[OSM-talk] Worldwide non-surveyed tag edits

[Andrew Guertin]

On 06/11/2014 04:03 PM, Marc Gemis wrote:
> irc will only work when that is an established communication channel in
> that country. So please, do not make that a requirement. E.g. in Belgium
> the best way to contact other mappers is the mailing list. I'll understand
> that this makes it more difficult for non-Belgians to fix the tagging here.

Oh, I'd never expect IRC specifically to be a requirement. I used IRC to 
show that (in my opinion) for a low-risk change, "discussion" can even 
include even include methods that are quick, informal, ephemeral, and 
even that don't reach the entire community. Something similar would be 
face-to-face discussion if you had other OSMers present.

For a low-risk edit, the real value doesn't lie in having someone else 
confirm that the edit is correct. The real value lies in having someone 
else confirm that *the edit is actually low-risk*. If you run something 
quickly past a few people, one of them can say "That's probably right 
but I'm not 100% certain, can you post it to the mailing list first?" 
and that's where I think the maximum value lies in staying out of 
people's way but still being able to catch things like the 
beer_garden/biergarten change in advance.

> Should I contact other people when I correct my own tags ? Or when I did a
> resurvey of the area and saw that it was really a restaurant and not a
> restuarant ?
>
> Why aren't we imposing the same requirements for people that just trace
> from aerial images?

In all these cases, you're editing with some amount of knowledge beyond 
what's already in OSM. When I posted my thoughts, I specifically applied 
them to only the cases where people edit with NO knowledge beyond what's 
already in OSM. So I would not apply these requirements to any of your 
cases here.

Editing OSM without any outside knowledge can have value (cleanups are 
good!), but it's inherently risky and that, in my opinion, is why extra 
requirements are needed.

> What if such a person connect two roads while in
> reality they are not connected ? Or when 2 intersecting buildings are
> separated ? Does (s)he risks to have his/her changesets reverted or
> eventually get blocked as well ?
>
> I'll agree with Jochem that people that are "gardeners' (to use wikipedia
> terminology -- people that try to fix existing tags) have to follow much
> more rules and risk more severe punishment than "tracers". It seems like
> the latter can't do anything wrong, unless it's pure vandalism.
>
> Can anybody tell me why a surveyor or tracer is free to keep adding
> restuarants without punishment, but a gardener should follow a long
> procedure to fix that ?

"Gardening" carries the risk that, when done incorrectly, it's not just 
the map that's impacted negatively, but the community as well. When 
people see their work improved upon that's great, but when their work is 
discarded or even worse edited to something that's wrong, that hurts.

Surveying doesn't carry that risk at all.

Tracing *can* carry that risk. It's much less frequent, but it's there. 
We don't have any specific policy that I'm aware of for dealing with it, 
but we DO have warnings all over the place that when you're tracing 
without local knowledge, the data that's already in OSM might be better 
than what you can get from your imagery. As far as I've seen, the 
community experience has been that having these warnings and dealing 
with problems case-by-case is sufficient.

Imports--using an external data source without sufficient local 
knowledge--have very similar risk to gardening. We do have extensive 
policy for them.

--Andrew