The political economy of data intermediaries

The Ada Lovelace Institute’s forthcoming Rethinking data report calls for a comprehensive and transformative vision for data that addresses power imbalances and serves people and society. One of our guiding questions is: how can we prepare the ground for long-lasting and ambitious change?

The Rethinking data blog series brings forward discussions, insights and reflections on what new types of infrastructure, governance, institutions and regulation are needed to reshape the digital ecosystem.

The second blog in the series continues our exploration with a critical reflection on the political economy of data. Read the first multi-part contribution here.

We cannot rethink the institutions and intermediaries of data governance without critically engaging with the political economy of data.

Data has become a core form of capital, crucial for establishing and maintaining new methods of value extraction and market domination. ‘Just as we expect corporations to be profit-driven, we should now expect organisations to be data-driven; that is, the drive to accumulate data now propels new ways of doing business and governance.’

The rise of profit-driven data intermediaries

Companies that once identified as part of various sectors are now rebranding and reconfiguring their business models as they all converge on being ‘data firms.’ Executives at legacy industrial corporations that made their name as giants of the last Gilded Age – like GE, Siemens, Ford Motor, BP and Shell – not to mention the digital tech giants of this Gilded Age, have all declared that, as a former Amazon executive wrote, ‘The data is the business model.’¹

Monsanto – owned by pharmaceutical behemoth Bayer – is actively transforming ‘from an agricultural biotechnology company into a data-science-driven organization’. In addition to spending billions of dollars on acquiring tech start-ups with expertise in ‘precision agriculture’ and ‘smart farming’, Monsanto has also ‘retooled its employees and increased its data science team from 200 in 2017 to 500 people in 2020.’

Monsanto is exemplary but in no way an anomaly. Its transformation is part of a business strategy built around what geographer Alistair Fraser has termed ‘data grabbing,’ in the tradition of land grabs by rentiers. Data grabs occur through innovation and investment, with the aim of accumulating as much data as possible by controlling – which often means intermediating – systems of data creation, exploiting (infra)structural asymmetries in power and capital, and dispossessing data subjects of all, or most, value associated with that data.

There is a co-productive relationship between the political economy of, and the institutional intermediaries for, data governance and it is no coincidence that corporate gatekeepers are the most dominant type of data intermediary in contemporary society. This present arrangement is the product of a long-term relationship between monopoly capitalism and neoliberal states.

No political economic system – capitalist, communist, or anything in between – can exist without institutions giving it structure, enforcing its rules, supporting its operations and propping it up when crises hit. The form those institutions take is fundamentally shaped by the demands and the interests of the powers in the system, the values it prioritises and the goals it pursues. We should understand the rise of digital capitalism as another project of capitalist political economy, renewing itself again by creating new instruments and new institutions to keep the same imperatives of accumulation going.

As they conform to these changes, companies across virtually every sector are (re)configuring their business models to become intermediaries, maximising their extraction of the entire data lifecycle – from creation and storage, to analysis and use.

To understand how these data intermediaries have gained power – and how we might rein them in and build alternative models – we must focus on institutions and relational data, more than on individuals and personal data (see Salomé Viljoen’s work for an in-depth analysis of the theoretical and legal aspects of relational data).

Companies like Google, Meta Platforms or Amazon do not care about your data as a unique person; they only care about our data as an aggregated mass of relations and patterns. In recognition of that fact, corporate gatekeepers have invested in creating large-scale infrastructure for constructing and capturing data, and equally large institutions for managing and analysing it.

Crucially, responses to these data intermediaries must fight fire with fire by focusing on large-scale interventions into infrastructure and institutions. In practice, as I argue in this post, we need to put in place the right kind of institutions, such as public data trusts, and ensure that they operate in tandem with reformed data infrastructures that have not been captured by the interests of private intermediaries. These responses are key conditions to enact real change.

Emerging alternative models for data governance

The fundamental question for governance is whether the social value of data is commodified for private profit or collectivised for the public good.

There are now a number of emerging proposals for alternative data intermediaries that seek to move away from the presently dominant, profit-driven model and towards varying degrees of individual ownership, legal oversight or social stewardship of data (see studies here and here).

