[geo-discuss] Re: Open Source geo data

[Jo Walsh]

dear David, Frank, all,
On Mon, Mar 20, 2006 at 10:19:12AM -0500, Frank Warmerdam wrote:
> >I'm still working through a discussion paper on Open Source geospatial 
> >data.

I'm glad to hear of papers being written on what seems like a pressing
and topical subject, and i would look forward to seeing your when it is
written, if it's going to be publically available.

> >So far, the response from users and vendors has been overwhelmingly
> >positive.  In the paper, I'm getting down to a few of the specific issues
> >that Open Source data raises.  Specifically, I'd like to point to any work
> >that has been done on open licensing for geospatial data.

I can offer you an overview of past work and recent discussion that
may provide useful context for you, if you have not seen it. 

Daniel Faivre began work on a Public Geodata License in 2003. It is
very strongly modelled on the GNU Public License for software. It has
a strong clause dictating that distribuition of a "derived work" must
be available on the same terms. Work on the license has been picked up
by a Canadian group. http://cemml.carleton.ca:8080/OGUG/pgl  
It provides a good discussion candidate for the most "open" position.

Richard Fairhurst of http://www.geowiki.co.uk/ recently wrote a
worst-case-scenario essay on the impact that such a ShareAlike clause
might have on peoples' willingness to distribute data under an open
license. How "derivative work" with respect to geodata is defined, 
is a hard question, as Frank notes below, and there's commentary and
discussion of this below Richard's article: 
http://www.systemed.net/blog/entry060311122655.html
http://www.systemed.net/stuff/sharealike/

His suggestions for resolution include a 'Lesser Public Geodata
License' modelled on the LGPL: "Permission to use a particular library 
[data set] in non-free programs [maps] enables a greater number of 
people to use a large body of free software [geodata]." and some
useful ideas about a "Collective Work" status.

This part of the discussion is really only beginning; it is an area 
which i hope the OSGeo Foundation's Public Geospatial Data Project 
will be able to help formalise, and produce a discussion candidate 
LPGL model license.  

The OpenStreetmap project, probably the biggest body of ground-up
contributor-created open geodata, uses the Creative Commons
Attribution-ShareAlike license verbatim.
http://wiki.openstreetmap.org/index.php/OpenStreetMap_License

Last year the Urban Tapestries group produced a license candidate with
advice from Creative Commons UK, for distribution of geodata limited
to "non-commercial" use. This isn't much use to those who want to see
real economic value creation from open geodata, but it provides a
relevant reference point on the open geodata licensing 'spectrum'. 
http://www.okfn.org/geo/geodata_cc_license_draft.rtf 

In thinking about open access to data issues drawing from a software
licensing model, i find the terms that Rufus Pollock uses in the Open
Knowledge Definition to be helpful: access to the work, the right to
redistribute the work on any terms, and the right to re-use the work.
http://www.openknowledgefoundation.org/okd/definition.html

> >Do you know of any work that's been done on data licensing that will allow
> >users to: 1) add value to proprietary geospatial data without constraint;
> >and/or 2) use or contribute that value-add data to a broader spatial data
> >infrastructure?

> Frank Warmerdam writes:
> I think a license that would allow adding value to proprietary geospatial
> data *without constraint* would be difficult to envision while retaining
> proprietary rights to the original data.  The problem being that many
> kinds of derivative work will contain much, or all of the original data.
> In a digital world this can be very problematic.  So for instance, it is
> difficult to publish online map images that include a proprietary image
> dataset without having to couch the map services in legal restrictions that
> significantly inhibit it's usefulness for anything other than simple
> interactive viewing.  This is why Google Maps has an acceptable use policy
> that inhibits use for many serious purposes.

I must admit that the question is phrased in a way that makes it hard
to attempt an answer. If data is proprietary, doesn't that imply that 
there must be constraints on how it can be re-used?    

I would agree with Frank, to the extent that this question can be
answered. Google Maps is a pretty good example here. They are not a
data collector or provider, but a "reseller" (except they are selling
advertising, and not location services directly). The geodata license
they have precludes them from re-offering that data in raw form, and
means they have to place a lot of restrictions on what can be called a
"derived work". They don't offer free address geocoding services, for
example, because that would contravene their own terms of use with their
suppliers. Where the distribution chain is so long, NavTEQ, Teleatlas
et al don't have an incentive to offer more freedom to users of their
data, so their users can't offer freedom downwards.

Google Maps in the US uses a lot of aerial imagery which is public domain,
provided by state or federal government. When they combine this with
proprietary geodata sources, the whole offering becomes non-free, no
matter what the licensing constraints or lack thereof in the parts.

A lot of information can or could be extracted from Google Maps that
would be a great deal of use to open source collaborative mapping and
spatial annotation projects, but people have shied away from that,
because they don't want to 'infect' their offering with copyrighted
data. It's a shame, because there is a lot of potential value to be
added, that is not being created. Yet this isn't really the problem of
the map service providers, but of the data service providers, when as
now, they tend to be different entities.

I am not a lawyer, but i would really like to meet one and have a
serious conversation that can be shared, about the legal meaning of a
"derived work" in the context of geographic information. 
 
> I'm not sure what the answer to your second question is.  Generally speaking
> if the source datasets have appropriate licensing and acceptable use 
> policies
> then value-add products can be contributed back to a spatial data
> infrastructure environment.  Alternatively "value add" can be provided as
> invokable algorithms that clients can invoke on datasets they have the 
> rights.
> However, I am at best generally versed in these matters, so I am cc:ing
> this to Jo Walsh.  Jo has now been elected as a board member of the Open
> Source Geospatial Foundation (OSGeo), and also appointed by the board to
> chair the Public Geospatial Data Project within OSGeo.

Thanks for nudging this onto me, Frank. I could ramble on on a
personal basis about some of the philosophy behind this, and the idea
of a participatory framework for spatial data infrastructures that
would not preclude commercial re-offerings or damage state agencies'
self-funding prospects, but that is probably best for another email,
and i have been sitting on this factual one for longer than i intended
to; i hope some of it may be helpful to you.


jo