This was the seventh Open Data Camp event, and Valtech were proud to be Silver Sponsors. Open Data Camp is an unconference entirely devoted to Open Data. Held across the weekend of November 2nd and 3rd, this was the first event I'd attended where the entire format was unconference. Having the attendees create the agenda meant that interesting discussion replaced PowerPoint presentations. There was a strong community feel with everyone being very welcoming and friendly. This was also the first weekend event I'd attended. This brought a greater level of focus, in part because everyone attending had committed to invest their own time in the event but also because of the distinct absence of the usual distractions. Nobody needed to duck out of sessions to respond to calls or emails from the office.
Climate change and sustainability. As people pitched their session ideas at the start of day 1, I was drawn to a session in the first timeslot on greening Open Data. Valtech recently ran an event exploring how public data can drive insight and solutions to climate change and sustainability
A challenge discussed at both events was the difficulty in connecting those collecting data with those using it to deliver tangible data-informed outputs. Better transparency of data held, collected and consumed would enable the end to end value chain of organisations involved to innovate further with better ultimate outcomes. Events like Open Data Camp are great for connecting people and sharing ideas about data that is available.
Bias in machine learning. A discussion on AI and machine learning brought up the risk posed by bias in data sets. How do you know what, if any, biases your source data set has that could lead to undesirable results? It's clearly important to monitor the outputs for anomalies. Measurement of error is also an important consideration, something my colleague Dr Jason Ward (Principal Data Scientist) has posted on previously - Know thy error
Selling public sector data. If government data has significant value, should the data be sold with the proceeds reinvested in public services rather than made freely available for all? Could the data be freely available, but revenue-generating services to deliver better timeliness or resilience be explored? Or is this idea fundamentally at odds with the principles of Open Data? This was an interesting question to pose at an unconference dedicated to Open Data. It led to a great deal of discussion, although we didn’t reach a firm conclusion.
Open Data Strategy. A session on Open Data Strategy and associated investment highlighted the conflict between typically speculative outcomes (given its emergent nature) and a need to provide tangible outcomes as justification. Also, there a need to join up Corporate, Data and Open Data Strategy rather than having them as islands. Valtech's approach of thin slice delivery fits well with an incremental approach to investment and early benefits realisation.
Excuses not to share data. It's easy to find them, more difficult to overcome them, but worth the effort! Another session explored typical objections around security and ownership, and how they can be overcome. Signposting can be effective in letting interested parties know where sensitive data exists and can be made available (under qualified circumstances). And data synthesis can be effective in anonymising sensitive personal data to open up access to the research community. The other key conclusion was that data doesn't need to be perfect before it's released to the public, just be clear about its limitations when publishing.
In conclusion, Open Data Camp #7 was packed with interesting discussions with a friendly community feel to it. I'd love to attend next year's event. In the meantime, look out for Valtech news and events as we continue our exploration of how public data can drive insight and solutions to climate change and sustainability.