cooperative economics law theory

What is a Data Cooperative?

Democratic solutions for data stewardship

The writings on this publication aim to normalize personal data as a form of property, with the ultimate goal of returning agency to the individual via legal ownership. A significant barrier to this goal is the challenge of mediating data transactions between millions of individuals and the big data brokers that wish to do business with them, leading to a lopsided data market akin to the labor market. Following this logic, it makes sense to apply the cooperative model that many workers, farmers, and small business owners have used worldwide — but for data.

This essay is predated by “Democratic Models for Ethical Data Stewardship” (Mendonça et al.), which provides more robust definitions for those interested.

Stewardship vs Ownership

The basic moral claim I wish to make across this project is that the individual has a claim over the data that is derived from them, that it should be recognized legally, and that a framework needs to exist to facilitate this. Under this view, the individual must maintain traditional ownership of their data, which implies exclusive control and the right to refuse access by others. Collective ownership, where a group pools its data and shares legal ownership over the entire pool, violates this protocol by adding friction to the process of opt-out. Collective stewardship, on the other hand, defers only the maintenance, governance, and facilitation of data to the collective while retaining individuals’ claims to ownership.

Mendonça provides a good roadmap by insisting on the clear delineation between data cooperatives and data unions. They define a data union as a legal entity established to pool data for stronger negotiations, without necessarily creating a statute of ownership within the pool. In practice, this would mean a member contributes their data to the pool under the promise that it will be fairly and properly managed by the union organizer.

Cooperatives vs Unions

Data cooperatives, in contrast, maintain a clear statute of ownership via a democratic voting bloc where members collectively determine what to do with their individually held data. This approach is more aligned with the ethos of Data as Property, though the facilitation of usage and enforcement of internal policies is less implicit and must be actively handled.

Mendonça also briefly describes two ideas that are less aligned with Data as Property, but are worth mentioning to more clearly define what we want out of a cooperative.

Data trusts, such as the Mayo Clinic’s patient data trust, are a way to ensure data is handled with care and proper governance by a trusted party. This method is the easiest on the part of the members; however, it has the least — indeed, no — democratic control, foregoing ownership for security and accountability. In contrast, a data cooperative’s purpose is to empower its participants with agency over their data, in exchange for an increase in personal responsibility.

Data commons, on the other hand, are somewhat antithetical to the idea of individual rights towards data: a data commons is similar to a data union concerning communal access and control; however, they also impose the idea that data is a shared resource. Data as Property takes the opposite stance — that data belongs to the individual — on the premise that the current attitude towards data has treated it largely as a common resource, and has resulted in various legal, economic, and moral harms against society.

What is a Data Cooperative?

Following these definitions, a data cooperative can then be defined as a voting body that facilitates the governance, operations, and compensation matters surrounding the use of member data. The guiding principles of a cooperative, established by the International Cooperatives Association and based on the Rochdale Principles of the 19th century, relate to individual ownership of data via:

  1. Voluntary Membership: The right to be deleted, a commonly held legislative right similar to the right to privacy, requires that members can opt out at will.
  2. Democratic Member Control: To maintain the individual claim over one’s data, democratic control must be maintained.
  3. Member Economic Participation: Individual claims over data also require fair and proportional economic incentives.
  4. Autonomy and Independence: This is the core argument behind Data as Property.
  5. Education, Training, and Information: One of the primary harms resulting from the current legal position around data is that there is an incentive against educating individuals about how their data is used. As such, education is a necessary component of trust and transparency.
  6. Cooperation Amongst Cooperatives: While different cooperatives may align on different goals, there is a technological incentive for cooperatives to share information: if cooperatives share a similar or identical governance or information structure, their barriers to adoption, sales, and recognition are drastically reduced.
  7. Concern for Community: Because data derives part of its value from its relationship between individuals, it is in everyone’s best interest that their community’s data is also treated fairly and with respect.

Governance

A data cooperative must set, manage, and enforce the handling of member data. This is both internal and external — governance relates to the voting structure, membership criteria, and personal and collective bargaining requirements when it comes time to negotiate. Roughly, a data cooperative needs to determine:

  • Negotiating structure
  • Voting model
  • Data policy (what can buyers do with our data?)
  • Participation rules (e.g., can a member opt out of one negotiation and maintain membership?)

Operations

A data cooperative must also manage the practical operations of data stewardship, such as:

  • Retaining lawyers for damages
  • Facilitating external functions (transactions, audits, etc.)
  • Determining technical operations
  • Carrying out internal functions (voting, policy enforcement, etc.)

The specifics of each would rely on the governance and economic requirements of the collective.

Economic

Lastly, a data cooperative needs to manage the economic aspects of its model. This entails both the pricing and negotiations for external use, and also the design of economic incentives for member participation to begin with. It is unclear as of yet if there is a preferable model, or if there even should be a preferable model, but examples include direct participation (1Kb = 1 cent, for example), weighted participation (medical data > retail history), and cumulative weight (compensation based on tenure and overall participation).

The economic structure of a cooperative is the most important aspect, as membership gain is both nominally dependent on payout, and directly contributes to bargaining power. At a minimum, members should feel that participation is always economically preferable to non-participation. To avoid the case of moral injury where destitute citizens feel forced to sell intimate knowledge of themselves, the economic reward of participating should be balanced with the moral reward of self-determination regarding opting in or out of a decision; something like a minimal incentive may work to this end.

Technological Framework

The last piece — which I exclude from the above three pillars, as it is not so much a guiding principle — is the technological framework used to facilitate the cooperative. The challenge with data cooperatives versus normal workers’ co-ops is the technical knowledge required to manage the data. In a traditional co-op, each worker knows exactly what the utility, operation, and storage requirements are for whatever they are contributing; this cannot easily be said today for user data. Moreover, due to its digital nature, the scale of these features requires specific infrastructure and treatment.

While any centralized platform could be developed, a blockchain infrastructure may be a worthy candidate. Voting records, governance logs, and transaction logs are all decentralized and tamper-evident, facilitating individual agency within a large collective while being transparent and stable. However, this must be treated as infrastructure and not the incentive structure itself, meaning the blockchain should be divorced from having a nominal value — that is, an associated cryptocurrency. Attaching an asset to the participation of the cooperative would introduce volatility and misaligned incentives and open the cooperative up to abusive behavior.

Conclusion

This essay describes what a data cooperative is meant to be: a member-owned, democratically governed structure for stewarding personal data while preserving individual ownership, consent, and opt-out rights. A well-defined data cooperative manages governance, operations, and incentives so members can collectively negotiate the use of their data without surrendering personal control. Technological progress is making a data cooperative more feasible than it was even a few years ago, with better tools for secure online voting and tamper-evident governance; these developments make the cooperative model a practical design for self-determination over personal data.