31 March 2014
On 31 March 2014 a group of civil society representatives, experts and government officials met to plan the data sharing policy process. The meeting, which took place over 3.5hours, consisted of four parts.
- Separate pre-meetings of civil society and government
- A discussion of the process with Francis Maude (Minister for Cabinet Office)
- Small group discussions on the process
- Separate wrap-up meetings of civil society and government
This meeting note sets out the key points of the discussion. To avoid repetition and aid ease of reading, related points are grouped together. Individuals’ comments are paraphrased and not verbatim.
Alan Harriman, Department for Work and Pensions
Amanda Smith, Open Data Institute
Andrej Miller, Department for Energy and Climate Change
Charlotte Piper, Department for Communities and Local Government
David Walker, Academy of Social Sciences
Emma Carr, Big Brother Watch
Guy Herbert, No2ID
Hayden Thomas, NHS
Hetan Shah, Royal Statistical Society
Ian Cope, Office of National Statistics
Ian Wakeling, The Children’s Society
James Edmonds, Office of National Statistics
Javier Ruiz, Open Rights Group
Jessica Adkins, Cabinet Office
Joe Taylor, Cabinet Office
Julian Moss, Cabinet Office
Kam Marshall, Cabinet Office
Lorna Horton, Cabinet Office
Lynn Wyeth, Leicester City Council
Marcus Besley, BIS (GO-Science)
Mark Newton, Ealing County Council
Matthew Bollington, BIS
Matthew Woollard, UK Data Archive
Mitch Riviere, HMRC
Natalie Banner, Wellcome Trust
Paul Arnold, Cabinet Office
Peter Lawrence, Cabinet Office
Richard Lumley, Department for Education
Rufus Rottenberg, Cabinet Office
Rupert Cryer, Cabinet Office
Sam Smith, independent
Sarah Young, Law Commission
Sara Rolin, Information Commissioner’s Office
Sean Kirwan, Department for Health
Simon Burall, Involve (chair)
Simon Meats, Cabinet Office
Sophie Faber, HMT
Tanvi Desai, UK Data Archive, University of Essex
Tim Holt, UK Data Forum
Tim Hughes, Involve (note taker)
Tracey Gyateng, New Philanthropy Capital
Vanessa Cuthill, ESRC
Part 1: Civil society pre-meeting
Simon Burall, director of Involve, introduced the data sharing policy process and Involve’s role as coordinators of the network.
It was questioned who had given Involve the space in which to begin the open policy process, and responded that it had come about due to the experience of coordinating a similar process with the Cabinet Office to develop the UK’s open government national action plan. Government and civil society both felt that that process had worked well, so Involve has been asked to help coordinate a similar process on data sharing. It was stressed that although Involve is being paid by the Cabinet Office to perform the role, it had been agreed that Involve would remain completely independent and would withdraw from the process if it was felt by civil society not to be working.
It was highlighted that the timing of the process just before an election period, with the possibility of a change of government, meant that the institutional context might significantly change and there is a danger that the work of the group might be lost. Concern was also expressed that the timeline might lead to legislation being prepared in haste. It was suggested that the civil society network should establish a boundary to the process that prevents this.
It was questioned what the remit of the discussion would be – specifically whether it would cover data sharing outside of government. It was responded that the majority of the current proposals related to data sharing within government, but a strand of the proposals in the research and statistics component did include sharing data with third parties (e.g. academics) through safe havens for research purposes.
It was questioned the extent to which Francis Maude commanded the full terrain that the process would be looking at, including data sharing by departments with longstanding data sharing practices.
It was suggested that the three components of data sharing – research and statistics; fraud, error and debt; and better public services – were pulling in different directions, and questioned whether they should be included together. Linked to this, it was identified that some issues that would be discussed would raise some political issues – which would need to be untangled from the rest. It was also suggested that careful consideration was needed over whether the three categories presented logical categories – or whether they needed to be disaggregated. In addition, there would be other more general questions of oversight and governance that would apply across the components.
It was suggested that the current proposals were extremely broad and could therefore be used to justify almost any data sharing. A key question, therefore, to Francis Maude would need to be why the government was looking at such a broad power, particularly considering in opposition the coalition parties had criticised the previous government’s attempt to introduce a broad power.
It was highlighted that both local government and business would need to be involved in the process.
Questions about the evidence base behind data sharing were raised, including whether the benefits are really as great as is claimed.
Full group discussion of process
At this stage the civil society and expert group and government group came together.
Introduction from Peter Lawrence
Peter Lawrence introduced the process and the purpose behind it. He outlined that the objective is to develop in a joint way good policy on data sharing. The big issues, he suggested, would be around how data sharing can be done properly, with the right safeguards in place that protect privacy.
