Ever since OpenStreetMap was founded three years ago, the licence has been one of the most debated aspects of the project.

At present, every OSM contributor agrees that their contributions can be used under the Creative Commons Attribution/ShareAlike licence, version 2 (CC-BY-SA 2.0, for short). This means:

• Anyone can copy OSM data.
• But if you incorporate it into something else, that “something else” also has to be copiable under the same terms and conditions (ShareAlike).
• When you copy it, you have to give credit to the copyright owner (Attribution).

It seems uncontroversial on first glance. But, as ever with things legal, the devil is in the detail. Do you have to give credit to ten thousand different mappers named individually, or just to “OpenStreetMap”? Does “incorporating it into something else” include adding extra layers of info over the top, as in a map mashup?

These, and similar questions, have kept the subscribers to our legal-talk mailing list happily arguing for over a year now. Many of the problems arise because Creative Commons (the “CC” of our licence), as the name suggests, is largely concerned with “creative works” – music, literature, art, and so on. OpenStreetMap, on the other hand, produces data: a factual, uncontroversial recording of the world around us.

When OSM started, we were on our own. No other website had a significant corpus of user-generated, publicly-licensed data. Few had open data of any sort. So it’s not surprising that no obvious licence was available.

Today, the situation is a little brighter. Open data is recognised as a fast-growing movement – and fortunately, this means that those with talents in drafting licences are applying their knowledge to the same problems we’ve run up against.

The situation today

As the elected OpenStreetMap Foundation, we’ve been discussing this in our board meetings for the past few months.

We are aware there are serious risks in continuing as we are. Partly this is out of a desire to do better: if a substantial number of people are unhappy with the licence, we want to see if we can find a solution that also satisfies those who are happy with the status quo.

Most importantly, though, there is a strong body of legal opinion that our existing licence is not valid (for our purposes) in most of the world. Creative Commons bases its licences on copyright. In Europe, however, geographical data is principally protected by database right, and in the States, the only available protection is contractual. OSM data is potentially in a curious unlicensed limbo at the moment, which will not protect us if a major geodata company, for example, decides to take our data without respecting the intent of the licence. (See here for a EU angle, here for a US angle; for Creative Commons’ position, see the third paragraph here, and here at §5.2.)

As the Foundation, we do not control the project, but we look out for its best interests. So we would be negligent if we did not look for alternatives.

Finding a way forward

So here are the presumptions that we have brought to our work:

1. We need to consider alternatives to CC-BY-SA 2.0 for future licensing of OSM data.
2. At heart, to gain widespread community acceptance, the licence needs to give our database the same three basic licensing elements (freely copiable; share-alike; attribution required) as it has at present.
3. It must acknowledge the reality that we are one database with many contributors.

Of course, it goes without saying that any recommendation we make must have the best interests of the project and the community at heart – however hard that might be for any of us who have our own strongly-held views on the matter.

The Creative Commons approach

Through its Science Commons initiative, Creative Commons has recently published a “protocol” on open access data. Its primary focus is scientific data.

Its approach is unusual. It mandates that all such data be placed in the public domain, with no share-alike or attribution clause. Instead, these requirements should be expressed as a “non-legally binding set of citation norms”. In other words – the data provider can ask that those using the data share-alike and give attribution, but cannot insist on it.

Why is this the case? In the protocol, Creative Commons restates the problem that copyright simply does not apply universally to factual information. “Many users choose to apply common-use licenses such as the GPL and CC in order to declare their intent […] But a user would be able to extract the entire contents (to the extent those contents are uncopyrightable factual content) and republish those contents without observing the copyleft or share-alike terms. The data provider, based on our research, is likely to feel ‘tricked’ by this outcome. That is not a desired result.”

This is an accurate summary of the situation. However, their solution – putting all the data into the public domain – is unpalatable to some OSM contributors. Here’s why.

