Somehow I think I managed to carry on talking sensibly enough to feel useful. I also took notes and recordings of the sessions I was in, but here are 5 points from each that struck some kind of chord with me. They're a mix of things people said, and stuff I thought, but I think I can remember which was which.
I've put up audio from each session here.
Session 1: Data Viz + maps issues + challenges
- Vicky Sargent
This seemed to focus largely on data quality.
- Poor quality data can be exposed through openness.
- Different users/uses want different levels of quality/reliability.
- Bringing together those who want reliability with those who want usability is hard.
- Getting useful "infoporn" is hard.
- Start by knowing what you're trying to achieve.
Session 2: Open Data as a Business Model
- John Sheridan
- John Sheridan
This went into how to sustainably fund both open data, and businesses based on it.
- It's not just a choice between "open"/public and "closed"/private.
- Perceptions of data reliability (including being up-to-date) are inherently linked to data management and its economics.
- i.e. Some people think you need a tightly controlled team / contract / business model to maintain data quality... Whereas others think openness is a viable form of reliability. (Cf. Wikipedia "vs" Encyclopedia Brittanica)
- Licensing offers multiple funding models depending on end-user, a la open source software. Chris Taggart doing a lot of this with openlylocal.com
- Does the data business model depend on the size/resources of the dataset/audience?
Session 3: LinkedGov tool to clean up & link data!
- Dan Paul Smith
- Dan Paul Smith
This was a demonstration of the really impressive work being put together by @LinkedGov (http://www.linkedgov.org).
- Software that extends Google Refine to let you easily link data structures and tidy data.
- If letting people edit data ("cleaning", "linking", etc), you have to be careful not to introduce "new" data such as assumed defaults for new values.
- Suddenly linked/semantic data is starting to look really powerful. I'm almost converted :-)
- The ability to "modularise" links to other external data lists has huge implications for data as a Distributed Ecosystem.
- Metadata for what's been edited needs to be accessible and clear, to understand who's done what over the lifetime of a dataset.
Session 4: Data Engagement / Everyday Data
This was an attempt to think about how to get data into something everyday and, not perceived as a "technical" thing. "Slides" and audio available.
- People love data if it's about something they love - e.g. football, F1, sales...
- Language used is massively important - often 2 groups will talk about the same thing, but in totally different ways.
- A range of "necessary" precision was brought up again - how can you transform the complexity of data into simplicity without misleading?
- Does data visualisation have to be 'comprehensively accurate', or can it just be enough to get people to ask more questions?
- Give data context and it's easier to turn data into feedback and so learn from it.
There were some great examples, and some amazing ideas coming out of this for me, which I'll blogpost properly soon.
Session 5: Network society engagement
- Catherine Howe
- Catherine Howe
This wasn't to do with data at all... This was about the advantages offered by moving to a more networked, more engaging approach to decision-making.
- Catherine claims that the current system of engagement/consultation is actually a method to mitigate our own ongoing disappointment in political participation.
- If done better, political participation can be enlightening, rewarding, and fun.
- Bringing people together as part of the consultation process can mean they understand it/others more, and are less disappointed if they don't get what they want.
- Feedback as part of the consultation is vital to success. In effect, consultation moves towards conversation rather than just gathering views. (Does this that changing people's views is an objective of a networked approach? Does this raise questions around the accuracy of the final result, or does a more involved process and more post-process feedback negate this?)
- While I generally agree, I wasn't sure how much success from trials was down to using a networked approach, and how much was down to just using a different approach (i.e. novelty can often be fun in itself) - more consultation iterations needed?
I'm running out of time so won't go over Mike Bracken's speech or the Closing Note. Here are 5 random, general points instead:
- The data landscape is slowly coming together in my head. I know I know something important about it, but I don't know what it is yet. Like Cooper in Twin Peaks. You know, when he has that dream. I need to mull it over and chuck some rocks at a bottle.
- It feels weird not having twitter usernames on name tags.
- GovCampers are a bunch of (mostly beardless) ale-swillers. Much to the surprise of the pub.
- The engineering/development going into the new gov.uk single domain is seriously good.
- The medium T-shirts this year are definitely smaller than the medium T-shirts last year. Or did I get a ladies' one?