[OSM-legal-talk] Geocoding as produced work (was: Proposed "Metadata"-Guideline)

[Eugene Alvin Villar]

On 9/23/15, Tom Lee <tlee at mapbox.com> wrote:
>>
>> I mean, nobody cares about a single on-the-fly geocoding result (this
>> easily falls under the "substantial" guideline) but if you repeatedly
>> query an ODbL database with the aim of retrieving from it, say, a
>> million lat-lon pairs to store in your own database, then how in the
>> world could this new database ever be *not* a derivative? Even if you
>> were to define a single geocoding result as a produced work, combining a
>> large number of them in a database would still get you a derived
>> database again.
>
>
> Can't the same argument apply to tiles? If you used tiles to recreate the
> OSM database (say, by tracing road geometry or by OCRing feature names) and
> then republishing under a different license, you would clearly be violating
> the ODbL.
>
> It seems as though the same approach can apply to geocoding: locate
> features to your heart's content, but if you use the results to create a
> general purpose geographic database that substitutes for/competes with OSM,
> you'll be in violation of the license.

A geocoding result is substantially different from a map tile. A
single geocoded result is arguably a single piece of data that is most
probably devoid of any copyright and by itself does not have any
database rights, while a map tile is a creative work that is clearly
copyrighted.

If you have multiple geocoded results, and if those results are
organized in a manner that enabled individual access, then you already
have a database (in the legal sense). I really cannot see how such can
be considered as a Produced Work when it is clearly a Derivative
Database when a source database was used to help produce the results.

Sets of map tiles, on the other hand, are not automatically a database
and because they are primarily creative works are considered Produced
Works. Just like a photo of a storefront may have embedded trademarks
in it, a map tile may have embedded data from a database in it too. It
is only when you extract the trademark from the photo, or the data
from the map tiles that you may possibly infringe on IP rights. Note
that the operative word here is 'extract'. No such extraction occurs
with geocoded results—they are pure data already.