In the science world, for which this protocol is intended, citation is an everyday part of life. Learned articles will always cite those whose work they are building on, and will be disregarded by peers if they don’t. When Science Commons speak of a “non-legally binding set of citation norms”, they can do so in the expectation that these norms will be respected. SC’s John Willbanks says as much in this interesting post: “It seemed we had to think about taking all these social goals and moving them outside the legal world, and into the world that scientists controlled – norms.”

That isn’t necessarily the case with geographical data. If a big mapping company like TeleAtlas or Navteq were to use OSM contributors’ work in their data, there’s no commercial imperative for them to credit us – let alone to make the rest of their data available on our terms. Our work will be used outside the world that we control.

Other OSM contributors say “so what?”, reasoning that the low cost, high quality and regular updates of free data will always win out over the commercial product. But even those with such views (among whom I personally count myself) do acknowledge that they’re not held universally within the community, making a licence on those terms unlikely to be adopted by all.

The alternative

So if Creative Commons’ existing licence may not be valid, and their new licence is unlikely to be palatable to many OSM contributors, where do we go from here?

The most promising answer, so far, is provided by the Open Data Commons project. The ODC is principally the work of two lawyers, Jordan Hatcher and Charlotte Waelde, specialists in intellectual property law as applied to electronic content and with significant geographical expertise. They have worked with Creative Commons (indeed, the open access data protocol is largely their work), and have been generously sponsored by scientific data firm Talis.

As well as the public domain dedication that they have provided for Creative Commons, they also drafted a pair of licences (the Open Data Commons Database Licence and the Open Data Commons Factual Info Licence) which together provide a set of provisions particularly suitable for OSM data. Key points include:

• The database is protected by share-alike (ODCD 4.4).
• Attribution to the project is required (ODCD 4.2b).
• The licence works through copyright, database right and contract, making it applicable across different jurisdictions (ODCD preamble).
• The extent of share-alike is expressly stated (ODCD 4.5).

In other words, the licence retains the fundamentals of OSM licensing – freely copiable, share-alike required, and attribution required; is unambiguously valid for data; and provides definite answers to those questions of interpretation (such as who to attribute and how far sharealike extends).

Where now?

Work on the ODC Database Licence and Factual Info Licence has unavoidably been slowed down by the Creative Commons public domain project. But both Jordan Hatcher and the project’s sponsors Talis have assured us that work is going ahead, and that a new revision will be available in the near future. As Jordan writes on his blog, “the Open Data Commons Database Licence and the Factual Information Licence are on hold during development of the new legal tools (the PDDL and Community Norms) but not forgotten.”

Paul Miller at Talis has already kindly given us his time to discuss the licence and we hope to meet with Jordan as the project moves on.

If the Foundation then believes that the licence is settled enough to be considered a viable option for OSM, we will recommend to the contributors – you – that we adopt this as the new licence going forward, for both existing and new data. We can’t change anything without you. You, not the Foundation, own the rights to your mapping.

The ideal situation would be that the ODC licences are suitable, and that not just the Foundation, but everyone involved in OSM agrees. If that’s the case, then we can publish the new terms without significant loss of existing data. If there are major objections, of course, we will either withdraw the proposal entirely, or weigh up the dangers of retaining the existing licence against the potential withdrawal of data. We are hopeful that the ODC licences are sufficiently in tune with the “spirit of OpenStreetMap” that this will not arise, but need to be alert to the possibility.

OSM isn’t just “open” in name – we believe in openly discussing the matters affecting the future direction of the project. If you have concerns or, conversely, would like to express your support, then please do discuss it on the mailing lists or forum; or, as at any time, contact any of the Foundation board (Corey Burger, Etienne Cherdlu, Steve Coast, Michael Collinson, Richard Fairhurst, Mikel Maron, Andy Robinson), or e-mail us collectively at team@osmfoundation.org.

Fixing “the licence problem” might be something that many have thought impossible, but we are hopeful that if we really want to resolve it, we can find a solution that will benefit OpenStreetMap and stand us in good stead for the years to come.

— Richard Fairhurst