He outlined that this was ab unusual process for government, trying to find compromise and consensus within a group that does not just include government, and then going out to the wider public. Involve had been asked by the Cabinet Office to form a network of those interested in having a role in the process, but would be completely independent. Peter Lawrence referred to the Open Government Partnership action plan development process where Involve played a similar role, and its independence had been robust and strong.
He outlined that the reasoning behind the process being open and collaborative is that it is such a contentious issue. It provokes strong feelings from citizens when they are confronted by data sharing issues, whether by the positives or negatives, and views change depending on how people are impacted personally.
Peter Lawrence briefly outlined the three areas of proposals that the government had been working on, including:
- Improving research and statistics – which could enable the use of administrative data to replace the census
- Fraud and error – which would help to detect and prevent fraud, and better manage the debt of citizens to government
- Tailoring of public services – where data would be used to achieve specific outcomes for specific groups
Introduction from Francis Maude
At this point Francis Maude (Minster for the Cabinet Office) joined the discussion. He began by outlining his vision for data sharing and the policy process.
He stated that the issue that the policy process needed to address was how to get the prize of the public benefit that is currently denied because of the barriers to data sharing, brought about by a patchwork of legislation and practice that causes government to be risk averse. It is right that government should be cautious because of the risk and that it should be respectful of the concerns, but it is important that the benefits of data sharing to the public are not denied.
He stated that experience from across government had led to the conclusion that not removing the barriers to data sharing was shortchanging the public. However, the Government had learnt the lessons from previous attempts to pass legislation and understood that the process of developing the policy needed to be done in a different way. It needed to be genuinely open and would only be successful if it achieves a high degree of consensus. The process should be open to all organisations with knowledgable interest at every stage – sharing and listening to concerns and working things out in a genuinely collaborative way.
He asserted that civil society should feel like co-owners of the process, but that this was not an attempt to catch their hands in the mangle – unless there is genuine level of consensus then it would be a failure.
Francis Maude finished by stating that he needed the civil society and expert network to tell him if we are unhappy with the process and outlined the idea of a joint policy paper, where areas of agreement and disagreements are set out.
Discussion with Francis Maude
At this point, questions were taken from those in the room:
Q: What’s currently hard to comprehend is how this is more restrictive than previous attempt. There is nothing obvious excluded from the current scope.
A: What is being proposed is very open to critique and change. Question is where the settling point is that people are going to be comfortable – between something so restrictive (establishing gateways through long, inflexible and un-transparent process) and broad power. Don’t think that anything nearly as all encompassing is being proposed. Important to wrangle through the possibilities so people don’t feel like government is trying to pass a fast one. Other solutions could also be on table instead of data sharing.
Q: How much are the current three components of data sharing hitched together? Can some elements be dropped?
A: Need to have broadest degree of demand. Dangers in being too technocratic. Quite big picture case to be made.
Q: Broad thrust of government is to diminish scale of state – but this could lead to stronger, more intrusive state. How do you square that philosophical contradiction?
A: There’s nothing about believing you should have a smaller state that says you should have an ineffective state. Philosophically, approach is work out what state needs to do and make sure it does it most effectively. It’s difficult to talk about in generalities – you need to get into the detail (e.g. data in save havens isn’t more intrusive – can make state less intrusive by removing need for census).
Q: There’s an opportunity to rationalise data sharing and increase transparency of what data is being shared. How does this fit with consultations that the Law Commission is currently being carried out.
A: The Law Commission is not going to produce a draft bill. It’s identified a problem which requires a sensible legislative solution. Only way that this has chance of happening is by developing a draft bill, with pre-legislative scrutiny, from both houses. Ensuring that there is proper scrutiny so no issues arise later. If enough consensus is found then there’s a potential that an incoming government could go with bill then.
Q: Reason care.data was such a fiasco – nobody felt able to say no to data sharing requests. How does this avoid that trap?
A: This is not meant to be a free for all. It’s meant to be very controlled process. The role of the Information Commissioner needs to be clear here. Where use of de-identified data is being suggested there are lots of restraints there (e.g. certified organisation, safe havens). Key is finding at what point there is enough control that is not so bureaucratic that nothing gets done.
Q: Are you open to increasing the powers of the ICO or introducing a new body?
A: Everything is in the scope of this policy process. This is not about us telling you what we want to do and get you to tell us how to make you feel comfortable with it. This is a genuinely open process. That includes telling us that there’s nothing that you can do that would make people comfortable.
Q: To what extent is there broad understanding across government?
A: Being pushed by other departments to go further. Do get ministers saying their bit is most important, but we’re pushing at an open door. Have a reasonably high level of comfort that the issues of civil liberties can be sorted out.
Q: To what extent will departments understand this process?
A: There won’t be an issue with the legitimacy of the process. Departments won’t care how we get to something, but they will care what we get to.
Q: How to overcome the issue of where people do currently have power to share data but are risk averse?
