In my work in Open Data I often talk about how the data should not be leading when exploring the possibilities of re-use of said data. Instead we should look at actual issues, questions, problems, and itches to scratch. And even then we shouldn’t look at the data, I think, but first take a closer look at all the stakeholders involved with the issue you want to address.

Especially so if you are a public sector body, such as a city government, that is trying to decide with which data to start, or which data to publish next (and while pro-active routine publishing as part of regular policy processes is not in place yet).

In my work with local authorities this is also how I build an internal business case, and how in workshops I help civil servants to perceive open data as a policy instrument they can use to work towards their own policy goals in a more efficient or effective way.

First take an actual policy issue or policy task. As concrete and tangible as possible. Then list all the stakeholder groups and organizations connected to that issue or task. Again as concrete as possible (so, not: seniors, but women 50-65 in neighbourhood x, and not: university, but research group y at faculty z). And as step three name the various types of data connected to that issue, regardless whether you have it or somebody else has it.

With those three points (issue, stakeholders, data sets), think about what new type of actions, connections or conversations become possible when all stakeholders would have access to those data sets you listed.

In short, open data works in triangles. (As I previously wrote in 2006 about social media working in triangles)

From each of the points of one triangle other triangles may be constructed. Stakeholders can be connected to other issues and other data sets, and vice versa.

It’s these triangles that make me say over and over again: I blindly accept the challenge to for any public sector body build a business case on how they can save money or be more effective with the same money by turning data into open data. Any takers?

(also cross-posted to the ePSIplatform blog.)