The Case for Data Cooperatives
Aapti Institute has commissioned this whitepaper to Open Data Manchester as part of a series that highlights learnings from stewardship in practice -showcasing the possible diversity in application across model type, use-case, and sector.
Julian Tait, Open Data Manchester
Data is playing an increasingly powerful role in our lives –through the services we use, the information we access, and how it is used to make decisions that affect us. The potential to control any of this can appear overwhelming, so it’s never been more important to increase our understanding of the power and value of the data we generate.
To do this, we need trustworthy organisations to help us take greater control of our data, allowing people to make real choices about how it is used, and to build trust and confidence in good data practices and the solutions derived from them. Such institutions or intermediaries, created to look after data on behalf of ‘data subjects’, have the potential to create fairer and more equitable uses of personal data, and the other data we create.
To become trusted intermediaries, charged with collecting, pooling, processing, and sharing our data, they must be underpinned by strong governance and fair decision-making mechanisms. In return, they give members control over how their data is used, and enable more and better data to be shared, providing greater value for people, places, and the planet.
Various ‘data stewardship’ models are emerging, including data trusts, with a fiduciary duty to their data subjects, as well as personal data management services, which have a contractual relationship with the individual. However, it is collectively controlled organisations, such as data cooperatives and data collectives, that may offer the best opportunity for data subjects to have the greatest control over how their data is collected, pooled, processed, and shared.
Solid historical foundations
Cooperatives are a well-established organisational structure –with more than one in 10 members of the world’s population being involved in one of the three million cooperatives in existence today.
These organisations operate meaningfully in many important areas of life, helping farmers, workers and producers act together, as well as facilitating financial relationships within credit unions and homeownership through housing cooperatives.
One of the oldest food coops in the United States, the Park Slope Food Coop, was formed in Brooklyn in 1973. It has 17,000 members collectively buying food wholesale and selling it with a small markup. Members give up a few hours a week to keep it running but it is not open to the public. The Gujarat Cooperative Milk Marketing Federation was formed in 1945 to combat the exploitation of India’s dairy farmers. It is jointly owned by 3.3 million producers and works with 18,600 village milk cooperatives, making an annual turnover of US$5.1 billion.
Whatever their purpose and whoever their members are, modern cooperatives share common foundational principles derived from those set out in 1844 by the Rochdale Society of Equitable Pioneers, considered among the earliest consumer cooperatives. They are:
- Voluntary and open membership–open to all persons able to use their services and willing to accept the responsibilities of membership, without gender, social, racial, political, or religious discrimination.
- Democratic member control–controlled by their members, who actively participate in setting their policies and making decisions.
- Member economic participation–Members contribute equitably to, and democratically control, the capital of their cooperative.
- Autonomy and independence–cooperatives are autonomous, self-help organisations controlled by their members.
- Education, training, and information–provide education and training for their members, elected representatives, managers, and employees so they can contribute effectively to the development of their cooperatives.
- Cooperation among cooperatives–working together through local, national, regional, and international structures.
- Wider society and community benefit–through sustainable development of their communities, through policies approved by their members.
These principles offer a useful foundation for data cooperative development, setting out how these new institutions will behave regarding their members and wider society.
Given these solid foundations, Open Data Manchester is an advocate for the development of organisations specifically designed to make cooperative principles work in the data-driven world.
In broad terms, we see data cooperatives as organisations that create an environment for members to pool their data so that it can be used to further their collective aspirations. By doing this, members have more control over data about themselves, they can maximise the value of the data they produce, and they can redistribute power through the collective value achieved.
To underline this, we accomplished pioneering research into data cooperatives, in 2013-14 and 2020-21, during which we identified some easy commonalities with the Rochdale Principles as well as several challenges, heightened by the centrality of data, that could not have been anticipated more than a century ago. These are:
- Governance–how data is managed through its whole lifecycle.
- Consent–the processes of consent needed to enable the cooperative to function.
- Sustainability–how the administrative, legal, and technological costs can be covered.
We already know that cooperatives from around collective challenges, or areas of common endeavour. Our work developing a data-cooperative model for small-energy cooperatives in 2020-21 found that members are more amenable to sharing data if it aligns with their organisational objectives and when data exploitation is seen as a means to achieving them.