A: Reluctant to use legislation for something that isn’t legally necessary – but current legislation is so complicated. There would be a huge amount of benefit in just having clarity.
Q: How will this dovetail into proposed EU data protection legislation, which is very strong on gaining consent?
A: Will need to make sure that we’re doing this process in parallel. Hope you don’t feel rushed in process. If people do then it will fail. Have always said right from the outset that we need to take enough time to do this. Will need to make sure we keep aligned with other processes. Public often think that if one part of government has data then the whole of government has it. They’re surprised government’s not doing it already.
Q: Lots of confusion exists over citizens’ rights over their personal data. Part of success of this process would be to consider how to legislate to ensure citizens right to privacy.
A: Yes – absolutely. Very good thought.
Q: People are surprised that data sharing doesn’t happen, but reason that care.data fell down is people did not know who it was going to be shared with. Transparency is important. Is data sharing going to discriminate against people?
A: It will discriminate against people conducting fraud, but I want it to. And the taxpayer wants it to. Will make government better coordinated in collecting debt from those who can pay but won’t, and more compassionate in dealing with those who can’t pay. It will give ability for government to act in a way that is holistic. Data sharing can force government to coordinate.
Q: What do we say to those outside to convince them this is a genuine process.
A: Better to prove by doing. Show our workings. If we have draft clauses we show them to you.
I will be involved along the way – there will be small “p” political judgements to make along the way. Need to hear very directly from those in the network to judge the magnitude of sensitivity versus the size of the prize.
The “What?” discussion explored what output or outputs would be developed as the result of the policy process.
An option discussed within the groups was a joint policy paper, which could become a joint White Paper. This would map the agreements and disagreements between and within the civil society, expert and government groups. The paper might, therefore, include some proposals where there is broad consensus and some where there are competing ideas.
It was discussed within the groups that in the process of developing the output, there would need to be clarity on the aims and broader objects of data sharing in this context, and challenging and critiquing of the assumptions behind this.
Outputs would need to outline the evidence for the need and impacts of data sharing, include further questions that need exploration and set out the next steps as agreed.
Linked to this, groups discussed needing to review the current landscape and conditions for data sharing, and the gaps that resulted from this. This would help to determine whether there was a case for new legislation, rather than alternatives such as clarification through guidance. It was suggested that case studies should be produced that show the need and benefits for data sharing, as well as the potential risks and how they can be mitigated.
It was suggested that the process would entail a series of smaller scale outputs covering these points, finally leading up to the joint policy paper.
There was some discussion on the optimum or minimum level of agreement that should be sought in each of the groups. Prior to the session, a scale was presented by Tim Hughes (Involve) to the group for levels at which agreement might be sought:
- Agreement on approach & detail
- Agreement on approach but alternative proposals on some detail
- Alternative proposals but agreement on the pros & cons
- Alternative proposals
It was suggested within the group discussions that the working groups should start at level 3, and seek to move towards levels 1 and 2. It was also suggested that implicit within the process was that working groups should be aiming for levels 1 or 2. It was also suggested that different areas might move more towards level 1 than the others.
It was also discussed that the three stands might progress at different paces towards different outputs.
The “How?” discussion explored how the process would be carried out, including the principles it would adopt and how they would be put into practice.
There was some discussion about the need for the available resources (which are limited) for the process to be transparent (including people, money and time) in order that it can be judged what is possible.
Clarity would be needed over the aims, assumptions and steps of the policy process, and what “the ask” would be of those involved in it.
It was discussed in the groups that meeting notes should be published online and include a list of those who attended. Linked to this, it was suggested that a list of organisations involved in the process should also be published.
It was proposed that a list of background documents should be published to enable and inform public discussion. Access to relevant documents should, as far as possible, be open and unrestricted.
It was also discussed that documents produced during the process would need to be clear and intelligible to those outside it. It was suggested that a layered approach should be adopted whereby summaries are produced for general consumption, but detailed documents are published for those who need it.
Some concern was expressed about the importance of developing clear milestones for this process, with timescales worked out retrospectively from the deadline for the finalised product. For example, it was proposed that the aim should be to develop first drafts of proposals by mid August, but that this would be dependent on group requirements.
It was discussed the extent to which it should be made transparent where there exist divergent views in Government. On the one hand, there was concern that this might undermine the level of honesty in the discussion, but on the other it was suggested that while this might be uncomfortable for Government, it would also be very valuable. In the same way that it should not be expected that all of civil society will be in agreement on proposals, it should not be expected of government either. The process will need to enable the exploration of divergent views within as well as between different groups.
Linked to this, there was also some discussion about balancing transparency with the need to have something new for Ministers to announce. It was highlighted however that Francis Maude had stated his aspiration for the process to be transparent.
It was questioned how the “outside world” should be informed that the process was happening, with suggestions about communicating the process to the media and involving the public in discussions.
