Kevin Pomfret from the Centre of Spatial Law just published The ODbL and OpenStreetMap: Analysis and Use Cases a white paper reviewing pain points in the ODbL - OpenStreetMap's current license.
2.5 billion OpenStreetMap nodes!
The paper provides a comprehensive review of issues broached in talks at State of the Map US (More Open, OpenStreetMap Data in Production) and State of the Map EU (The State of the License) and discussions thereafter. It offers an assessment of legal risks and includes a series of case studies focusing on legitimate use cases of OpenStreetMap that are currently impeded or complicated by the ODbL. At both State of the Map conferences I have heard requests from the Licensing Working Group, the OpenStreetMap Foundation board and others for a more solid summary of problems and actual real world use cases that are impeded by the license. This is why over here at Mapbox we have supported the Centre of Spatial Law to compile this white paper.
Here's an overview of issues identified by the paper in order of appearance in the ODbL license:
- License does not cover contents - the ODbL covers the database, but not its contents. OpenStreetMap does not make clear under what conditions the actual contents of the OpenStreetMap database are available.
- Rights of contributors is uncertain - neither the ODbL nor the Contributor Terms protect a licensee from third party intellectual property claims. Note that a third party here is not limited to contributors, but would also include parties whose data has been imported to OpenStreetMap.
- Uncertain if and to what extent "share-alike" applies - the delineation between Produced Work and Derived Database is fuzzy and the crucial concept of Substantial is entirely undefined. This makes the extent to which share-alike applies to data that is combined with OpenStreetMap data guesswork.
- Uncertainty as to which jurisdiction's law applies - the ODbL states it will be governed by the laws of the relevant jurisdiction in which the License terms are sought to be enforced. - the global nature of OpenStreetMap together with (1) makes it unpredictable as to in which jurisdictions to expect claims.
- Lack of a cure period for a breach - there's no grace period to make amends. If you're in breach of the license you have to stop using it right away.
- Unclear governance - there is no authority to ask for definitive clarifications around the license. When posing questions on related mailing lists or the OpenStreetMap Foundation the standing practice is to defer to license interpretation and non-existing case law.
The paper's case studies illustrate how potential OpenStreetMap users don't use OpenStreetMap at all - or not to the extent they could - due to the problems outlined above. This is a crucial issue - we're not a community of givers on the one side and takers on the other, there's a large overlap between data users and data contributors and the more we can get OpenStreetMap used in the real world, the more exposure we have to potential contributors, the more contributors we'll have.
Here are some of the case studies:
Yale University does not use OpenStreetMap in research under HIPAA or similar privacy regimens because of concerns that ODbL's share alike provisions could force researchers to open sensitive data - for instance when geocoding research data with OpenStreetMap data. This example highlights the issues with share alike (3) but also with governance (6). Some of the concerns expressed by Yale may be based on a conservative reading of the ODbL, but the absence of license governance in OpenStreetMap (6) and the understandable desire to avoid any risk of violating federal law rule out OpenStreetMap as an option where it should be a prime candidate.
As the Wikimedia Foundation is exploring opportunities to integrate tighter with OpenStreetMap they are running into incompatibilities between Wikipedia's CC-BY-SA license, Wikidata's CC0 license and the OpenStreetMap's ODbL. It should be a no brainer that OpenStreetMap and Wikipedia should work as close as possible with each other for the benefit of both projects. Maybe a good real world use case we can get all moving on?
Foursquare is not using OpenStreetMap for reverse geocoding where they could due to concerns about share-alike extending to Foursquare data. Foursquare has been an awesome engine for driving people to become contributors and they show willingness to contribute data but can't commit if the extent of the commitment is not clear. This is a great example of where we're loosing out on contributions with a license that tries to take it all. To hear it directly from the source, listen to Dave Blackman's talk at State of the Map US.
The National Park Service is working on standing up their own OpenStreetMap like service where they could otherwise use OpenStreetMap directly to power Park Service maps. This is due to the fact that OpenStreetMap's share alike provisions are not compatible with the National Park Service's policy to keep their data in the public domain.
- Join me for a birds of a feather session on licensing at State of the Map in Buenos Aires.
- Geocoding is one of the issues that has come up most in the paper's case studies. Let's unlock permanent geocoding with OpenStreetMap and create clarity - join the discussion on legal-talk.
For a full read of the white paper, head over to the Centre for Spatial Law blog.