In September 2020, the Department for Digital, Culture, Media and Sport published a consultation on the National Data Strategy. This response was originally submitted on 9th December 2020.
Chapter 1
The objective of the National Data Strategy is to build “a world-leading data economy while ensuring public trust in data use”. These are very worthwhile aims; but data and its uses across the wider economy and society is too broad and varied a set of issues for a single national strategy to meaningfully cover.
The Strategy includes a focus on public sector use of data, coordination between departments, and digital capability in government, and we think these are the right places to start. We also strongly welcome the emphasis on inclusion, data ethics and the opportunities for using data well. However, to make real progress, the Government needs to be more ambitious in rationalising the institutional and structural “plumbing” behind public data governance and data policymaking. The omission of opening up postal address data from the Strategy, which would require significant coordination between public bodies, reflects a wider need to reform organisational responsibilities and capability around data to improve public sector data infrastructure. In summary, this response argues:
The government should focus on a public sector data strategy as a first priority instead of a one-size-fits-all, society-wide approach
Overcoming organisational and cultural barriers to an open address system could be a very valuable first step in providing effective public data infrastructure
Progress depends on clear responsibility, authority and accountability for making data across the public sector work
Chapter 2
1. The government should focus on a public sector data strategy as a first priority instead of a one-size-fits-all, society-wide approach
In order to be most efficient, the Government should identify the areas where national-level top-down action is clearly essential (for instance, putting in place foundational legislation for data privacy, cybersecurity and international data flows). Beyond these specific cases, it might not be possible or desirable to develop a comprehensive, top-down policy approach. Aggregating everything related to ‘data’ into a single, national framework risks being too shallow and broad to be effective. It would be better to take each question in turn and establish the right governance approach on a case by case basis. For example, this would mean disaggregating the question of whether government should act to encourage access to private sector data to how it should regulate the carbon impact from datacentres.
This is arguably reflected in some of the Strategy’s potentially-conflicting ambitions: for instance, it seeks an EU data adequacy decision while also exploring divergence to “improve on international transfer mechanisms”. It also aims to give individuals control of data sharing across public and private sectors, but doesn’t want to limit innovation and digitisation.
These are all honourable ambitions, and many of the goals in the strategy’s pillars, missions and opportunities are uncontroversial. But government policy should be honest about the tensions and trade-offs, for instance minimising compliance costs for private organisations versus exploring divergence in data regulation, or putting users in control and promoting responsible data whilst also removing limits on how organisations can innovate with data.
However, one area where the Government (and only it) can deliver faster progress is on management and openness of public-sector data. We think the National Data Strategy could go a lot further towards providing comprehensive, effective and trustworthy public data infrastructure. A dedicated strategy for public sector use of data – including upgrading digital capability in government, departmental collaboration and opening up access to public sector data – would be a worthy priority with significant potential economic and social benefits. Working on this area first could also help to inform the Government’s wider policy approach towards data in the rest of the economy.
Chapter 3
2. Overcoming organisational and cultural barriers to an open address system could be a very valuable first step in providing effective public data infrastructure
Within the field of public sector data we think there are two projects that the Government should focus on. These would help to serve as pilots and proofs of concept for going further in opening up public data:
Open Address Data
One of the pieces of Government data infrastructure with the greatest impact would be an open postal address dataset. Specifically, a list of all UK residential and postal premiseswith the street address and postcode is not available under an open licence, despite the significant spillover effects this would provide. For example, there was a 3000% return in Denmark on creating an open dataset of unique property identifiers: €62 million (£55 million) in direct financial benefits were generated for the period 2005-2009 after an initial investment of only €2 million (£1.8 million)). In the UK this is likely to be far higher.
The lack of open address data in the UK seems counterproductive. The Ordnance Survey, which currently owns this data, is also owned by the UK Government but charges for access to this data as part of its independent funding. Other public bodies have been granted a carve-out via the 10-year Public Sector Geospatial Agreement, but significant costs remain for small businesses – limiting access to high-quality geospatial data. We think this fails to recognise the broader economic value of open data. Neither the Geospatial Data Strategy nor the National Data Strategy have addressed this.