In order to make the process accessible, it was suggested that meeting dates should be set well in advance in order to enable attendance.
It was proposed that regional meetings should be held and how devolved administrations should continue to be engaged in the process.
Questions were also raised how the policy-making process could be made more accessible to those groups with limited resources.
A general principle of involving the right people, at the right time, in the right place was proposed.
It was noted, however, that to make the process manageable there needed to be a way to bring people on early into the process, and then to broadly lock the group down so they can get on with work, rather than having to continuously repeat conversations which have already happened for the benefit of new joiners.
It was agreed that collaboration between the different groups would need to be based on mutual honesty and trust. However, it was flagged that the group should be prepared for things not necessarily going to plan, as well as acknowledging that different groups will have varying “red lines.”
Linked to this, it was highlighted that those involved should flag problems and concerns early in the process in order that they can be dealt with. It was discussed that outputs should capture areas of agreement and disagreement, which should be shared along the way to highlight and resolve issues.
There was some discussion over what should happen where no consensus is reached. For example, should groups be able to veto the inclusion of proposals, should there be votes or should the loudest voices win? However, it was discussed that the output should reflect where positions diverge. Full agreement, therefore, should not be sought across all areas, but individuals should facilitate agreement where it is possible. It was suggested that discussion and agreement may need to be at the level of components, rather than the proposals in their entirety.
It was discussed that clarity over the aims and assumptions behind the process would be needed in order for groups to collaborate effectively.
Questions were raised regarding the power of representatives to represent their organisations in discussions. Linked to this were questions about how “sign-off” would take place – who would need to sign off papers, when and how? It was agreed that there would be a parallel sign-off process in Government and civil society. Also connected to this were conversations over who should write the papers, with the suggestion put forward that the Cabinet Office should draft and then circulate for comments.
The issue of how communication across the civil society network would operate was discussed, and it was suggested that an online forum should be set up to aid collaboration.
The “who?” discussion explored who would need to be involved in the process. In the “what?”, “how?” and “who?” discussions it was emphasised that the process would need to reach beyond Westminster to include regional representation.
Groups that should be involved included:
- Interest groups (e.g. service recipients)
- Campaign organisations (e.g. 38 Degrees, Avaaz, Change.org)
- Service users / beneficiaries
- General public
- Departmental ministers
- Service deliverers (public, private & charity)
- Operational staff
- Data protection practitioners
- EU stakeholders
- Law enforcement agencies
- Central and local government
- Industry leaders
- Data providers and users
- Legal advisors
- Independent information lawyers
- Statistics user groups
Whole group discussion
The groups joined back together to discuss any issues that had emerged from the small group discussions.
It was raised that groups should work towards a mid-August deadline, but that times to reflect on whether this would be enough time should be set along the way.
It was questioned the extent to which Involve had capacity to manage across such a complex area, and where limits would need to be put in place to make it manageable.
It was stated that it need to be distinguished the organisations that must be involved, organisations that should be involved, and organisations that it would be good to be involved.
It was questioned what private sector representation there should be, including large data users such as Google, Facebook and Apple. It was suggested that engaging with the members of the Royal Statical Society would be a good proxy for reaching private sector organisations that manage data well.
It was stated that the process needed to clarify assumptions, and there needed to be a clear articulation of the main steps of the process. Linked to this, the intended outcomes and assumptions behind them would need to be clearly set out.
The question of how the discussion could be had outside of London and the southeast was raised, particularly given limited resources. It was highlighted that there is a significant amount of willing from local government to engage – so the cost of regional visits might be lower than thought.
Civil society wrap-up discussion
The meeting finished with civil society and government groups dividing again to have a wrap-up conversation.
There was some concern raised over the tailored public services proposals. Clarity was needed over how, why, and what for data would be used. Government needed to be transparent about where it wants to do things to people. There also needed to be some firm examples of where data sharing is necessary.
It was stated that the trend seemed to be that government wanted to share data about vulnerable people, which raised the question of whether this is paternalistic? And if not, is it going to apply to everyone else?
It was suggested that the problem definition is going to be important, as it would enable principles to be identified that tackle the issue but don’t become expansive.
It was suggested that this would lead to legislation being needed for different cases, but responded that other processes for ethical oversight could be put in place.
It was commented that there are good reasons to rationalise and to make transparent how data sharing takes place. It was suggested that because government would always seek to make more data sharing possible, the process is a good opportunity to ensure that there is a robust institutional and procedural framework that the government must follow.
It was highlighted that the current system is not very robust, with some gateways going through Parliament with no debate. There are many different possibilities of systems that could be put in place.
It was questioned whether the process could be future proofed for possible new governments, perhaps by involving think tanks as proxies for different political parties.
Finally, there was some concern that process could become unmanageable. It was stated that the network would be important in supporting Involve’s coordination role.