Data Intermediaries: An Urgent Need for Democratic Ownership and Governance
Over the last five years, opinions about the use of personal data have shifted dramatically. Concerns about data have grown throughout the public as a result of the work of scholars and activists, new data legislation, and documentaries like The Social Dilemma.
This period also saw the creation of the first data intermediaries, organizations aimed at developing a fairer data economy. Data intermediaries, also known as data collectives or data coalitions, are “collective bargaining entities that ordinary natural persons, data producers, [can] interpose between themselves and the businesses that collect or use their data.” Many of these organizations provide their members with the ability to monetize their data.
Due to the collective nature of data and to prevent the owners of the data intermediaries from taking advantage of their members, early proponents of these entities, such as Julian Tait, Trebor Scholz, Jaron Lanier, E. Glen Weyl, and Divya Siddarth, advised structuring them as cooperatives.
A worrying trend
Few entrepreneurs took the advice. Data intermediaries incorporated as private companies without democratic ownership and governance vastly outweigh those organized as cooperatives. Decentralized autonomous organizations (DAOs) that have chosen to integrate cooperative principles also represent a small minority of intermediaries.
In addition to outnumbering data cooperatives, non-democratic intermediaries have significantly larger memberships. Companies like Gener8 and Swash, which crowdsource browsing data, have 600k and 320k members, respectively. Others, like DIMO, which specializes in vehicle data, and Unbanx, which focuses on banking data, each have close to 10k members.
These intermediaries are experiencing rapid growth, fueled by venture capital. The additional funds allow them to spend large sums of money on acquiring new users. However, venture capital comes at a cost. Venture capitalists expect a return on their investment, which can only be realized if the intermediary is acquired or if it completes an initial public offering (IPO). Thus, increasing the incentives to prioritize profit over members interests.
Data as labor
To understand the issues that occur in the absence of democratic ownership and governance, it’s critical to treat data as labor rather than capital.
Data, like other forms of labour, is generated through human activities requiring physical or mental effort. Recording a workout on Strava, solving a reCAPTCHA to access a website, and using Google Search to find information are all examples of physical or mental activities that generate data. Even if these tasks are enjoyable, they nonetheless have the characteristics of labour. These examples demonstrate how data labour differs from traditional labour; it’s a form of work produced primarily while consuming digital services.
In today’s data economy, users barter their data contributions in exchange for the free services the technology giants provide. Instead of selling their services or products to their users, tech companies use the data generated by their users to improve the targeting of the advertisements presented to them. In this system, the customers are not the users but the companies paying for advertising. Data intermediaries offer an alternative to this model by enabling collectives of data producers to profit from their work. They also give their members greater bargaining power.
If data is labour then data intermediaries should treat their members as employees rather than users.
In organizations that lack democratic ownership and governance, owners and workers have irreconcilable interests. The former seek to maximize profits to survive in the market, which leads them to pay workers as little as possible for the most amount of work, whereas the latter seek to obtain the highest wage. In the end, employees’ interests are subordinated to those of the owners because workers have no influence over how the organization is managed.
This reasoning also holds true for data intermediaries. An intermediary organized as a traditional private company will inevitably exploit the people it tries to help. Owners will have to compensate members as little as possible for their data in order to generate profits. Members, on the other hand, will want the highest possible compensation for their data. Owners’ interests will take precedence over those of the members who have no control over the intermediary.
Owned and governed by the people producing data
It’s no coincidence that the first advocates for data intermediaries proposed structuring them as cooperatives. Cooperatives reconcile diverging interests by granting complete ownership and control of the organization to its members. It’s a simple feature that allows organizations to fulfill their mission without exploiting their members.
Entrepreneurs, it’s not too late to convert your intermediaries into cooperatives. And, future data intermediary founders, please consider the incentive structure built into the organization you’re forming. For inspiration, look to the following data cooperatives: Driver’s Seat, MIDATA, polypoly, Posmo, Salus, SAOS, Savvy, and Schluss.
Ultimately, data intermediaries handle private data, but this is only one aspect of the bigger picture. Public data from the internet, often referred to as the digital commons, needs our attention. The recent rise to prominence of generative foundation models, such as ChatGPT and DALL-E, which rely on extracting data from public sources, requires us to reconsider how we manage these common resources.