Integrated Data Platform
The flagship government-facing proposal of the Data Strategy is to develop an “Integrated Data Platform”, a “safe, secure and trusted infrastructure for government’s own data”. While the ability to join up data sets is important, the government should clarify if a single, unified system is planned. While the public sector should certainly aim to use leading technologies to manage, store, analyse and provide access to its data, its focus should be on setting up long-term, reusable infrastructure with high spillovers rather than a single system ‘to rule them all’. A ‘once and for all’, centralised, technical solution would risk creating legacy technology, vendor lock-in and limiting the innovation enabled by decentralised solutions with lightweight central rules. It is also unclear how this project aligns with the responsibilities of the Data Standards Authority.
Notably, the digital government lessons of the last decade that GDS have pioneered should highlight a different approach: that connecting data across government is often best done via individual data registers, each with an API. This is where projects built around authoritative, reusable sources of data (such as GOV.UK Registers, while resource-intensive) were promising. More modular solutions enable far more innovation, minimise interdependences and improve resilience over time, enabling data access without the need for duplication - and subsequent risk of corrosion - of datasets. Decoupling these data functions from policy functions in departments might also promote better use of data and reduce the incentive for risk-averse senior leaders to shy away from opening up data, lest it be low quality and lead to errors.
Chapter 4
3. Progress depends on clear responsibility, authority and accountability for making data across the public sector work
In general, the Government has tended to talk up the opportunities of data whilst under-resourcing the messier work of improving quality and accessibility of government data. For example, the NAO has reported that departments found data infrastructure investments hard to justify as the benefits were often realised elsewhere in government. The Strategy should confront and resolve the institutional and structural issues that give rise to these misaligned incentives. The National Data Strategy does commit to several steps designed to improve public sector data foundations, but many of these lack a plan of action (for instance, “we will…commit to resolving the long-running problems of legacy IT and broader data infrastructure”).
In 2019 the NAO recommended setting up clear, cross-government accountability, governance and funding for data to support delivery of data strategy, however:
Responsibility for government use of data is now to move back to Cabinet Office (CO), having been moved from CO to the Department for Digital, Culture, Media & Sport (DCMS) in 2018. This is a positive step given a strong centre is necessary to make progress, but the Data Strategy still sits with DCMS so there is a clear risk this strategy lacks the necessary authority to deliver on its aims. Meanwhile it is the Office of National Statistics that is hiring a ‘Director of Integrated Government Data Programme’ to be charge of the Integrated Data Platform, not the Government Digital Service (GDS) / Cabinet Office or even DCMS.
The Chief Data (and Information) Officer position at GDS has been unfilled since 2017. GDS also currently has an Interim CEO, with the permanent appointee to report to another new role: Government Chief Digital Officer.
GDS is now leading a new ‘Data Standards Authority’ who will promote adoption of standards to improve data sharing in government. This is a positive step, but departments have historically been reluctant to sort out their data and this new team appears to have no enforcement powers to achieve their aims.
Digital identity policy has made strong progress in recent months, and every use case shouldn’t be wrapped up into a single digital identity strategy. But the proliferation of identity schemes across government should not happen at the expense of meeting user needs, of which simplicity, consistency and convenience are important. Opening up driving licence and other identity data, which are not currently part of the Document Checking Service pilot, would also be greatly facilitated by establishing an infrastructure of data registers independent of policy departments.
Clarifying this confused landscape is not a matter of putting everything in one place, given data governance and management affects every department. A strong central authority equipped with levers and political capital is a prerequisite for encouraging central government departments to adapt in how they manage data. At the same time, it should resist constraining innovation by imposing too-restrictive common standards.
Redesigning the institutional environment isn’t a one-shot undertaking, but it is a worthwhile and necessary starting point for improving the governance and quality of public sector data. Making with this work would put the Government in a good position to understand and improve the way data is used in the rest of the economy.
We would also like to highlight several other consultation responses and reports which informed our analysis: