[Talk-GB] Ordnance Survey
richard at systemeD.net
Tue Apr 6 22:02:17 BST 2010
Tim François wrote:
> Also, be aware that a discussion is also ongoing in the discussion
> page: http://wiki.openstreetmap.org/wiki/Talk:Ordnance_Survey_Opendata#Attribution
> (you may be aware, just making sure!)
...and we had quite a lengthy discussion about it on IRC earlier, too.
The following is my understanding of how it can work.
The new Government licence terms (including OS OpenData) are
specifically designed to be compatible with the CC-BY family of
licences. The one cited is CC-BY 3.0, but there's nothing that
conflicts with CC-BY-SA 2.0, which is of course what we use. The
Government, and OS, have gone out of their way to make sure that we
can comply with the OS licence by fulfilling CC licence terms. So
let's look at what those terms say.
CC-BY-SA 2.0 says:
> If you distribute, publicly display, publicly perform, or publicly
> digitally perform the Work or any Derivative Works or Collective
> Works, You must keep intact all copyright notices for the Work and
> give the Original Author credit reasonable to the medium or means
> You are utilizing by conveying the name (or pseudonym if applicable)
> of the Original Author if supplied
> in the case of a Derivative Work or Collective Work, at a minimum
> such credit will appear where any other comparable authorship credit
> appears and in a manner at least as prominent as such other
> comparable authorship credit
This is what we're talking about here.
Crucially, it is no different from the problem we have with any
contributors. You or I could demand that, as rights-holders in the
data which we post to OSM, we should be individually "given credit"
every time the data is used. Indeed, the letter of the licence appears
to say that.
We do also know that CC licences were certainly not designed for data,
and not expressly for mass collaborative projects either. Spatial law
as a discipline is in its infancy; so is open-source law. Open-source
collaborative spatial law? I suspect there's only one lawyer in the
entire world who even begins to understand that. Whatever. The
point is that a licence cannot be read as an isolated text: it exists
within a rich legal system.
The crucial line in the clause above is "reasonable to the medium or
means You are utilizing".
Now anyone with the mis^Wgreat fortune to read legal-talk will
recognise that, within our rich legal system, this is the whole
"substantial" caboodle twice over. Firstly, like "substantial",
"reasonable" is a "how much?" argument, not a binary one. Secondly,
the "substantial" question itself is relevant: if it ain't
substantial, it ain't protected, therefore you don't need to attribute.
If you print an OSM map of Charlbury, CC-BY-SA might say you need to
attribute me, because I did lots of it. If you print an OpenCycleMap
small-scale overview of the UK National Cycle Network, it might tell
you to attribute me and Gregory W, at least. But, and this is a big
but, it only needs to be "reasonable", at the same prominence as "any
other comparable authorship credit". If you're publishing a glossy
large-format printed atlas that names the individual surveyors you've
employed, you probably need to name me and Gregory too. If you're
distributing a little iPhone app, you can simply link to OSM, where
the contributors should be listed.
This is a rather long and winding e-mail, but I hope you can now see
where I'm going.
The important thing is for us to get our own house in order about
attribution. We've often been able to get away without it - mostly
because OSMers by tradition waive their right to attribution,
though also because many map views do not involve the substantial work
of one contributor. Large-scale data derivation challenges both of
The first step of getting our house in order is to correctly attribute
those contributors who require attribution. Earlier today I committed
a new, clearer, and more compliant copyright notice to the core (Rails
port) site. It presumably needs to wend its way through OSMF
committees for approval or otherwise, but you can see it on the dev
(linked clearly from the LH sidebar at all times)
This, I believe, would put OSM in compliance with the OS and other
The second step is to help end-users comply with attribution
requirements. I believe that, in the vast majority of cases, a link to
OSM (where the attribution will be clearly displayed) will be
sufficient. Not only will it usually be "reasonable to the means",
most imported data is rudimentarily factual and therefore does not
pass the "substantial" test .
However, it is possible to imagine a scenario where explicit
attribution would be required - especially if we were to relax our
evolving disdain for bulk imports (a disdain which I share). If, for
example, we were to nuke all our existing river data (which we
shouldn't) and import StreetView-derived rivers, a rivers-only extract
from OSM may require OS attribution.
The copyright statement (linked to above) puts the onus on the end-
user to comply: "it may be reasonable to credit them". This is
perfectly acceptable, but we could aim to do more. At its simplest
this would be a set of community guidelines on "how to attribute"; if
there was the will for more, it could include some automated analysis
of source= tags.
That's the theory and I hope it's some help. I don't see any reason
not to trace from OS data, but I think we do need to do a little work
in improving our own standards of attribution, and supplying the above
notice is my attempt to do something about it.
Can I close this far-too-long epistle with two pieces of realpolitik?
The first is that the Government _wants_ this data in OSM. The idea of
releasing open data is that it's used openly by geo hackers, and we
are the biggest set of geo hackers in the country. We should comply
with the licence, of course; but we shouldn't think that there are
hawk-like Government lawyers waiting to trip us up on a tiny
The second is that we already have a problem. New Zealand and Canada,
I believe, have both had imports from massive attribution-required
datasets (LINZ and GeoBase/GeoGratis/Canvec). There are probably
others, and there will be more.
 FWIW I believe it is also compatible with ODbL
 disregarding for now the old "is it copyrightable at all?" argument
 and you know who that is :)
 though until we move to ODbL, we will have problems if anyone
decides they do want to be attributed
 I also believe it provides much more clarity to potential data
users about our licensing, and indeed it's something I've been wanting
to do for ages, aside from attribution issues
 ...and bear in mind that OS's attribution requirement is drafted
to cover their cartographic copyright, which we are not touching,
regardless of any rights in the data
More information about the Talk-GB