Yesterday I went to ScotGovCamp in Edinburgh and had a lovely time. Spent more of it chatting in the hallway than participating in the sessions; but have detailed notes from the Open Data session led by Chris Taggart of Openly Local, and scatterings from elsewhere.
Chris cites his membership of OKF’s Open Government Data Working Group, the London Datastore advisory body, and the Westminster Local Public Data Panel. Good, we now know we are dealing with a pretty serious guy.
His focus has been on the “English Experience” and he’s come to make contacts in Scotland. Citing as recent developments with impact yet to be fully felt, the Ordnance Survey Open Data release and the disclosure of Westminster MPs’ expenses. Looking for “drivers and levers” that will surface as yet unseen issues in local government.
It’s much less clear (at least here in the UK) how local, as opposed to central, communications and decision-making networks actually work. Local authorities are in an unclear legal situation – European PSI law should oblige local government to publish more data, but the knowledge of the law is often just not there (people are too busy).
OpenlyLocal has been going for a mere 15 months. It was inspired by a Manchester version of They Work For You and by the ScraperWiki project. OpenlyLocal collects information about local government data sources and critically the people involved, the social networks involved in decision making at council level. The site now has some amount of data (scraped from websites and republished as Linked Data) for 158 councils in England and Wales – but for only 4 in Scotland. One ultimate aim is to encourage local authorities to re-adopt the data, and the practises, being created by Chris and the contributors to OpenlyLocal. Other motivating things for publishing local administration info, as pure data:
- Accessibility concerns. Publication of data, as opposed to pictures of data (like PDFs) avoids accessibility concerns. Creation of interfaces to data is expensive and incurs a maintenance burden…
- Possible to tie in to other hyperlocal resources – a good example in Edinburgh is Greener Leith
- Creation of an index, or directory, to existing council resources, that is easier to explore than a conventional website
Chris outlined 4 key reasons why open local data is important (though the reasons seem to alter with every re-telling).
- Transparency – we can see for ourselves, and draw our own conclusions.
- Engagement – citing Planning Alerts – casual engagement is possible, you don’t need to be obsessive
- Equality – “open data is about equality of access, because all this data is currently available for a price, and that’s not right”
- Relevance – to local temporal reality of affairs – less decoupled synthesis of prepared or reported data – just data.
“Quality of data is important and opening that helps (and is used as a blocker) but not as important as other points”
Can we make interfaces that work for our grandparents?
“There’s a much bigger step between creating nothing, and creating something, than between creating something stupid, and creating something great… just make a start, somewhere, anywhere.”
To local administrations – “it should cost nothing to release open data. If it doesn’t cost nothing, you’ve got a really bad outsourcing deal”.
To everyone else – “Fundamentally, it’s our data.”
Questions about quality
Recently, I’ve been thinking a lot about data quality within the geo ghetto, so it surprised me to hear several audience questions from local administrative workers, directly asking about data quality. How imperfect/unreliable/uncertain is the data? Given inevitable uncertainty, how is this doubt stopping us (or the decision makers for whom we are responsible) from opening data?
Data quality problems can have severe cost and social effects – one case cited was a database recording details of children, in which 5% of dates of birth were wrong, so 5% of people are being treated administratively as children when they are not, or treated as adults when they are not (at least according to the administrative definitions, processes etc).
It’s quite possible to measure quality, to test and to describe it. Data package tests, like software package tests, extracting what’s useful from the formal standards thinking on quality. But this is too much of a digression, some of which is here, some of which is on the way.
Law and Computers
An interesting session which i only caught the end of and is more fully described on the ScotGovCamp blog by my EDINA colleague, Nicola Osborne. My notes say this:
German reform in the early 19C. | Biblical census. Legislation | Standards | Influence e-Care records, ATOS Origin distributed versioning in citizen data - propagation, provenance, merging. Robot Queen? Automaton? Target specification - e.g. music education, department of education directive. specifications, models, records management overspecification in law, cost, fear. Westminster Information Act (ontology-like)
Dropped into the session on cuts, which wasn’t all gloom and doom, but more vendor optimism about shared services. Asked vendors about whether they made free software, or could find a place where business benefit to themselves and organisational benefit to their (public administration) clients could be created by freeing their software (in parallel to building shared hosted services). Not sure there was an answer.
Wondering about open demographic data, social credit data, and what’s the non-proprietary answer to Experian.
Good comments from Chris Taggart in this session too – “specialising in one thing, as a service provider. Low barriers to entry – low barriers to exit equally important”. Wondering about a JISC-like body for stewardship of shared services for local authorities. Would probably become a beast.
Fragments of insight
The big consultancies that form consortia to do government work, work by mimicry – by mirroring the hierarchical administrative structures that they are serving. But then internally, they actually do iterative micro-procurements – as in EU consortia the bulk of the actual work is done by very small providers. Many large and small companies work across local authorities, and it would be fascinating to see the map of who and where they are, which Chris is beginning to derive from spending data.
Shadow networks, shadow systems form, inevitably, in organisations at scale. But a paradox – the more superficial openness there is (coming from cultural change, or coming from legal or quasi-legal mandates, or meeting in the middle) the less is actually recorded. Data implies audit, audit invokes fear of loss. So organisation becomes about emotional concerns – perhaps it would be helpful to recognise this more?
Note, i corrected a bit of this, Equality rather than Quality, with which i must be temporarily obsessed. Thanks Chris for notes. Thanks Tim Howgego for insights.