[OSM-legal-talk] CC-BY-SA for source data, other license for rendered derivatives

[80n]

On 2/28/07, Richard Fairhurst <richard at systemed.net> wrote:
> Etienne wrote:
>
> > On 2/28/07, Richard Fairhurst <richard at systemed.net> wrote:
> >> What I'd say is that it is certainly possible to come up with a
> >> licence where the _illustration_ needn't be copyleft, but the source
> >> code would still be copyleft.
> >
> > I think this idea is worth investigating.  I think what you are saying
> > is that we need some way of making the rendered version publishable
> > under some other license but requiring the corresponding source data
> > of that rendering to be CC-BY-SA.
> >
> > I can see some interesting issues with this, but it seems to better
> > describe how most people would like to see the data licensed.  It
> > would maximize the use of OSM data while still protecting the source.
>
> Yes, spot on. It would also mean that additions to the data could be
> fed back into OSM, which CC-SA doesn't require.
>
> The key, I think, is that we require that the geodata in the "map"
> (rendered version, derived work, call it what you will) is still
> licensed under our terms. So:
>
> > 2) What would be the status of data that is digitised from a rendered
> > image?  Especially if the quality of the resulting data was a near
> > match to the original source data.
>
> The geodata is copyright OSM Contributors, and licensed according to
> our terms - no matter how you extract it.
>

So we need some way of distinguishing between "geodata" that is
somehow encoded in an image and the image itself.

Lets assume, hypotheically, that a sophisticated piece of software
exists that could take an image and reverse engineer it back into
something close to the original XML data file.  Such a tool would make
it easy to extract geodata from an image.  If someone publishes an
image in what way are they not then publishing OSM geodata?

We have to be able to say that when the data is represented in a
raster form (or something) it is not subject to the same copyright
restrictions, but when represented as "geodata" then it is.  I wonder
if its possible to express that in terms that would be unambigous
enought to specify in a license?  Is digitisation the term we are
looking for?  Or is the distinction raster vs. vector?

> > 4) What form of license would be appropriate for the rendered
> > versions?  What restrictions would necessarily be required.
>
> You would be required to impose the same licence requirements on the
> geodata. So you could sell (say) a proprietary cycle map as long as
> the underlying geodata was still available under OSM's terms.
>
> > 1) What constitutes a rendered version?  Is an SVG file a rendered
> > version or just a transformation of the source data?  How would this
> > be defined in a license?
>
> That's actually defined by local law, not by our licence. In the UK,
> CDPA 1988 I.I.1.9.3 refers ("in the case of a work which is
> computer-generated...").
>
> > 3) When OSM source data is combined with other data to create a
> > rendered version,  if the  rendered version is then published, should
> > there be a requirement to publish the other data under CC-BY-SA, or
> > can it remain unpublished?
>
> This is the crux, I think.
>
> What I'm suggesting here, as a way forward, is a "data copyleft"
> licence. That would require that the other data were published as
> ShareAlike/copyleft.
>
> Allowing it to remain unpublished is little different from full PD (or
> PD-with-attribution). Personally I'd prefer this, but other
> contributors such as Oliver disagree strongly. I think more OSM
> contributors, on balance, might be satisfied with a "data copyleft"
> licence than with any of the alternatives.
>
> cheers
> Richard
>
>