[Talk-us] Changeset to revert (or defend?)

[Nathan Edgars II]

Richard Welty wrote:
>as far as Nathan's editing goes, i think Nathan needs to back off a little
>and be more cooperative when local mappers have different positions from
>his. this edit first, discuss later stuff is getting old.
Honestly, while there are some places, such as on the wiki, where it's
stated that we should be careful and respect the styles of local
mappers (which I try to do, except in cases where said local mappers
are actually tagging incorrectly, such as blatant tagging for the
renderer), this isn't actually how I've seen others editing. For
example:
*The TIGER import. Yes, there are still places where pre-existing
roads duplicate the TIGER roads.
*The recent bot elimination of duplicate nodes. There's been a bit of
complaining but no action on reverting it.
*Someone drove through an area and took notes, then changed ref tags
away from the consistent local style to match his notes, despite the
previous tags being just as correct per the wiki. Several typos were
introduced in the new ref tags.

Sometimes this is not a bad thing (while I'll admit the TIGER import
had problems, I think it was a net positive), but other times it can
be harmful. But what we really need is a way of marking certain
guidelines on the wiki as "yes, this has consensus, and you shouldn't
go against it, even locally". I tried a few months ago to get some
sort of consensus on highway classification (this has to be at least
partially subjective in the US, since there's no consistent polished
system otherwise:
http://wiki.openstreetmap.org/wiki/User:NE2/classification_FAQ) and
there was far too little interest for me to feel comfortable going
forward with the process. I, personally, don't care which way things
are done, as long as the guidelines are as well-defined as
practicable, and consensus is on their side (so that eventually actual
use will approach the guideline). But the current mishmash in many
areas helps nobody, making it harder to edit the data, process the
data, and find definite errors (such as, here, a long segment of
street with intersections tagged layer=-1 because somewhere along the
way it passes under a railway).