The Data Governance Deep-Dives held by Aapti institute and Digital Commonersaim to build communities of concern around alternative data governance and models for data stewardship. Every month, we host informal, closed-door conversations with practitioners, experts, and researchers to explore how these models can be brought to life. Our fifth session was structured as a ‘Community meet’ and featured presentations by three organizations.

1. Designing collaborative data governance structures: what to consider

The first presentation featured an organization that is exploring open data sharing in the context of biomedical research. The methodology they adopted to uncover flexible governance practices included observations from participant-led research. The goal of the study was to design inclusive governance structures to manage mental health databases of global youth across three countries.

It engaged two sets of stakeholders, youth and researchers, who were chosen through panels constituted in participating countries. Based on observations, the study offered preliminary insights on guiding questions for choosing appropriate governance designs:

• Access -who can access the data and what do they have to do to access it?
• Hosting -where is the data hosted?
• Controls -who controls the data?
• Costs -what are the expenses involved in managing the data?
• Visualisation -how can people see the data?
• Usage -how is the data employed/used by researchers?

Through their analysis of data sharing in medical research, they deduced that there is no ‘one-size-fits-all’ solution for governance. Hence, they recommended flexible data governance models which were created from observed research collaborations (‘governance structures’) and reused patterns of data governance systems (‘governance designing patterns’) with demonstrated success. Their aim is to:

• help emerging organizations to create legal governance structures to unlock the value of data in a responsible and safe manner while respecting the autonomy of data subjects;
• incorporate flexibility within governance structures to cater to the varied needs of data collaborators
• promote human-centric design as essential attributes within different approaches to data governance by including personas and scenarios created from interviews and observations

The choice of specific governance structures depends on the following attributes among parties collaborating in the data-sharing model:

• the degree of data available ( potential size of the user base for data)
• the degree of freedom to use such data (constraints such as purpose limitations and end-use restrictions placed on data processing entities)

They also provided successful illustrative models for data sharing deployed during the pandemic, created on the basis of observed governance structures and design patterns suggested in this paper.

Participants had divergent views on how data could be accessed by the public. The youth panelists favoured open viewership on the server or synthetically recreated datasets as a suitable means of access for the public. The researchers, on the other hand, preferred downloading of data. Similarly, on the issue of control over data, the youth preferred a community-controlled model as opposed to the researchers, who preferred a managerial model of control over data.

Through the discussion, attendees raised questions on how the youth and researchers were engaged in the experiment. The organisers explained that the youth and research advisory boards were created in different nations. They met twice a week and played the role of co-researchers to the traditionally trained researchers. Additionally, attendees were curious to understand whether the participating panels shared a common understanding of the term ‘data’, given general ambiguity around the term and how it affected ideas of “ownership” and “control”. This as all participants were informed of the data being specifically mental health data of youth collected through an application.

2.Empowering indigenous communities with control over administrative data

The problem many indigenous communities face around data governance is two-fold: first, data is collected and defined by a set of outdated and colonial metrics; second, existing indigenous ways of knowledge and labelling do not easily translate onto data management systems.

Recognizing these challenges, the second participant organization developed a prototype that supports indigenous communities by giving them ownership and control over their own data which is stored in public databases. The app generator codes the application for the community in the background of the app webpage, permitting communities that are not always technically savvy to access their own database with ease. It enables standardisation of datasets, granting communities greater autonomy over their data. The prototype sources data from different databases, and through its interface enables the interoperability of different data sets. This gives communities the flexibility to create their own data by matching data from different points.

However, going forward, it would be interesting to test this prototype within particular communities in a manner that is accessible to these communities. Another challenge would be to ensure access and interoperability of those datasets which are not openly available. This would require building customisable APIs. This also requires addressing questions around community consent and how consent can be sought in an ethical manner. Given the flexibility of the app, this would depend on the controller and community operating it for their benefit.

3.Data Coalitions to collectively bargain for rights of data providers

The third organisation considered was a community movement concerned with democratising the data economy. Through a multi-stakeholder brainstorming process involving artists, academics,and entrepreneurs, the organisation seeks to create a legislative tool for data providers, to collectively bargain for their rights on data sharing.

To address the market failures inherent in the present digital economy, the organisation seeks to build coordinated ‘data coalitions’ within the legal systems to create obligations on data buyers to be responsible to their data subjects. Collective bargaining seeks to prevent monopolistic hold over data by a single entity.

A few of the challenges mentioned were:

• Embedding ‘data coalitions’ within legal systems/frameworks in a manner that ensures that data buyers cannot bypass these intermediaries.
• Ensuring better coordination between the different data coalitions. This requires the presence of frameworks that can adjudicate disputes between coalitions.

In closing, the question that remains is would turning intermediaries into a public good be the right way to walk towards achieving equitable data economization? How can data governance structures and global regulatory frameworks assist intermediaries in balancing the rights of all stakeholders in the ecosystem to unlock the social value of data? If intermediaries are founded in law but not publicly provided, would that imply people with better resources receive greater data protection?

Overall, the session highlighted the complexities of data decisions relating to governance mechanisms, design choices, and collaborations. The experience of these organisations holds useful insights for implementing data stewardship.