“Under this strategy, data and data use are seen as opportunities to be embraced – rather than threats against which to be guarded.”
This was the pledge of then-digital secretary Oliver Dowden, in his foreword to the government’s first National Data Strategy, published in September 2020.
The document, which set out a wide range of initiatives and ideas to improve the use of data across the country, was more than two years in the making. During this incubation period, many of us unexpectedly gained a great deal of experience in data analysis, as a result of months-long periods of making daily studies of case numbers, R-rates, and vaccination uptake.
The urgency of responding to the coronavirus crisis also helped enable and encourage the sharing of data between government departments, as well as with other bodies across the public sector and beyond.
Another two years on from the release of the strategy and, according to participants in a recent webinar discussion hosted by Civil Service World, it is important to try and maintain the momentum gained during the pandemic – and guard against familiar barriers to data-sharing creeping back in.
Sue Bateman, interim chief data officer at Central Digital and Data Office, says it is crucial to have “sustained leadership and sustained focus” now government is moving beyond crisis-response mode.
“The pandemic… has shone a spotlight [on data use] for ministers and senior leaders, and that is the thing I think we need to sustain,” she says. “Some of the work we now need to do in delivering the strategy is probably a little bit dry, from their perspective – but it will lead us to being able to say that we are transforming things.”
Some such things are already well on the way to being transformed, she adds, picking out numerous examples where data is being used in support of not only back-office operations, but in the delivery of some of government’s most critical services. Bateman and her colleagues are looking at such examples to see what can be learnt from them to be standardised and replicated elsewhere.
“There is a lot of work around operational services and real-time decision making, whether it is to support someone in getting a benefit, or making real-time decisions at the border,” she says. “We are trying to look at that and find where is the commonality among such a range of use cases, but also such a range of domains and sectors – [including] criminal justice, energy, regulation, benefits, tax and health – which are all really different use cases; we are trying to come in and find those areas that really are ripe for standardisation.”
A lack of standards or common terminology was one of a number of potential hurdles to data-sharing identified by panellists on the webinar, which was sponsored by SAS, Capgemini, and Intel.
The division created by departmental boundaries is another issue that has long been recognised. Steven Burgess, senior consultant for government at SAS UK, says: “Some of the challenges are due to the sensitive nature of data and [concerns about] the purpose for which data is shared and the approach by which it is shared – but also then controlling the ongoing usage of that data once it has been shared for a particular purpose. Because of the way some departments operate, we have lots of siloes of information for which there are both technical and policy barriers – and that represents a significant challenge.”
Simon Pearson, vice president, tax and trade at Capgemini, says that, for government data teams, removing barriers is also a process of “getting the enablers” right to achieve their ambitions.
“It is about really understanding what the end-to-end needs to look like, and how you overcome some of the cultural issues around data-sharing – with an acute focus on standards, definitions and the things that affect data quality,” he says.
Similarly, government’s heightened focus on the risks created by sharing sensitive information can translate into an awareness of data-security issues that can, in turn, enable siloes to be broken down.
Paul O’Neill, at Intel says: “Government and departments, as they are sharing data, should be looking at new ways and new methodologies to share data in encrypted format – so they can retain technological control of data as they move out of these siloes. With advancements on artificial intelligence and analytics, encrypted data has to be in the spotlight.”
A bias view
Andromachi Tseloni, a professor at Nottingham Trent University and the academic lead for the Ministry of Justice’s Data First programme, also recognises that, in addition to new opportunities, AI and machine learning will create new challenges for public sector data sharing. To meet them, the civil service needs to ensure it maintains the necessary expertise.
“Institutional memory [is important] to tackle AI bias,” she says. “We need to have a very good understanding of what data talks about: who does it cover and what issues – this is a lot of knowledge investment and, from my experience, the turnover of government analysts is a big limitation.”
According to Tseloni, such issues mean that progress on sharing and using data for public good is often “not a straight line – it is more like a zigzag”.
But, even if it is a little up and down, such progress is being made, according to Bateman, with CDDO and the government’s Centre for Data Ethics and Innovation having just completed some pilot initiatives to encourage algorithmic transparency.
“We have been getting organisations to report where they are using algorithms, including what are they doing, and what data they are using – and bring some transparency to what can feel like a hidden area,” she says.
Register here for free to watch the full webinar discussion, including lots more insights on how government can get the most out of its data.