Government's Big Data dilemma - building public trust during a data science skills crisis
- Summary:
- Big Data could revolutionise UK public sector service delivery, but there are barriers in the way, not least the need to build trust among citizens, says John Manzoni, Chief Executive of the Civil Service.
Public trust is crucial if government is to tap into the potential of Big Data, but of equal concern is how capable civil servants are in their ability to exploit and manage that potential.
Those were the two main challenges posited by John Manzoni, Chief Executive of the UK Civil Service, at an Reform event yesterday which looked at how Big Data might transform public service delivery in the UK and beyond.
Much of what Manzoni had to say was couched in all-too-familiar terminology, bordering on the cliché - Domesday Book and Rosetta Stone are data, information is power, need the public sector to catch up with online retailers use of data etc etc.
But alongside all this was a clear rationalisation of both the potential and the challenges of Big Data in government, even if some of the practical steps to tackle these seemed a tad more long term in their impact. Manzoni said:
With the evidence of data we can spend less time developing policy and services that don’t work, and instead focus on continuously improving those that do. I want people to turn to digital public services as readily and confidently as they do when shopping, socialising or checking bus times. By doing so, we can actually change the way citizens interact with us - making the relationship we have with them more transparent, more responsive, and based on increasing levels of trust.
He cited some early exemplars of this sort of transformative activity at major government departments and agencies, including:
In DWP, for example, providing job seekers with more targeted advice, and opportunities that closely match their personal profiles. The department is also working on data-informed tools, such as interactive visualisations of benefit claimant trends.
In June last year, for example, Land Registry and partners published the first UK House Price Index, and provided a single source of information as opposed to the multiple competing versions which existed before. Land Registry data has also been used to create a range of information services. From whether rude-sounding street names have an impact on house values - they do! - to more serious matters, such as whether your home is on a floodplain. Land Registry’s Flood Risk Indicator service uses data from the Environment Agency to identify flood risk for any registered piece of land within England and Wales.
The Companies House Service gives us free access to real-time information on companies. It’s receiving millions of search requests every day from people checking supplier and customer information. The service can also be used for more mundane but practical reasons - if you’re getting in builders to do work on your home, you can go on the Companies House website and check them out first.
Big Data can help to deliver societal benefits, argued Manzoni:
The Home Office Child Abuse Image Database has transformed the investigation of child abuse crimes and child protection. It won the Civil Service Innovation Challenge in 2015. The database brings together all the images of abuse that police find. Using the images’ unique identifiers and metadata, they can check devices they’ve seized from suspects against the material on the database much more quickly.
Previously a case involving, say, 10,000 images, would typically take up to three days to review. Now, it can be reviewed in an hour. So, we have a process that is cheaper, less labour-intensive and more efficient. This is all good. And it makes the investigation and prosecution of these appalling crimes vastly more effective.
There’s also potential for Big Data exploitation to help prevention activities in the cash-strapped NHS, although the example Manzoni cited of Moorfields Eye Hospital involves a private sector third party provider - in this case, Deep Mind - having access to healthcare data. This is something that still provokes splenetic outrage in Daily Mail-land, which doesn't help public confidence, at least in certain quarters.
That said, it’s a powerful use case. Manzoni explained:
At the moment, clinicians rely on complex digital eye scans. 3,000 of these scans are made every week at Moorfields. But traditional tools can’t explore them fully, and analysis takes time.
Moorfields will share a data set of one million anonymised scans with Deep Mind, who will analyse them using machine-learning technology. This can detect and learn patterns from data in seconds, to quickly diagnose whether a condition is urgent.
With sight loss predicted to double by 2050, the use of cutting-edge technology is absolutely vital. The right treatment at the right time can prevent many cases of blindness or partial sightedness. Up to 98% of sight loss resulting from diabetes, for example, can be prevented by early detection and treatment.
Building trust
So far, so clearly beneficial - but that Daily Mail-land reaction to sharing healthcare data is a cautionary reminder of the challenges that accompany all this. Manzoni acknowledged that winning over the public’s trust is critical to any data-driven transformation of service-delivery:
Now that we are openly releasing information, we have to do so responsibly. Trust means giving people confidence that their data is used appropriately and effectively, and that it’s secure, particularly when it’s being shared by different authorities. That trust has to be earned.
As well as the inevitable name check for the Verify digital identification scheme, Manzoni pointed to the Government Digital Service’s publication of an ethics framework for public sector data science and the Office for National Statistics adoption of the‘The Five Safes’ framework - Safe people, Safe projects, Safe settings, Safe data, Safe output - as examples of initiatives designed to build that trust.
It’s also important that the public is clear on what data is being held by government and the use to which it is being put:
Transparency is part of this - transparency of evidence, ‘showing your working’, and opening up to greater scrutiny the data and analysis on which we base policy decisions. For transactions such as driving licence and passport applications, users can now see the data government holds about them and change it if it is wrong.
All of this is clearly well-intentioned and ticks a lot of the right boxes in terms of data management and public trust, but while the policies might be emerging, there’s still a major problem - the civil service doesn’t have the data science skills it needs to have in place for citizens to be confident that they can put all this into practice.
Manzoni pushed the angle that other countries have the same sort of problem, citing predictions that by 2018 the US will be short of 190,000 data scientists. But that’s something of a distraction. Manzoni's not in charge of the US civil service, but the UK one and that’s where the public needs answers from the top about what’s going to be done to tackle this skills crisis on the domestic front. Manzoni said there are moves afoot:
In the UK Civil Service, we are growing the specialist data science community in a variety of ways - from direct recruitment to training to defining new career pathways for analysts. The Data Science Accelerator Programme is tapping into the 3,000 or so analysts from other disciplines looking to develop their data science skills
A Data Science Campus opened its doors at ONS’s headquarters in Newport last October. And the first intake for a new Apprenticeship in Data Analytics started work on their two-year vocational training programme at the end of 2016.
And because everyone at every level should have an appreciation of the power of data, we’re developing a programme in data literacy for non-data specialists. The Digital Academy will provide skills training right across government for up to 3,000 people a year.
But all this is going to take time and it needs to move faster than it currently is. The word Manzoni used to describe the impact of the initiatives above was “nudging” the government towards the necessary cultural shift. Given that ‘nudging’ is hardly the sort of revolutionary terminology that might be hoped for, it is at least reassuring that Manzoni admits:
We have to step up our efforts.
My take
The theory, principles and the policies articulated by Manzoni are hard to argue with, but the speed at which this data revolution is proceeding still leaves much room for improvement. The civil service boss concluded by pointing to plans to appoint a new Chief Data Officer, whose role will be to oversee this agenda, and a cross-government senior Data Advisory Board. Those are going to be critical appointments if the Big Data agenda is to be met in UK government circles.