[Talk-GB] OS OpenData and ODbL OK

Robert Whittaker (OSM) robert.whittaker+osm at gmail.com
Tue Jul 5 13:51:54 BST 2011

On 5 July 2011 13:14, Richard Fairhurst <richard at systemed.net> wrote:
> I think that's a red herring, isn't it? ODbL imposes additional requirements
> over and above DbCL. OSM is not distributing OS OpenData under DbCL alone,
> nor does it permit anyone else to do so (subject to the usual 'Substantial'
> test, which is of course Database Directive stuff and therefore governs OS's
> existing data distribution business anyway).
> What circumstances can you envisage in which OSM-distributed OS OpenData
> might not be subject to the provisions of ODbL?

My understanding is as follows:

ODbL only covers the database and not it's individual contents. (The
example given by OpenDataCommons is a database of photographs, where
the database could be covered by ODbL and the photographs themselves
by a suitable contents license -- which could be DbCL or CC-By or
something else.)

Without a separate license for the contents you're actually unable to
distribute any of the contents of the database. (See the preamble in
the legal text for ODbL, along with clauses 2.2a and 2.4.) OSM wants
to distribute things under ODbL for the database and DbCL for the
contents. This means that any submitted data needs to be compatible
with both of those licences (or at least the two in combination).

DbCL is essentially a copyright waiver on the individual contents, so
downstream users only have to worry about complying with ODbL. You're
right that it doesn't impose any additional requirements on downstream
users. But it does grant them additional freedoms. So you therefore
need additional rights to do so from any upstream suppliers.

In the context of OSM, the fact that the contents will be under DbCL
will enable users to make use "insubstantial extracts" without having
to provide any attribution or share-alike or anything else. (Clause
6.2 of ODbL says that ODbL does not impose any restrictions on such
extracts, so you're just left with DbCL to comply with.) Without the
contents under DbCL -- for example if they were under CC-By (which is
fairly close to the OS OpenData Terms), then you'd have to provide
attribution even on insubstantial extracts. (Unless of course you
tried to use fair use / fair dealing arguments, or claim geodata facts
aren't copyrightable -- but I don't think we'd want to rely on these
in all jurisdictions.)

Because of this difference, I believe we would need permission from OS
to distribute derivatives of their data under the specific ODbL+DbCL
combination rather than just the ODbL part. It may be that we already
have this and Mike just didn't mention it, or it may be that
"ODbL+DbCL" is implicit in OS's agreeing to "ODbL", but I think it's
something that should be checked carefully.


Robert Whittaker

More information about the Talk-GB mailing list