Although many cooperatives have a purpose of returning economic value to their members, economic return for participation is only one type of value derived. Such ‘altruistic’ aims neatly map onto the Rochdale Principle of ensuring wider community benefit and are mirrored by established data cooperatives such as the Salus Cooperative and Healthbank. The cooperative is granted access to sensitive health data, and then allows access to it to relevant research organisations, so that the development of medical treatments is furthered.
Educate, train, inform
many, data is an abstract concept, with the impacts of its use, either good or bad, little understood outside of ‘data native’ circles. Therefore, a challenge for data cooperative development is raising awareness and understanding within the cooperative, so that informed decisions can be made around data sharing, with the impacts properly understood. Given that cooperatives are organisations with education, training, and information at their core, the problem can easily be addressed. Only when there is a common understanding of the power of data can it then be collected, used, and shared with confidence and responsibility.
Cooperatives have mature governance and decision-making processes because they have evolved over almost 200 years, yet, different cooperatives will have slightly different models depending on need and function. Data cooperatives, in developing governance structures for data processes, will need to build on and extend these for decisions about how data is collected and used.
With the data cooperative model for small-energy cooperatives that we developed in 2020-21 as part of the Open Data Institute’s Data Access Stimulus Fund, we identified five data flows that could shape the governance of such a cooperative:
- Member-to-cooperative –data is shared within the data cooperative for internal use, and is held and processed by the cooperative to improve service provision.
- Member-to-member (intra-cooperative) –data is shared between cooperative members, facilitated by the cooperative. This could be done through a member granting access to certain data, such as for benchmarking or for understanding the performance of a particular intervention.
- Federated –sharing could occur between organisationsor other cooperatives that have similar aims and data governance processes.
- Third party –traditional data sharing or licensing agreements with other organisations.
- Open data –creating a common, open asset available to everyone.
To ensure the value and utility of the data being managed by the cooperative, solid, foundational data governance processes need to be implemented so that any data being shared is fit-for-purpose and high-quality. Legal, as well as member-specified data protection requirements, must also be complied with.
Many of those who participated in the data cooperative model development were already participating in an energy cooperative that had a stated purpose, so they saw that better data use would help the cooperative and, by association, themselves, achieve that purpose. This leads to the question as to whether a data cooperative could be:
- Aseparate entity with aims that align with an existing cooperative, which then manages the data of that cooperative.
- A data management function within an existing cooperative.
- A cooperative with its own purpose –to collect, pool, process, and share data within the parameters set by the members.
Each of these will have a bearing on what a cooperative can and cannot do. This, in turn, will impact the routes to sustainability that the cooperative can take. Salus, Healthbank, and Driver’s Seat fit the latter model and demonstrate a need to operate at scale, or at least in an environment where the data collected return enough value to cover the costs of maintaining the cooperative.
As discussed, data, and its use, are often intangible and abstract, and managing consent regarding data can be an onerous task. It is a rare experience to visit a website and not have to decide whether to share your data or not.
So how do data cooperative members consent to data sharing? When we were developing the data cooperative model for small-energy cooperatives, members said they would freely share data if they could be confident that this sharing aligned with the cooperative’s aims. Yet, even though there may be alignment between the member and the cooperative, is this enough to allow the cooperative to manage and transact data efficiently, with a variety of data users, without referring back to each member?
A flat ‘worker type cooperative structure, where everyone has equal rights within the cooperative and all members vote on all possible data-sharing arrangements would quickly become impractical and unviable. So, creating a system that builds on the collectively agreed purpose but delegates decisions to a process that upholds that purpose, offers a way to reduce the burden on individual members and enables the cooperative to reach decisions regarding data use and sharing.
Working with the energy cooperative and other cooperative members, we looked at different mechanisms that would reduce the burden of data sharing decisions, and enable fast and efficient request processing. The basics of a ‘consent mechanism’
- Different conditions would be agreed upon regarding different kinds of data held, which could then be coded into the data set and, if met by a requesting organisation, would enable data exchange.
