Systeme D

13 June 2006

Towards a free geodata licence

I posted previously about practical problems with using a ShareAlike licence for geodata.

Most of the discussion in the comments focused on matters of scope - in other words, to what extent do the "viral" parts of a ShareAlike licence mean you have to share the non-geodata parts of your work? I think the consensus was that I was being overly pessimistic, and I can see that.

But the boundaries are still very ill-defined for geodata. To use Creative Commons ShareAlike phraseology, when does a "collective work" become a "derived work"? Here's an illustration I prepared to demonstrate the difficulties.

Existing licences

The reason we have these ambiguities is because Creative Commons licences weren't written with geodata in mind. To look at the three most common types of copyleft/ShareAlike licences:

  • Creative Commons licences are expressly intended for "creative works".
  • The GNU GPL is expressly intended for software.
  • The GNU LGPL is also expressly intended for software, usually libraries.

Which one of these is closest to geodata? Well, geodata is clearly not software. But then... it is source code, of sorts. End users don't consume raw geodata, but they do use the 'compiled' products of it: maps, route-planning applications, or database-driven mashups of the sort MySociety are so good at.

And is geodata a creative work? I don't think so. At its best, it is simply a faithful, uncreative recording of what's on the ground. Creative effort is involved in drawing up the schemas for categorising the data, and in making the artistic judgements that result in a beautiful map, but not in collecting and recording the geodata.

(There is therefore an argument, especially if you follow Bridgeman vs Corel, that geodata isn't entitled to copyright protection - other than under the shorter-lasting EU concept of database right. To quote from the judgement, "Slavish copying, although doubtless requiring technical skill and effort, does not qualify. As the Supreme Court indicated in Feist, 'sweat of the brow' alone is not the 'creative spark' which is the sine qua non of originality." Which is pretty harsh on those, like me, who have sweatily trudged the streets with GPS units recording for free geodata projects. That's kind of a side issue and I won't follow it any further, but it does illustrate the danger in treating geodata as a "creative work".)

Choosing the ideal licence

To recap a little, we're looking at share-alike licences because many of the people involved in free geodata projects like to use this sort of licence. Openstreetmap is a case in point: of those who care about such matters, the vast majority appear to be GPL advocates from a free-software background. Other contributors largely don't care, with the lonely exception of one PD/BSD advocate (me :) ).

So we're using a GPL/ShareAlike licence. This is good, this is how democracy works. What we now need to do is find the best one.

It is my humble(-ish) submission that Creative Commons ShareAlike is not the best possible viral/copyleft licence for geodata. This is largely because of the ambiguity expressed above. Ambiguities are not good in licences, unless you're a lawyer, and free geo projects generally don't have "enriching lawyers" among their aims. (Seriously, this is really important, because where you have legal ambiguities, you have lawyers; and with lawyers, generally the best-paid wins. National mapping agencies like the OS, and megageocorps like Navteq and Tele-Atlas, can afford good lawyers. Free geodata projects can't.)

The best possible viral/copyleft geodata licence would preserve the "freedoms" that GNU enthusiasts value; would maximise the amount of free geodata in the world through the ShareAlike provision; and would do all of this unambiguously, for the avoidance of lawyers.

Creative Commons took their cue from the GPL. Unfortunately, from a geodata point of view, they borrowed the wrong bits.

Finding the best bit

There's one crucial clause in the GPL which is missing from the Creative Commons licences.

"3. You may copy and distribute the Program (or a work based on it)... provided that you also... Accompany it with the complete corresponding machine-readable source code; (and) Accompany it with a written offer... to give any third party a complete machine-readable copy of the corresponding source code"

The reason this isn't in the Creative Commons licences is that creative works don't have source code. But as discussed above, geodata is closer to source code than to a creative work.

A successful ShareAlike geodata licence could be based around this "source code" provision. You would be required to make the raw geodata ("source code") available for any derived work that you produce using this free geodata.

If you wanted to produce a map or route-planner from free geodata, you could do so - but only on the condition that any other geodata you included in the work also became ShareAlike-licensable (or "free" in GNU parlance). There is no longer a need to define what's a "derived work" and what's a "collective work". You simply impose this one source code provision.

To summarise such a licence in two bullet points:

  • You may copy and distribute this geodata, or a work based on it, provided that you accompany it with the complete geodata source for the full work - or a written offer to give any third party this complete source.
  • Apart from this, you may do what the hell you like with it.

Any attempt to make a real usable licence out of this would require much more thinking about specifics, and in particular, licence compatibility (so that maps produced from such geodata could be included in Wikipedia and the like). But I think this has real promise.

(I owe a hat-tip to Schuyler, whose insightful posting on the Openstreetmap lists set me off along this train of thought.)


Comments

Posted by on 29.11.06 20:50


Add a comment

Your name:

E-mail address: 

Comment:

your comment. (E-mail addresses will not be visible, but a server-based mail link will be provided. To guard against spam, no comments which include 'http' or 'www' will be published.)