Skip to the content

Leading government's 'data revolution'

07/04/16

Share

Interview: Paul Maltby, data director at GDS, talks about the challenges of data science, registries, open data and finding the balance for sharing

A conversation with Paul Maltby about data creates an intriguing sense of excitement tempered by patience. He's an evangelist for government doing a lot more with data to save money and improve services, but also emphasises that there are barriers to overcome, plenty of unknowns and a lot of work to do before realising the promise.

“We're only at the foothills of understanding the improvements we can make, and bringing in the technology, skills and processes to be on that,” he says. “But it's real progress and data science is one of those expressions.”

As director of data in the Government Digital Service (GDS) Maltby is quick to say that a lot of important work is being done by individual departments' data teams. But he is Whitehall's lead official in the field, overseeing the “data revolution” talked by up Cabinet Office Minister Matt Hancock, and is active in pressing it forward on every front.

One of the most ambitious elements is in getting to grips with data science, with GDS teaming up with the Office for National Statistics (ONS) and Government Office of Science in the Government Data Science Partnership. It is a field for which public and private organisations have found their own definitions, often with sharp differences, but Maltby is clear on how he sees it.

“At heart it's people with analytical capability, plus the ability to code and access the libraries through RPython (an advanced coding language) and other tools that would not have been available to analysts, plus some of that mindset altered by the technology,” he says.

Experimentation

This mindset goes beyond that of the traditional analyst, into an eagerness to experiment with data, often with no set idea of what might emerge.

“It's sort of a hacking culture. People can run the data quickly and try lots of ways into a problem. Instead of setting out with a hypothesis in a strict delineated way, you can run the data in different ways to see what sort of patterns you can find.

“Over the years I've seen fabulous government analysts who have that curiosity and playfulness, but not everywhere.”

On a basic level this can produce easy visualisations of data, and Maltby shows how the Department of Communities and Local Government has produced graphs and maps revealing the causes of fire and rescue call-outs. But it also has the potential to provide more complex insights, sometimes unexpected, to feed into policy-making.

Capability priority

Although data science has been on the agenda for a couple of years, there are still relatively few specialist practitioners in government; Maltby estimates 120-150 in the “non-secret” parts. Building up the capability is a priority, he says.  

“To support the effort we have the accelerator programme, matching people with mentor, machine and project, and for now it's small scale with about 20 people going through it,” he says. “We're trying to keep our demand and supply in a degree of equilibrium.

“Part of of our approach is recruiting people, part training our own people, and building the awareness among those that are the decision-makers – policy, operations and commercial people – that there is a different game in town. We haven't started that in detail yet, but there are things like work with the ONS for the policy profession on building a statistical capability into the training.

“Part of what we do is showing what is possible to a wide audience, whether it be permanent secretaries or ministers, through the projects that we do. Ultimately it won't be me and GDS changing the thousands of analysts across government, it will be the big players, like HMRC, working at scale.

“We wouldn't necessarily want thousands of data scientists without an understanding about how to use them and how the career path would work, but it's relatively early days.”

If there's a measure of success for this, it could be in making data science as significant as open data. Maltby's team is promoting the cause inside government, and he believes there has already been a big change in attitudes, pointing to more than 20,000 datasets on Data.gov.uk, the Department of Environment, Food and Rural Affairs' big campaign to release its data, and the way it has transformed information on public transport.

Tipping point

“We're past the tipping point,” he says. “There are individuals in the mix who still struggle with it, but overall we're seeing departments racing ahead in being able to do this. They realise what it can provide.”

There is a possible impediment to progress in some areas, as data is in the hands either of public organisations expected to fund themselves or private companies running public services, both concerned with the commercial value.

Maltby acknowledges it is not a straightforward issue and hedges his bets on the outcome, talking about finding the right balance between making it easier for analysts and developers. But he makes an approving reference to the Government's plan to require private bus companies to provide more real time information, and emphasises the relationship between public funding and opening up data.

It conveys a perspective that the impediments are mainly there to be overcome rather than to obstruct the campaign.

One of the challenges comes from differences in the same data held by different sources. This underlies one of the key planks of the Government's data infrastructure programme, the creation of an ecosystem of registers with common naming standards and application programme interfaces.

Unified registers

The first step is step to establish a set of unified registers for Whitehall, with the Foreign and Commonwealth Office having recently released the first – a list of foreign countries. Maltby says this will make things easier for developers and analysts, removing the confusion created by the previous existence of 14 central government lists, and that in the next few months there will be registers of local authorities, companies and schools, with further plans for an open address register.

“We're thinking of a number of domains rather than specific datasets, because as we look at those domains we find other things,” he says. “For example, on the countries register we also need one of territories – UK territories, overseas territories – so we're working with the Foreign Office on one to complement the countries register.”

It prompts a question as to whether there could be competition between organisations to ensure their register becomes the single source. He responds by pointing out that it is a big change, everyone involved is finding their way, and that organisations have had reasons for using their own registers. But again he expresses faith in the process, as long as it involves some patience.

“It's important that when we go through this process and start to publish some of the technical specifications, that people go through user research and understand there are users of their data who are not just in their department.

“So we've been talking about a Register Design Authority. There's a bit of progress to understand how that will pan out over time. We have a group of data leaders and a ministerial group if there is disagreement. We have governance mechanisms for the first time; but so far we're just trying to make progress.”

Sharing dilemma

While all this is going on, his most pressing priority is the Better use of data in government consultation, which began at the end of February and is due to close on 22 April. It includes proposals for some tightly managed mechanisms for sharing personal data, an effort to get to grips with the dilemma that sees the potential for better services while being anxious over violations of privacy.

This has been a bugbear for government for several years, but Maltby suggests that changes in technology are making it a little easier to overcome.

“In the past we thought of data sharing as a bulk dataset, but as the technology changes there is the ability to query data through APIs,” he says. “It's not the same mechanics and that changes the game a bit.”

He says there should also be legislation to clarify what is acceptable, and that civil society activists had been consulted in forming the proposals. But the timeframe for any parliamentary bill is much more in the hands of politicians than civil servants.

“We're trying to get right balance between not having unnecessary rules and recognising that personal data should not be shared willy nilly across a range of people. We share the concerns about that.”

It all conveys a careful approach to a revolution, but something with more purpose and control than an evolution. Maltby and his team are going to have a high profile in the effort, but he is also eager to make clear that it will be spread around central government.

“We lead the programme in GDS, but it relies on the efforts of data leaders across Whitehall, and there is an understanding at a political and senior leadership level that this stuff is going to make a difference.”

Register For Alerts

Keep informed - Get the latest news about the use of technology, digital & data for the public good in your inbox from UKAuthority.