On Thursday 19 November 2015, the Bodleian Library, in collaboration with the Department of Politics and International Relations at the University of Oxford, hosted a one day event for Oxfordshire Sixth Formers to learn more about parliamentary representation and the work of select committees. It formed part of ‘Parliament Week’, a programme of events and activities that connect people across the UK with Parliament and democracy.
Members of the Committee joined the seminar in the Divinity School of the Bodleian Library to discuss the big data with two experts. The sixth formers present were also able to ask questions. (The seminar also discussed the Ebola outbreak, and we included notes on that subject in our Science in emergencies: UK lessons from Ebola report.)
The panel comprised:
The panel noted that the benefits of big data cannot be easily valued or costed, in part because it overlapped to some degree with other long-established analytical techniques and processes. 90% of data was created in the past two years. New skills are needed for big data, beyond those usually captured under the ‘STEM’ heading, concerned with how to store, access and protect and share data, as well as the processing of it.
Ethical issues arose where big data involves dealing with personal data. In particular, issues around obtaining ‘informed consent’ from ‘data subjects’ were difficult to deal with because big data often involves bringing diverse datasets together to bring out new insights and in the process re-analysing personal data that may have collected a long time beforehand. It could be difficult to trace those who had given the original consent to use the data when first collected, to be able to ask them for consent for the new use of the data. The bringing together of different datasets which have been ‘de-anonymised’ can also potentially make it possible to ‘re-identify’ individuals. We need to establish a balance between security and privacy, and the trade-offs made between the two should be made much more explicit.
Privacy issues may relate to individuals, but also groups of people collectively who share particular traits or circumstances. The Enigma decodings in the Second World War presented the Prime Minister with a dilemma about whether to use the acquired data to evacuate Coventry before a bombing raid—new unforeseen dilemmas might also arise from the results of big data analysis.
The meaning of ‘informed consent’ may need to be rethought. It will depend to some degree on different cultures. Scandinavians were generally more amenable to their data being used by the State, which is more open with the public about what consent means and what the government will and will not do with public data. Penalties with appropriate repercussions for improper use of big data need to be part of a consent regime. Consent is generally also likely to be easier to obtain in cultures where the big data benefits for society are clearly seen, where processes and results are transparent, and where people can readily see the policy trade-offs being contemplated from those results. People need a vehicle for have an intelligent discussion about how data will be used to allow them to give consent which is ‘informed’. Consent should be built in to everything we do with data.
Problems arise when people think their data is being sold and feel powerless to stop this. Particular care is needed where data relates to minors, who may need additional protections about how their data are used and before their consent is obtained.
A danger is that technologies are how data can be used are advancing more quickly than the skills are evolving which will be needed to use the data safely and to protect privacy. It is people rather than technologies that go wrong in most systems.
Some big data are not about personal data, but commercial or research data. CERN is collecting massive volumes of non-personal data from the Hadron Collider.
Prepared 11 February 2016