[OSM-legal-talk] ODbL and publishing source data

Frederik Ramm frederik at remote.org
Mon Nov 28 21:33:44 GMT 2011


On 11/28/2011 12:55 PM, 80n wrote:
>     I could render a map from OSM and then render something else on top
>     of it, say a commercially acquired set of hotel POIs. That would
>     clearly be a Produced Work; I could point anyone asking for the
>     source data to the planet file and the rendering rule, and keep the
>     hotel POIs to myself.
> This is an overlay on top of a Produced Work.  Whether it's produced by
> layers at the browser end or by compositing two separate images doesn't
> seem to be materially different.

I agree that it should not be different, but technically "layers at the 
browser end" is combining two images after disseminating them, and 
"compositing two separate images" happens before disseminating them, so 
any license for which the act of publication triggers something is 
likely to distinguish between the two cases. (It would be interesting to 
find out if there are lawyers who say that publication happens at the 
time when the Javascript in the browser combines several data streams 
into something viewable, rather than when the data leaves our server. If 
that were the case, then several of today's typical uses like overlaying 
transparent non-CC-BY-SA pictures over OSM tiles would cease to work.)

>     I could also remove all hotels from my OSM copy and add in the
>     commercial hotels instead, then render a map from it. Unless the
>     commercial dataset is missing data, the resulting map could look
>     100% identical to the map from the first process, but this time I
>     would be required to release the hotel dataset because it is part of
>     the derived database used to create the produced work.
> Leaving aside the step about removing content for the moment, I don't
> see how this is materially different from the first example.  You've
> simply overlaid your hotels on top of the OSM data.  I don't think the
> mechanics of how you achieved this are, from a legal perspective,
> important.

I agree that they should not be important but I am pretty sure that they 
are. - Of course, if there are several plausible ways to arrive at the 
same end product then you can always *claim* to have used process A 
rather than B and nobody will be able to prove the opposite.

> Any process can be considered as a series of inputs to a
> black box and some outputs.  If the inputs are the same (an OSM database
> and a set of POIs) and the output is the same (a map with an overlay of
> the POIs) then it shouldn't matter whether it was achieved using a
> complex machine or monkeys with typewriters.

We don't currently require of users of ODbL-licensed data that they 
demonstrate what processing chain they used to arrive at something 
(though they might do that of their own volition in order not to have to 
release a database). I can imagine, however, certain produced works 
where it is really not plausible if people claim to have created it 
using "monkeys with typewriters" - assume someone makes a map where all 
the roads are coloured according to the number of accidents that happen 
there. If someone were to publish such a map, and upon being asked for 
the source data, were to claim that he was not using a derived database, 
just a regular OSM database where he adds some things ad render time, 
and therefore had nothing to release - I would be skeptical. But I guess 
it is *impossible* to make a map that way (just unnecessarily complex).

I'm not sure what I can do in such a case. I mean even today someone 
could use OSM outright and deny it, in that case my only recourse is in 
the courts; I guess the situation above is similar - if someone claims 
to be using a process that requires them not to release their extra data 
but I believe he has an OSM-derived database where he merged in other 
data, then I might have to sue in order to find out.

>     Same thing with your reply to my "pencil" example - depending on how
>     exactly you update your produced work, you might or might not have
>     to release a database.
> If this were to be possible then it would be a very undesirable flaw.
> The intent of ODbL was to protect OSMs database and ensure share-alike.
> If it can be circumvented then it fails one of its main purposes.

Oh, it does protect OSM's database all right, but drawing lines onto a 
printed-out image is not making a derived database (and frankly I 
wouldn't be all that interested in the geometry of those).


Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"

More information about the legal-talk mailing list