You are not logged in. LOG IN NOW >

How to Evaluate the State of Open Data

BY David Eaves | Tuesday, May 8 2012

By tedeytan from Washington, DC (Library of Congress Reading Room Open House 7)

The Open Knowledge Foundation recently announced that it will organize and coordinate an Open Data Census. The intent is to create a basic baseline against which governments can measured around how much (and how relevant) their open data is.
If nothing else the census will be useful is forcing governments to reassess what “open” means. The OKFN has the very reasonable assessment that data is only open if it meets specific criteria, e.g. available in a machine-readable, publicly available, free of charge and openly licensed. (They also include available in a digital form, but one presumes that if it is machine-readable, it is digital). There are, in fact, a number of “open data” portals – particularly among local governments – that fail to meet these basic requirements. Seeing this fact reflected in the census may cause them to alter their licenses and formats.
More intriguing are the data sets the OKFN is opting to count. The details about the selection process are sparse, however. According to their announcement, data sets that will be counted were chosen based on “their breadth and relevance” and what “most governments could reasonably be expected to collect.” This list includes:

Election Results (national)
Company Register
National Map (Low resolution: 1:250,000 or better)
Government Budget (high level – spending by sector)
Government Budget (detailed – transactional level data)
Legislation (laws and statutes)
National Statistical Office Data (economic and demographic information)
National Postcode/ZIP database
Public Transport Timetables
Environmental Data on major sources of pollutants (e.g. location, emissions)

This is an excellent list. There is a solid balance between data that can encourage innovation and participation in democratic institutions and processes (election results & legislation) to data sets that can also help boost productivity across a variety of sectors (demographic and postal code data).
More challenging is understand the methodology of the census. There is a risk that the British biases – with its highly centralized national government – have strongly shaped the census. (OKFN is based in the UK.) Thus, while the census evaluates countries some of the data sets being counted are not controlled by national governments. For example – will national governments Canada or the United States get counted for public transport data if any of their cities release transit data? Indeed, transit data – while demonstrably useful – strikes me as an odd duck choice since it is almost always not managed by national governments. The same can be said for company/corporate registers, in which the most important data sets are managed by sub-national entities. Resolving – or at least making clear – these methodological issues will be important to foster a true apples to apples comparison.
There are, of course, a few data sets I would like to see added to the list. These include:
-Access to Information (ATIP or FOIA) made, completed, rejected and average response time, broken down by government entity.
-Government procurements and contracts, broken down by government entity
-Electoral Boundary Data
-Voting Booth Locations
-Land Registry Data
-Payments to Government for Extractive Industry Resources
-Foreign Aid Data
-Campaign Finance Data
-Registered Lobbyists List

As with the OKFN’s list, some of these data sets are targeted at reducing corruption or enhancing democracy. Others can both enhance democracy and reduce the friction costs in the economy. For example the land registry is important in both that knowing who owns what land is important for marginal groups as it enables them to gain access to capital, helps reduce the transaction costs around exchange land and, enables more transparent political discourse over who has influence, interests and resources in issues surrounding land ownership.
Finally, I would love to see another metric added to the census: the number of data sets that are shared in established schemas or standards. For example, a country that shares its foreign aid data in an open format should be congratulated. However, a country that shares its data in the International Aid Transparency Initiative data standard should receive a higher score still.
Personal Democracy Media is thankful to the Omidyar Network for its generous support of techPresident's WeGov section.