Terence Eden from the Government Data Service had one of the most reacted-to pitches at Open Data Camp 4. Surely, he suggested to the more than 100 attendees packed into Cardiff’s Pierhead, data should always be released as pdf?
Of course, this was a joke. And at the session on ‘what open data standards do we need’ he said he had insisted that government departments released data in open document format.
This wasn’t openness for openness sake, he added. It was because he didn’t think it was reasonable for open data users to be expected to buy licenses for expensive, proprietary database and software projects where good, open and free alternatives existed.
If they used proprietary standards, he added, government departments, councils and other organisations could find themselves releasing data, and even thinking they were releasing open data, without releasing “usable data”.
Eden argued that users could and should demand open standards. “One thing I have discovered is that councils are very competitive,” he said. “If you go along to a council and say: ‘You are releasing data in this format, while [another council] is releasing it in this lovely open format, and I can do loads of things with it’, you can get them to do something about it.”
The FOI lever
The Freedom of Information Act could also be a lever, he suggested. “Sometimes, people say: ‘We don’t want to release this data as open data’ and I say: ‘Could this data be FoI’d’ and they say: ‘Suppose so’ and I say: ‘Well, think how much less hassle it will be to be able to say it’s all on the website’…”
Eden really wanted to use the session to find out what standards open data users wanted. Which, inevitably, raised the question: “What is a standard?” Eden said that by “standard” he meant “the way two computers communicated with each other to exchange information.”
Or: “If you give me data, this is how it is laid out – so html is a standard, cvs is a standard.” Naturally, as participants pointed out, this doesn’t mean that other factors, from the culture of data release, to where data is held, to how it can be accessed and searched, to rules about attribution and presentation, to the quality of the data itself are not important.
Approved and desired standards
To focus debate, Eden outlined a couple of standards that have been approved – the Universal Coded Character Set, or Unicode, which is an international standard for the consistent coding and representation of text, and the International Aid Transparency Initiative, which is an inter-government initiative to release data on where aid is directed and what it achieves.
Where else could standards be usefully applied, he asked? One suggestion was humanitarian response – at the moment, official reports on who responds to a big disaster, such as the Nepal earthquake, take time to issue; open data to agreed standards could enable this information to be collated faster.
Another suggestion was data on modern slavery. Other participants asked for a standard on data on the location of polling stations – which are, apparently, surprisingly difficult to identify. Other ideas can be sent to @edent.