[OSM-talk] Good practice, and should we rely on defaults?

[Mateusz Konieczny]

Apr 6, 2022, 14:59 by zelonewolf at gmail.com:

>
>
> On Wed, Apr 6, 2022 at 7:59 AM Frederik Ramm <> frederik at remote.org> > wrote:
>
>> Hi,
>>  
>>  my thought is that, in general, the default should not be added to OSM. 
>>  For example, I don't want 95% of streets in Germany to receive a 
>>  surface=asphalt or a motor_vehicle=yes!
>>
>
> I tend to agree as a general rule, and I've been annoyed, for example, of StreetComplete prompting me with quests to record the surface value on every street I walk by.  Of course I can configure SC to turn off this or that quest, but it seems like it would be a good cue to editor software generally to de-emphasize prompts for users to add information that is normally assumed to be a certain default in a particular country or region.
>
> Another tagging that I think falls in this bucket (at least in my area) is when to tag lanes=*.  It seems like it would not be terribly useful to tag lanes=2 on every highway=residential in my town, while it would probably be appropriate to tag lanes=* for perhaps any highway=tertiary and above regardless of value.  I generally go with "common sense" on this kind of thing but it sure would be nice to be able to reference documentation on assumed defaults.
>
Personally I think that it makes sense to record typical values explicitly 
in at least some cases. 
Otherwise systematic survey is basically impossible and on seeing any
cobblestone street during survey you would need to check OSM is surface
mapped already (and it would need to be done by every mapper).

Especially with speed limits you need some way to mark 
"defaults apply here", otherwise every mapper in area would need to check
every such location on their own and coordination would be impossible.

Obviously, not every value should be mapped (say, motor_vehicle=yes on
motorways) in cases where it is blatantly obvious except extremely tiny minority
(what qualifies here is going to be depending on mappers and on edge is
going to result in discussions...).

Curiously, I am against explicit access=yes on roads despite similar arguments 
apply to it as to surface=asphalt (maybe because here it is more likely that missing
values will be mapped? And it is rather on minor roads of little importance?)

>> We need to establish good ways to make country-wide (or region-wide) 
>>  defaults available in a machine-readable way. Tagging those defaults 
>>  onto every single object in one country because the default in this 
>>  country might be different from the default in the next country would 
>>  (a) create too much data inflation and (b) mean an edit orgy every time 
>>  the default changes in a place.
>>
>
> Wikidata is a sensible place to store this information as it is here today, performant, and OSM data consumers are already using it to conflate externally-linked data with OSM.  It would be very straightforward to document something like a default speed limit in a city using wikidata.
>
Main problem here is that we would be putting critical part of OSM infrastructure
in system not maintained or controlled by OSM.

name-suggestion-index had already problems with some items used by it getting
deleted by Wikidata admins as were no considered as passing some standards there.

It would also make mandatory for OSM contributors to accept rules of some external
project if they would want to fix some issues with that data

> It would be very straightforward to document something like a default speed limit
> in a city using wikidata.

Can you give example of that?
I expect that for people who have no experience with Wikidata it would be not
be very straightforward.

I have some experience and I am not sure how it would be done (which property
code would be used?)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20220406/94c29a3d/attachment-0001.htm>