[OSM-legal-talk] Updated geocoding community guideline proposal

[Paul Norman]

On 2014-07-14 8:15 AM, Martijn van Exel wrote:
> On Mon, Jul 14, 2014 at 8:44 AM, Alex Barth <alex at mapbox.com> wrote:
>> This is also how I'm reading this. Obviously the sticky point is the 
>> definition of what's a database in this sentence: "systematically 
>> recreate a database from the process". You can't abuse geocoding to 
>> recreate OpenStreetMap without triggering share alike. 
> The definition of 'substantial' is key here, isn't it? In one of the 
> examples I added, the result of OSM-based geocoding actions would 
> potentially be stored on a client in a collection of 'favorites' 
> together with other favorites that may be the result of tainted 
> geocoding. There's really two questions here - 1) is this collection 
> of favorites 'substantial' and 2) does this mixed storage trigger 
> share alike in an of itself?

Given that any database of geocoding results is going to be clearly 
based upon the Database [OpenStreetMap], and that any interesting uses 
of OSM are probably going to substantial, I don't see the definition of 
it mattering.

In most of the cases raised in the wiki page, there's a derivative 
database of geocoding results and some other non-derivative database of 
something not taken from OSM, e.g. non-OSM POIs with just an address. 
You then take this collection of data sources and create a produced 
work, e.g. a page showing what the user has showed.

Once you start taking actual POI information from OSM, not just 
addresses, then your POI database will also be a derivative of OSM.