[Spotlight] New types of data institutions
Generating value through data trusts
by Implementation Working Group
Every month we host informal exchanges with a group of passionate open data practitioners to discuss challenges and opportunities to do open data well. This year we kick off the calls welcoming our new co-chair, Florencia Serale, an open data and digital government expert on Latin America, working at the IADB Digital Cluster in the Innovation for Citizen Services Division. She joins Paul Stone, Open Data Lead from the Government of New Zealand, to help deliver an exciting new action plan for the group.
We were joined by guest speaker, Aimee Whitcroft, board member of the Open Data Charter currently working at the New Zealand Transport Agency to talk about ideas on Data Trusts and Trusting Openness.
As a relatively new concept that is receiving a lot of attention, it was helpful to start with the data trusts definition from the Open Data Institute — “a legal structure that provides independent stewardship of data” (ODI, 2019). This means not seeing data as a property, but rather as the right to control how my own data is being used by a third party (being government, business or whoever). Essentially, it’s about giving control and agency back to those who originated the data, by establishing a legal duty for trustees to manage data on behalf of others.
Yet terms to describe different types of data institutions to increase access to data are still complex, so we focused on practical examples:
- Medical data. A data trust here could be very beneficial, where the person allows their own data to be used, for instance, for a specific kind of purpose to advancing research, but it does not give permission to share that same data with another person/institution/business.
- Artificial Intelligence- One of the reasons that the AI industry is not so intelligent is lack of access to data. Data trusts could potentially be a way for AI systems to both access quality data and be accountable for how they are used.
- Digital identity: as the debate heats up over who should control our digital identities and how, data trusts could provide an alternative to some of the current options by giving individuals control over the components of their digital identity for different uses, without requiring them to try manage and control every aspect, every time (a nigh-on impossible task even for the digitally highly literate). Read more about it in Aimee’s previous blog.
- Indigenous data. Several indigenous groups are talking about “Indigenous Data Sovereignty”, as “the right of Indigenous Peoples and nations to govern the collection, ownership and application of their own data”. More information can be found on our previous blog about the CARE Principles recently launched.
- Open Libraries. Librarians have a deep understanding of data. There are some examples from Boston (where the government partnered with its public library to revamp the city’s open data program) and Toronto (where the Toronto Region Board of Trade called for Toronto Public Library to oversee Sidewalk Labs’ data).
- Agriculture. With controversy in the US farm community about farmers’ ownership of their data and its uses by companies that are selling services. Two proposals for the Smart Cities Challenge in Canada referenced a type of data trust for agriculture, including the creation of a Data Utility in Guelph, Ontario and four counties in Alberta.
- Data Ventures is a data brokerage idea that has matured into a form of data trust, whereby they take sensitive (private) data from mobile phone operators and other sources and transform the data into safe, useful data products. The first product being Product Density, which gives the population of a suburb at any given hour.
- Data for Children, and collaborative working to bring about better outcomes for children around the world: https://www.dataforchildrencollaborative.com/
- Financial data. This paper in the UK recommends that anonymised personal banking and financial services data should be held in new public data trusts
With few models being tested in practice, there remains a series of challenges for the design of institutional and legal arrangements that increase access to data while retaining trust .
How would people know these data trusts exist? Where would they go to look for them? Some participants argued that if the community is engaged at the point of creation of the data trust (which should ideally be the case), then discoverability may not be a huge issue. It depends on the governance model that is set up — you could envision a data trust where the governance consists of one-third community representation, for instance.
In the case where we see communities actively engaged in the design and creation of their data trust, digital literacy would need to be addressed in advance . How do we make sure people meaningfully understand and control what kind of power they are placing in the hands of others?
Sharing and privacy
How do we ensure that data is securely transferred, in particular sensitive and personal information? How can we share cross-border data? Another grey area is the fact that multinational corporations like Google & Facebook operate horizontally across nations, while the discussed models of data trust are envisioned to operate vertically within a city or country. So this might create a challenge for them in fulfilling their desired role, especially as they navigate changes in legal data protection and access frameworks.Someone raised the GDPR case, as a possible “international open data trust”, which reaches many countries from Europe, so this example may show how these corporations may actually reach their aim.
Statistical agencies around the world are exploring how to use administrative data to replace expensive surveys to complete a population census, and there are concerns around privacy. Can data trust model help to maintain trust and social licence here?
What’s needed to trust Data Trusts?
Principles of openness in how new or existing entities undertake a stewardship role over collecting, maintaining, and providing appropriate access to data will be essential to fostering trust., As organisations are thinking about forming data trusts, we believe it is important to think about the entire spectrum of the data value chain, including decision on what should be made available as open data, to enable collective solutions.
Join us at the Data Justice 2020 where we will be continuing to explore this issue. If you are interested in sharing your examples or joining this event, please contact us at firstname.lastname@example.org!