- These conditions could reflect individual preferences or those of the cooperative as a whole.
- The process essentially represents an automated checklist that generates different weighted scores, which are then fed into an algorithm that decides if data should be released to the requesting entity.
- The challenge with the system is that, although this was understandable, members involved with the research were hesitant about the automation and advised that the process would have to first be manual and then, if it behaved as expected, become automated.
- Once deployed, the system would also need to be subject to regular monitoring.
Even with an automated process, there would need to be agreed by members on what the assessment criteria and weightings would be. Yet, taking part in defining these criteria could be an attractive, trust-building function of the data cooperative. These criteria would need to be periodically reviewed as more members joined, and also when data needs within and outside the cooperative altered.
While data governance and consent are operational processes that are key to building trust and ensuring regulatory compliance, how this work is sustained, especially when data is often undervalued by those who create it, needs careful consideration. In terms of sustainability, data cooperatives can return direct economic value to their members through access to insights gained by enabling members to pool information.
Driver’s Seat Cooperative in the United States is one such cooperative –it empowers gig-worker drivers to make better decisions about where and when they work by sharing their own collected data to challenge the work allocation algorithms operated by the likes of Uber and Lyft. Giving workers access to the data that is used to manage and control their work practices, especially when many are considered free agents by both these companies and the government, has the potential to counter exploitative employment practices. Value is further returned by anonymised data being sold to city governments to support transport policy. This data has the bonus of being ethically supplied by Driver’s Seat members.
The Swiss Healthbank is the world’s first people-owned health data exchange that allows users to choose which research studies they participate in through sharing their data. The basic service is free and users are not required to be members but they can join the cooperative and receive a share for a 100 Swiss francs subscription fee. As of 2019, Healthbank had more than 200,000 users and had attracted 6.3 million Swiss francs in private-equity funding. This indicates a hybrid model –closer to a traditional company, limited by shares.
Sustaining these structures means people need to be involved –and incentives for this involvement, including and beyond economic value, remain a challenge. Even though data may be intangible and abstract, there is a cost burden regarding the creation and operation of data cooperatives. They are a legal structure with members and statutory obligations. Data has an infrastructural cost too, such as the creation of databases, related software development, and various data transformation functions.
There is also the human burden of the maintenance of the cooperative, ensuring the database is operating properly, that governance processes are adhered to, and general administration. Meeting these costs may be a challenge for all but the largest data cooperatives, which can aggregate and sell insights at scale. The founder of The Good Data Cooperative (now dissolved) stated that to become sustainable, they would need to attract 500,000 users. For comparison, Swiss Healthbank has 200,000 users.
Driver’s Seat Cooperative, with its 800 member drivers across the US, pools data and then sells insights from that data to cities –to help them with transportation strategy and the like. Local decision-makers are attracted by the fact that the data has been collected ethically, and increasing membership will generate more detailed insights for them, along with helping them move towards sustainability.
For small data cooperatives, sustainability may come from the development of subscription models, where members directly contribute to the operation of the cooperative, with the value being derived from the ability to maintain and exchange data through shared insights or new services created. There is also potential for data that is collected to bring value as a representational tool that forwards the aims of its individual members. Additionally, a data cooperative could act as a data management function of a host organisation, where the host supports the costs of the cooperative in return for better data access.
Data needs stewardship and data cooperatives must have a role in this –because they build on tried and tested means for people to come together to cooperate when faced with common challenges. While the above provides an idea of the issues related to their development, identified through several research projects, this is the beginning, rather than the end, of the story.
What’s needed next, as ever, is greater cooperation – bringing together those who work within such organisations, those who observe them, and those who benefit from their existence –to make it easier to develop some of the newest, oldest organising models in the world. From our research, we have outstanding questions about how to create and build trust in, data cooperatives –on issues such as the transparency and legibility of the system, particularly any automation involved, and of course, the time and financial costs of setting up and sustaining such a structure.
The issue of trust versus scale cannot be ignored for data cooperatives but it presents an interesting opportunity for further research: how can data cooperatives achieve trust at scale? And, conversely, how can small data cooperatives set up and sustain themselves, when the scale is not required?