[OSM-legal-talk] Would The ODbL and BY-SA Clash In A Database Extracted From a BY-SA Produced Work?

Rob Myers rob at robmyers.org
Sat Sep 4 22:13:53 BST 2010


On 09/04/2010 03:38 PM, Anthony wrote:
> On Sat, Sep 4, 2010 at 8:15 AM, Rob Myers<rob at robmyers.org>  wrote:
>> Would removing the word "individual" from the CT improve it?
>
> Sure, it'd make everything (except the database schema) DbCL, and DbCL
> is better than ODbL.

It would make contributions DbCL. The database aggregation of them would 
then be ODbL.

>> OSM ways aren't generally representations of artistic works, though.
>
> They're not artistic, but then neither are most other maps, or an
> encyclopedia article, or a newspaper article, or a police sketch, or a
> textbook.

Agreed, but in the example you give any copyright will be a product of 
the way representing an original artistic work.

> If you create a way representing the outline of a lake freehand from
> memory, it's hard to see how that's any less copyrightable than any
> other crude sketch of reality.  Now most lake outlines are not done
> freehand from memory, but some might be, and I wouldn't feel too
> comfortable copying a way which might be copyrighted.

That's what the DbCL is for, to ensure that all Ways are covered by 
precisely the same level of rights.

>> And even if any are copyrightable the DbCL will handle that.
>
> That's what I'm wondering.  It's not really clear.  If the
> "individual" in "individual contents" means a database row, and a way
> is copyrightable, then the DbCL probably doesn't handle that, because
> a way is a creative arrangement made up of many database rows.

If the CTs specified the levels of organization of the DB, e.g nodes, 
ways, meatadata, etc. would that be better? Or would saying "any 
contribution" suffice to cover any level of organization or abstraction?

I think the latter should do it, but I understand that something more 
precise might be more convincing.

> And having a single database row as the "individual" in "individual
> contents" fits best with the spirit of ODbL (which, oddly enough,
> seems like it was never adapted for something like OSM even though it
> was (*)).  In a database of white page addresses, or a database of
> flickr images, or a database of baseball stats, or NCBI’s Entrez Gene
> database, the definition of individual is obvious: a single white page
> entry, a single flickr image, a single baseball stat, a single gene
> name or pathway or protein product, which is probably going to
> correspond to a single row.

I think that you are right that this is an important distinction, but I 
also think that the levels of structure in OSM are easily identified and 
descibed (nodes, ways, tags, etc.). The example you give of a row 
doesn't preclude an entity that consists of multiple rows being covered, 
indeed it makes it more covered as it contains many covered entities.

> It would be interesting trying to convert OSM into an unordered
> collection of facts and then converting it back into a structured
> database, in a clean room type operation.  You'd probably wind up with

Teleportation. ;-)

> a public domain database in any jurisdiction without database rights
> or sweat of the brow, and possibly also in any jurisdiction with
> database rights but without sweat of the brow, if you can argue that
> no single person or corporation was first to make the "substantial
> investment in either the obtaining, verification or presentation of
> the contents" required to obtain database rights.  Notwithstanding the
> ODbL, of course, since writing down what you want to be the
> restrictions on a database does not amount to creating a law allowing
> you to put such restrictions on that database.  (You can only bind by
> contract people who want to be bound by contract.)
>
> (*) I guess because it was adapted under the premise that OSM was an
> uncopyrightable collection of facts.

I was recently reminded of this blog post that explains more generally 
why the DbCL (or FIL as it was called at the time) is needed:

http://www.opencontentlawyer.com/2007/09/open-data-commons-contracts-and-factual-information/ 


>>>   If such a database were released with "DbCL 1.0 for the individual
>>> contents of the database", would the way be DbCL, or would it be ODbL?
>>
>> The way would be DbCL, and use of it would be covered by the ODbL or not
>> like the rest of the database.
>
> That's not the way licenses work, though.  If something is covered by
> two licenses, you can apply *either* license, you don't have to apply
> both.

Your examples of various kinds of potentially copyrightable Ways show 
why the DbCL is a pre-requisite for OSM applying the ODbL. And the ODbL 
applies to the combined DbCL work collected by OSM. So both licences 
certainly apply.

> I guess that's one part I don't get.  Because if you had to apply
> *both* licenses, it makes sense.  But if you can apply *either*
> license, it doesn't.

You have to apply both. Or, rather, the DbCL has to be applied so that 
OSM can then apply the ODbL to the results. I think. In any case, the 
DbCL is what allows BY-SA Produced Works to be created.

> That said, I guess the answer is that the "individual" in "individual
> contents" is *meant* to mean any non-substantial extract (i.e. any
> extract on which the ODbL does not apply).
>
> Of course we then still have the question as to what is "substantial",
> but at least that's a question for which there is likely to be some
> case law (as opposed to "individual contents").

Yes I'd love to know what "substantial" means. But if we take the 
example of Fair Use, it can actually be harmful to try and set 
quantitative rather than qualitative guidelines. The UK's useless Fair 
Dealing system resulted from trying to quantify Fair Use.

>> It's a licencing stack, with each level (data, database, map tiles) covered
>> by an appropriate licence.
>
> It's certainly strange that the middle of the stack would be the most
> restricted level, isn't it?

Licencing funnel? Licencing hour glass? ;-)

The middle of the stack is the level that protects free availability of 
data(bases), so from that point of view it's not surprising that it's 
the most protective.

>>>   Would it make a difference if the way were split into 1,000 different
>>> connected ways?
>>
>> I wouldn't have thought so. If you can re-assemble the copyrighted work then
>> you have the copyrighted work (teleportation doesn't strip copyright,
>> remember :-) ).
>
> Yes, but my question was over the application of the DbCL, not the
> application of copyright law.  1 way, perhaps, is "individual
> content".  1,000 different connected ways isn't.  If it were, then you
> could make substantial extracts and have them covered by DbCL.

If we can tell that they are connected, and if they amount to a single 
entity (at some level of abstraction) why should that entity not be 
considered "individual"?

> Actually, it's not really clear what the DbCL accomplishes.  The
> point, as was explained to me in another thread, is to accomplish
> nothing - saying that the individual contents are DbCL is like saying
> that the individual words in a Wikipedia article are DbCL.  Trivial,
> obvious, and unnecessary.

Sure, but that doesn't mean either that the DbCL isn't needed or that 
the CTs don't need to make it clear that contributions will have the 
DbCL applied to them. It's at worst unclear wording (which I agree 
should be addressed).

> Maybe that's all it does.  Or maybe there's more to it.  I don't know.

The DbCL is designed to work like this:

http://www.kugelbahn.ch/sesam_e.htm

or this:

http://makeprojects.com/Project/The-Most-Useless-Machine/91/1

;-)

If part of the contents of the database would otherwise be encumbered by 
copyright, DB right, or whatever, the DbCL neutralises that. If it 
wouldn't, it doesn't. When you don't need it, you don't need it. But 
when you need it, you *really* need it.

The DbCL creates a level playing field or baseline for the ODbL to build 
on. This is important both so that using the Database doesn't fall foul 
of any hidden variations in rights and so that Produced Works can be 
covered by an arbitrary licence, including BY-SA.

(IANAL, TINLA.)

- Rob.



More information about the legal-talk mailing list