[OSM-legal-talk] Exception in Open DataLicense/Community Guidelines for temporary file

[Tobias Knerr]

Richard Fairhurst wrote:
> David Groom wrote:
>> However where you *don't* add data, but merely process the OSM data, 
>> either by extracting some sub-set of it, or simply by transforming it from 
>> one form of database to another, then what is the point of requiring 
>> compliance with ODbL clause 4.6.
> 
> You seem to be assuming that compliance _would_ be required. I'm not sure
> whether it would be. It's possible that the diff between (say) a full planet
> and a regional subset isn't qualitatively Substantial and therefore doesn't
> need to be released. That may also be true for transforming from one form of
> database to another.

The only motivation for data SA I can somewhat understand is to open up
data that can be contributed back to OSM.

Even leaving aside that it would still be problematic for us to accept
ODbL-compatible, but CT-incompatible data, this intention does not
provide a reason for the (likely) requirement to publish
* a database calculated exclusively from content of the original ODbL Db
* a database calculated from content of the original ODbL DB and another
Db that is publicly available under an ODbL-compatible license.

Thus it would be great if ODbL did not require publishing the derivative
databases in these cases, but unfortunately, I currently have to assume
that it does indeed require publication.

> (In any case, even if it is, the requirement could be
> fulfilled by putting a single line on a wiki page somewhere with the
> Osmosis/ogr2ogr/whatever command line you used.)

That might work well for a member of the OSM community who intends to
offer a regularly-updated map or something. It's not a nice solution for
a casual user who does not want to set up a service of any kind, but
just uses a piece of software he just downloaded to create a produced work.

At the minimum, that user would also need to make sure that the program
he used remains available somewhere as long as he wants to publish his
produced work, wouldn't he? After all, it would be pointless if people
could just say that they created a derivative database using a piece of
software that you cannot download anywhere.

> But let's say that you produce a Derivative Database that contains a very
> cool optimised routing graph from OSM data: no new data, just processed. The
> share-alike viewpoint might be that it's genuinely useful to OSM to have
> either the full Derivative, or the algorithm. That would be "the point".

As for that hypothetical routing graph, I don't understand why
share-alike-for-data supporters would want it. It seems to make no sense
to request share alike when the routing graph was created by publicly
available software. But even some evil proprietary routing graph would
only be of limited usefulness without access to the similarly
proprietary routing engine.

Altogether, it seems that ODbL does burden creative users of OSM with
significant restrictions, even though in many cases there is no reason
why anyone would even want that databases they are forced to share -
because it will usually be much more practical to just use the original
databases.

> (Again, IRMFI, as you probably know I'm a PD sort myself.)

It's funny that several of these legal threads contain mostly exchanges
between PD people and PD people who, RMFI, explain the share-alike pov.

-- Tobias Knerr