[OSM-talk] Ground truth for non-physical objects

Tomas Straupis tomasstraupis at gmail.com
Sat Dec 15 12:30:24 UTC 2018


2018-12-15, št, 13:57 Andy Townsend rašė:
> If I want to find the border
> between Ireland and Northern Ireland, for example, I might not (yet)
> find anything stopping me driving through but I will see something along
> the lines of "speed limits now in mph" or the reverse.

  And then the borderline in OSM will be drawn by simply connecting
those scarce points on the roads with straight lines?

> The fact that we can't get some boundaries from an on the ground survey
> doesn't mean that we have to rely on "documents" for all of them, and
> for a good reason - "documents" often contradict each other, even from
> the same organisation.

  If you find an error in "documents" - why don't you inform the owner
of the document so that they could fix the error (and maybe fix data
in OSM until official document is fixed)? My opinion is that
OpenStreetMap should COLLABORATE, not ISOLATE itself. Collaboration
does NOT mean OSM will become just a fusion of data from different
official documents.

  The fact that there are some errors in official documents means that
we want to resort to guesswork and extremely simplified geometry for
ALL non-physical objects?

  How many border vertexes are actually mapped by observing physical
reflections in OpenStreetMap? 0,00001% or less?

  And if I was to use an "ad absurdum" argument type: a lot is mapped
by using so called "satellite imagery", which is not a photography,
but a number of them processed and merged to produce an
"Orthophotographic MAP" which is a DOCUMENT. Where will it lead us to?
To methods of ~2006 when in order to map a building we had to walk
around the building with GPSR (with some distance in order to minimise
building obstruction to GPS signal), then measure the building and use
that information to map ONE building? And note, building is a physical
object which can be observed directly, but is quite difficult to
measure with tools most mappers have...
  (Note: the point of ad absurdum argument in discussion is to point
out flaws in initial statement, not to ridicule, so please do not take
is as an offence)

-- 
Tomas



More information about the talk mailing list