Post: 2 July 2017
Britain came joint second in OKFN’s Global Open Data Index (GODI) for 2016, tying with Australia and behind Taiwan. Taiwan won in 2015 also, with the UK dropping to second after leading the table in 2013 and 2014.
2016 was the first year in which Northern Ireland was listed separately from Great Britain in GODI. Northern Ireland has made substantial progress in its open data programme over the past year or so and was ranked in tenth place.
GODI is a multi-dimensional index that ranks countries based on the quality and characteristics of the open data they publish across a range of categories.
There are 15 categories in GODI 2016. Administrative Boundaries is a new addition, and Legislation has been split into Draft Legislation and National Laws. Pollutant Emissions has been replaced by Air Quality.
I was involved with most of the GB submissions this year and I had some concerns about the design of the 2016 process. In this blog post I mainly want to look at the gaps in key GB open data identified by GODI, so I won’t go into those concerns in any detail. However it’s important to understand that the submission format and the underlying scoring method were somewhat more nuanced than may be apparent from the summary information linked in this post.
Full marks in six categories including Air Quality
For 2016 GB scored full marks (100%) in six categories: Government Budget, Company Register, National Maps, National Statistics, Administrative Boundaries, and Air Quality.
That’s a slight improvement on 2015; previously UK lost marks on the basis that air quality monitoring data wasn’t available in bulk, but OKFN has now accepted that Defra’s download facility meets that criteria.
Legislation: marks lost for lack of bulk data
Whereas in 2015 UK scored full marks for Legislation, in 2016 GB scored only 85% in the National Laws and Draft Legislation categories because of barriers to accessing the information in bulk. Legislation.gov.uk, which publishes enacted legislation, does make bulk downloads available but the backlog in incorporating revisions may have been a sticking point.
Procurement: marks lost for lack of bulk data
As in 2015 GB scored only 85% in the Procurement category because the current Contracts Finder site does not provide any bulk download facility. (Data snapshots are available via Data.gov.uk, but only for each day’s activity.)
Water Quality: a big improvement in England
After some considerable discussion GB scored 85% in the Water Quality category, losing marks only because the available data isn’t frequently updated. This is a significant improvement over 2015, when the UK scored only 10% in this category. The main difference has been the launch of Environment Agency’s Water Quality Archive site, which currently provides data for England up to 2016.
Weather Forecast: marks lost for registration requirement
In 2015 UK scored 100% in the Weather Forecast category, but in 2016 GB scored only 85%. The difference was a change in OKFN’s criteria: GB lost marks because Met Office’s DataPoint service has a registration requirement and its open data is therefore not “publicly accessible” to the same standard as data on the public web.
Government Spending: marks lost for timeliness and lack of bulk data
GB scored only 70% in the Government Spending category because some departments and agencies are more diligent than others about keeping their data up to date and because there is no central facility for downloading spending data in bulk.
These are long-standing issues, and it is difficult to make excuses for GB’s lack of progress in this category given that spending data was the central theme of the UK Government’s open data policy when it was launched in 2010.
Locations: marks lost for no open addresses
GB scored only 50% in the Locations category, compared to 100% in 2015, because OKFN’s criteria have changed: the Index now expects geocoded location data to be open at address level, not just postcodes. GB has had open geocoded postcode locations since 2010, but address data remains closed.
The national address dataset is widely considered the largest gap in the UK’s open data infrastructure, so in principle I’m glad OKFN is pushing for this as a standard requirement worldwide. However it’s worth noting that Northern Ireland, which has neither open postcodes nor open addresses, also scored 50% in the Locations category. That points to a flaw in OKFN’s approach.
Land Ownership: no open data
GB scored 35% in the Land Ownership category, based on datasets held by Land Registry and Registers of Scotland. This data isn’t open or freely available but meets the criteria for timeliness and open formats.
Election Results: nil marks
As in 2015, the outcome in the Election Results category is contentious. The UK has good quality open data for results of national elections; see my recent blog post. However OKFN’s criteria require data at polling station level.
In the UK votes are not counted based on polling stations; there is a statutory requirement that ballot boxes are mixed prior to counting. Accordingly there is no results data available at polling station level, open or otherwise, and both GB and NI have scored nil for this category.
Although the UK has no recent history of significant electoral fraud, there are potential analytic benefits from publication of more detailed election results. Ideally I would like to see the Electoral Commission encourage returning officers to publish vote counts and other data based on counting places, not just constituency.
However, as in the Location category, it seems unfair that lack of granularity should over-ride all other considerations. Nil marks does not accurately reflect the availability of UK open data for election results, and I hope OKFN will rethink its approach for next year.
Image: open data (scrabble) by Justin Grimes, which I have lazily re-used yet again under CC BY-SA 2.0.