[OSM-talk] Lithuania fancy tagging rules (was: Re: Andy Townsend is above local community?)

[Tomas Straupis]

  Mateusz is ignoring the main issue that he having no relation to
Lithuanian community (and again without any discussion) is altering
wiki which is supposed to represent OUR opinion. He is focusing on GPS
requirement which is obviously a remnant from the times when we had
only landsat images and position accuracy was very bad. It is
obviously outdated but I see no point in editing wiki as Mateusz is
all over wiki doing whatever edits his mind thinks of.

  Here in this blog post you can find an image of keepright errors
placed on a map (you do not need to read the text). And you can
clearly identify Lithuania:
  https://blog.openmap.lt/2016/03/25/klaidos-baltijos-juros-regione/

  OSM data is not just a pile of bits. It has it's process:
requirement for change is identified, data is collected and
added/updated, requirement for update/deletion is identified, data is
checked updated/deleted. Humans can do that, but then the result is
very heterogenic - result coverage is very dependant on existence of
active mappers in the region. Therefore some automated (or
semi-automated) system must help humans to identify missing or
excessive data and you need high quality data for this to work.

  QA is also part of this process. You cannot just expect that all
errors will be noticed by casual mappers checking some specific
position in a reasonable time. You need both patroling changes of new
mappers as well as constant re-runing of QA checks (Note that
patroling idea was later adopted by Mapbox). This is a known fact in
testing: the faster you notice the problem - the cheaper it is to fix
it. By fixing I mean fixing the data as well as educating the user on
what the actual problem is.

  If there was no patroling, error like this (12 hours old, adding non
existing motorway) would not be noticed for days or maybe even more:
  https://www.openstreetmap.org/changeset/121994892

2022-06-06, pr, 11:29 Frederik Ramm <frederik at remote.org> rašė:
> It is important to remember that QA tools are only made by humans and
> these tools can be buggy (or intentionally wrong) too. For example, a QA
> tool that would ask you to upgrade a 20-metre stretch of tertiary road
> to secondary just because it has surface=paved is certainly buggy and
> requires fixing.

  Yes exactly, and there were three options: wait some years until all
these roads will be paved, add exception on rules themselves, augment
data to identify rule exception. No big deal. Why was Andy unable to
contact Lithuanian community and say that waiting is now not
acceptable, lets agree on a solution acceptable to both sides?

  People (most of whom do not have any actual information on that)
claim different things on my role in Lithuania mapping and yet you're
afraid to try to contact those "oppressed" other Lithuanian mappers
and actually ASK them what they think.

  Saying that most of my comments on bad changes raise a complaint at
DWG is either a lie, either I do not see full numbers of it (why can't
this information be forwarded to me?). I see most people thanking me
for explaining, for fixing the errors etc. I've participated in
commenting 1,429 changesets and can remember roughly 10-20 occasions
when DWG contacted me (and in most cases DWG actually agreed that I
was right).

P.S. I've raised a question on talk-lt if rigid quality control is
wanted by the rest of Lithuanian community, we'll see.

-- 
Tomas