Accueil

Should open data standards use an open license?

Innovation & technologie

Earlier this month, Robin Berjon published a refreshing piece: W3c specifications should use an open license. Insightful and right in my current subject of interest.

As we are developping Open511, the question of the license becomes essential. Our context is slightly different than W3c but not that much: open data “standards” with high adoption, like GTFS and Open311, are de facto standards (see our discussion on this topic on the OpenNorth’s blog). These specifications have unclear license and governance issues when regarded as “standards”. We anticipate it will have a negative impact in the future.

In order to improve the governance side, we evaluated official standardization bodies (and W3c is a first rank candidate) but they come with restraining licenses. Not the kind of limitation that is expected in the world of open data. A recent discussion with Karl convinced me that we should target an open license.

On top of Robin’s arguments, I would say that an open license (with the implicit acceptance of fork) would dedramatize parallel efforts to develop new features. As we see for some open data specifications, organizations forking the specs tend to do it on their own since they feel rogue. With a right-to-fork philosophy, people experimenting will feel accepted to share and discuss their modifications. It would be the opportunity to evaluate with the community if the changes are indeed useful and meet some business cases. Given what I see on some mailing lists about open data standards, I have the feeling that such an approach could help a lot.

Open license probably makes a lot of sense in the context of open data. I like the approach of Robin saying something like “fork and fragmentation in the spec does not mean it will be the case for implementations.” But how can we be sure? Open511 is still a young project and like a child, we want the best for him. To continue with the child metaphor, alternative education, as interesting as it may seem, comes with some risks: kids will usually be more open to the world but in some case they will have some academical lag compared to other kids.

To come back on the standard question, do we have some experiences of specifications efficiently managed with an open license? Added to that, open license is not everything. How do we ensure that besides the open license (which will necessarily be outside of a standardization body as it is), we set up an environment and a community where both the specification and the use of the open license will provide the best outcome? Do we need safeguards to prevent fragmentation of the implementations? Would the implementors and users (mainly governments) be able to keep up with the pace of change? Those are tough questions when at the same time we are evaluating how to increase the number of implementations and adoptions. But the license and governance question is part of the “number of implementation” equation.

We are preparing various documents about the governance of Open511. When more stable, these documents will be shared with the community and we will try to choose the best model, including the license, for our case.

Stéphane Guidoin est un amateur de chocolat qui s'essaie à l'occasion à une analyse du monde à travers la lentille des données et plus généralement des technologies, tout en essayant de demeurer fidèle à Descartes en faisant usage d'une bonne dose de doute.

Billet suivant

Précis de mode de scrutin à l'attention des québécois et des autres