[OSM-talk] Abandoned Rails

moltonel 3x Combo moltonel at gmail.com
Sun Aug 23 21:25:18 UTC 2015


On 23/08/2015, Balaco Baco <balacobaco at imap.cc> wrote:
> I don't think so. Wrong data happens to Bing Maps, to Google Maps and
> probably to any other map we can obtain, electronically or in a paper.
> Maps have dates attached to them - or should have, most of the time. The
> fact is that: if I browse around my city, looking for streets and
> brindges created in the last fews years, I will see mistakes (as I did
> before).

Yes, all maps have errors, wether it is outdated data, data that was
never right, or missing data.

> But it's better to have there what existed, as it was before,
> than have just an emptyness in the area.

How so ? Say I'm walking along an old railroad which OSM led me to
believe continued for 10km, but is impassable at various points
including a wheat field and a housing estate. Or I'm heading to the
convenience store only to find it has closed years ago. Outdated data
is wrong data, it is misleading and lowers the overall quality of the
map.

> Until someone fix it
> (hopefully) or at least mark it as old, potentially wrong (without
> deleting until an update is made!).

That's just the normal mapping workflow, nobody is arguing against
this. Nobody is proposing to "delete first, improve later". We make
the best map we can, within the bounds of our knowledge and time
constraints.

> In the context being discussed here, recent changes should also cause
> data deletion, but that's wrong, in my opinion. Data may be *replaced*
> with newer data, with everything that's needed.

Same thing here, we improve the map as much as we can. But maybe you
don't have enough time right now to map everything, or meadows are so
far off in your todo list that you never bother with them. But leaving
known-outdated data in place just because you can't yet make a fully
detailed mapping of the area doen't make sense either.

> If people are doing something that is fiercely against the community
> idea of OpenStreetMap, that it could be deleted. But that really seem
> far from the truth.

There are very few rules in OSM, but one of them is that we map on the
ground and that we don't map historical features/events
http://wiki.openstreetmap.org/wiki/Good_practice#Map_what.27s_on_the_ground
http://wiki.openstreetmap.org/wiki/Good_practice#Don.27t_map_historic_events_and_historic_features

Ways like https://www.openstreetmap.org/way/366610457 (and presumably
many other railway=dismantled) fail these checks.

> So that should be preserved, and its deletion, if
> decided to be made, should give a reasonable opportunity for the data
> contributors to backup that data - so their work is not lost, but may be
> used somewhere else.

Yes, giving a heads-up to a mapper when you edit a lot of his work is
good etiquette. As for backup, the data is versioned in the db, you
can always get the old data back.

> Further, I'm not one of the users that would be confused with that. I
> would find it unsual to see in a map. But being tagged and noted
> somehow, should not be a problem at all. And to say the users who would
> be confused with it are the "majority of them", is an vague argument you
> do just to give some apparent strength to your idea. And I repeat: I
> would not be confused with it, I don't think the majority of users would
> be.

Since OSM has always had a policy of containing only current data, it
stands to reason that the majority of users only expect to find
current data in OSM (or rather that anything that isn't current
anymore needs to be fixed, and that it was current when it was added
to osm).

Wether you get confused when stumbling uppon data which violates that
rule depends on what you're doing with the data. Remember that
interpreting osm data is actually a lot of work. Very few people have
the manpower to verify what railroad=dismantled actually mean to
decide wheter they want to use or filter out that data. Most of them
will just match railway=*, plus perhaps some special cases for
railway=rail and railway=subway. Now they're looking at historical
data without even knowing it. They are confused.

> P. S.: this mailing list does not add a "Reply-to" header to mail
> messages, as I'm used to. So I initially sent the answer to just one
> person. This should be changed - may it not confuse the majority of
> users!?

It normally does, not sure what happened here.



More information about the talk mailing list