Mapping Hacks

Recently there has been a new level of licensing discussion for open geodata. Daniel Faivre has been socialising the Public Geodata License, a draft of which he worked on years ago and which is being picked up by a group in Canada, in particular. The PGL is very strongly modelled on GPL v2.

Richard Fairhurst wrote a really interesting analysis of problems with a ShareAlike clause covering open geodata, for people who want to create and publish “derived works” from open geographic information. He makes some good points covering “what is a derived work”, and what restrictions on use based on use might mean in terms of people sticking with proprietary alternatives for their mapping needs.

The Open Geodata Manifesto talks about the benefits of a ShareAlike clause in encouraging commercial use while getting better contributions to spatial information resources collected and distributed by states. In a discussion about open geodata on Ed Parson’s blog i made a comment about this, i thought i would repost it here because i don’t like comments that much, it seems pointless to have two ‘levels’ of contribution to a discussion.

In the open source worldview there is a concept of a ShareAlike license. Data can be made fully and openly available for all uses including commercial uses, with the constraint that if improvements and enhancements are made to the data, those changes are made available in the public domain, covered by the same license. The Public Geodata License (http://cemml.carleton.ca:8080/OGUG/pgl/ provides a good model for this.

This isn’t the only model for making geographic data available in a more open way, but it’s a good one for state-collected information and public sector information in general. A ShareAlike license applied to distribution of Ordnance Survey data would help fulfil the purpose for which it is suppose to exist, to collect and redistribute the most current and accurate description of the UK that is available.

Richard offers a consideration that applications using web maps may not be able to use ShareAlike maps because information they don’t have a right to redistribute becomes part of a “derived work” with it; presentation as well as re-annotations of geodata, and collections of related information that being used in a “mashup” sort of a way, could become covered by the terms of a demanding license.

Richard offers positive suggestions to resolve these issues:

The simplest way to avoid all these problems is to use a BSD-style licence, or an unambiguous declaration of ‘public domain’. These do not restrict the works that can use the data.

This may have two practical disadvantages. One is that some contributors are implacably opposed to BSD-style licences; another is that it will lead to the development of two distinct free geodata streams, one ShareAlike and one PD/BSD.

The long-term solution, then, must to be to create a dedicated LGPL-like geodata licence that addresses the problems above, setting out what you can and can’t do in terms that are unambiguously relevant to geodata and mapping.

The problem of two distinct streams of open geodata is a deep one. I don’t think there is one licensing stance which would be optimal for everybody; some people are more determined to create very free works, even if that takes a bit longer to build because it impedes some potential contributions.

Right now, peoples’ motives in collecting and distributing public information and the motives of enthusiasts look quite different. Cost of maintenance of updates is the biggest burden that state-funded information agencies claim to be bearing. A ShareAlike clause in that context would guarantee that enhancements to public domain geodata should be returned to the public domain. I can understand that this doesn’t work for everyone. When i worked on music metadata systems with state51, Paul offered a convincing argument that GPL, Gnu Documentation License type licenses can be unsupportable for small organisations, a lot of whose value is contained in the enhancements they can bring to an outline of publically available information. There is a surprising amount of value in, for example, the catalogue metadata owned by a niche independent record label.

The “Collective Work” stance that Richard is suggesting is, on these grounds, pretty interesting. A collective trust maintenance of a body of data would work a bit like a collecting society, except it would be a redistribution society.

All this helps throw light on a question that is interesting to me: at what point does data become metadata, and to what extent does “metadata” become covered by the terms of a license that covers “data”? Each shape that is contained in the Ordnance Survey’s data set has a TOID, an alphanumeric code that identifies it uniquely, and which is then keyed to a polygon described as a set of latitude and longitude coordinates. A TOID, arguably is not part of a description of a thing, it is a reference to a description of a thing: a unique identifier is a piece of metadata about a spatial object. What about the coordinates themselves? They are a description of a thing in geometry but also a reference to a thing in space; this information could also be interpreted as metadata. As Richard’s potential future scenarios outline, PDFs or web pages containing information about what is happening nearby, could by being combined with vector descriptions of things in space, become metadata about something that is described by the map and thus part of a work derived from it that ought to be covered by the same terms. The PGL covers metadata cases explicitly. Perhaps it’s hard to make a very firm description of where the line between data and metadata is, and some people argue that it doesn’t really exist.

One reason for getting involved in the Public Geodata project at the burgeoning Open Source Geospatial Foundation (OSGeo) is to help set out decision aids for licensing of data, with licensing options that can be included and explained in software packages like Geoserver, Mapserver, that enable easy publishing of the vector data underlying web mapping services using the Web Feature Service standard protocol. The transaction mode of WFS-T really has good potential to allow more direct and delegated contributions to public domain geodata. Simple standards exist which can allow much easier and more copious sharing of data between different public agencies and well as between citizens themselves. Simple licensing standards which guarantee the most public value in the data could really help with this. I like the LPGL idea a lot.

Posted in geodata, licensing |

You can follow any responses to this entry through the RSS 2.0 feed. Trackback from your own site.