motivation to map Scandinavian nature with any sort of detail, as it
then just becomes a theoretical exercise with no practical end goal.
There are two show-stoppers that makes this impossible. Neither is
technical.
One is that as OSM is only interested in providing a geo database it
makes it very hard to discuss geo services, such as maps. Limitations in
data representation that cause limitations in what geo services you can
implement are considered as non-issues as OSM doesn't prioritize any of
these services itself, and thus it's next to impossible to make any kind
of progress in these areas.
The other more concrete showstopper is the purist interpretation of the
verifiability principle which makes it impossible to name nature in a
way that that can be rendered properly on a map. Confusingly enough,
some of this functionality has slipped in and is used in practice (bays
and straits), but turns out that's due to an individual developer going
against other developers' will, thus if it wouldn't for a glitch in the
development process it wouldn't be in.
In Scandinavian rural nature we have about 5 - 10 of these names with
"undefined" borders per 10x10 km square, so it's more than a few. Named
sections of forests, named sections of water, named sections of a
mountain or hill, named sections of broad ridges, named peninsulas,
named valleys (often extremely wide), named plateaus, long streams which
in some undefined place change name, huge wetlands which in some
undefined place change name, etc.
According to the purist interpretation of verifiability all these should
be named as points of undefined size, or maybe in some cases as a line.
(Actually you could make a polygon of these natural areas verifiable in
the purist way, just by making it small enough so that all points of the
polygon is verifiably inside the area, but it's not up for discussion)
Point-naming nature in Scandinavia is only useful for very small natural
areas, say up to 500 meters span. As soon as the natural area spans more
than that a point without size information is not sufficient for any
data consumer can represent it properly in relation to the small areas.
With this type of data in the database it's impossible to render a map
with names in nature properly so you are better off leaving them out
alltogether.
And that is what mappers have done in Scandinavia most of the time so
far, and also what I will do. Mapping nature with high detail but
leaving out names does not feel meaningful to me (as it still just means
that one will have to use a traditional map on the side), so I will
personally just stop doing landcovers alltogether.
I will of course keep an eye on OSM and how it develops in the future,
so if its strategy evolves in a way that makes mapping of these natural
features meaningful I'll get back to work. Until then, I'll only map the
things that OSM can do in well-defined ways.
I'll unsubscribe from the list now as I won't need to ask any questions
when mapping basic stuff, so if you need to reach me send an email
directly to me.
/Anders
More information about the Tagging
mailing list