These proposals include relatively minor reforms to the status quo, such as legally requiring companies to act as ‘information fiduciaries’ and consider the interests of stakeholders, who are affected by the company, alongside the interests of its shareholders.

In a Harvard Law Review article, David Pozen and Lina Khan provide detailed arguments for why designating a company like Meta Platforms – Facebook at the time when the article was published – ‘a loyal caretaker for the personal data of millions’ does not actually pose a serious challenge to its underlying business model or corporate practices. In fact, such reforms may even entrench the company’s position atop the economy. ‘Facebook-as-fiduciary is no longer a public problem to be solved, potentially through radical reform. It is a nexus of sensitive private relationships to be managed, nurtured, and sustained [by the government].’

Attempts to tweak monopolistic platforms, without fundamentally restructuring the institutions and distributions of economic power, are unlikely to produce – and may even impede – the meaningful changes needed.

Other models take a more decentralised approach in the form of ‘data sharing pools’ and ‘data cooperatives that would create a vast new ecosystem of minor intermediaries for data subjects to choose from. As a different way of organising the data economy, this would be, in principle, a preferable democratic alternative to the extant arrangement.

However, in practical terms, this approach risks putting the cart before the horse by acting as if the political, economic and infrastructural support for these decentralised intermediaries already existed. Under present market conditions – where competition is low and capital is hoarded by a few – it seems much more likely that start-ups for democratic data governance would either fizzle/fail or be acquired/crushed.

The bias towards the large-scale centralisation of power and wealth in the hands of a few tech giants is no accident, but very much the product of government policies that allow them to grow unimpeded, enforce their ownership claims over concrete and digital assets, and furnish them with lucrative contracts for services ranging from health and welfare to military and immigration.

This centralization is so thorough that governments have outsourced data analytics and machine learning services, across such a wide breadth of agencies, to the same company, as in the case of Palantir. Thus reinforcing the idea that their worldview, methods, and technologies are universally applicable to solve all problems and govern all situations.

Our efforts to build alternative forms of data governance must focus on changing its political economic foundations. We should focus on advancing two related strategies for reform that would pave the way for a new ecosystem of decentralised intermediaries.

Infrastructural disintermediation and democratic data trusts

The first strategy is to disintermediate the digital economy by limiting private intermediaries’ ability to enclose the data lifecycle.

The digital economy is hyper-intermediated. We tend to think of the handful of massive monopolistic platforms that have installed themselves as necessary middlemen in production, circulation and consumption processes. But there is also an overabundance of smaller yet powerful companies that insert themselves into every technical, social and economic interactions to extract data and control access.

Disintermediation means investigating what kind of policy and regulatory tools can constrain and remove the vast majority of these intermediaries whose main purpose is to capture – often without creating – value.

For example, disintermediation would require clamping down on the expansive secondary market for data, such as the one for location data, which incentivises many companies to engage in the collection and storage of all data possible for the purpose of selling, sharing or servicing third-parties such as advertisers.

Even more fundamental reforms could target the rights of control and access that companies possess over data assets and networked devices, which are designed to shut out regulators and researchers, competitors and consumers from understanding, challenging and governing the power of intermediaries.

In a recent example, after many years of fighting against lobbying by tech companies, the US Federal Trade Commission has voted to enforce ‘right to repair’ rules that grant users the ability to fix and modify technologies like smartphones, home appliances and vehicles without going through repairs shops ‘authorized’ by the manufacturers. This represents a crucial transference of rights away from intermediaries and to the public.

The second strategy consists of the construction of new public institutions for the democratic governance of data.

In addition to pushing to expand the enforcement of data rights and privacy protections, efforts should be directed at policies for reforming government procurement practices and expanding public capacities for data governance.

The political and financial resources already exist to create and fund democratic data intermediaries. But they are directed at outsourcing government services to tech companies, propping them up with the state’s coffers, rather than insourcing the development of capacities through new and existing institutions.

In a recent article, Salomé Viljoen, Meredith Whittaker and I propose the formation of public data trusts that seek to turn data from private capital to collective resource and outline the institutional and legal steps necessary to achieving that goal.

In short, the type of public data trusts we advocate for would provide a powerful form of intermediary that not only reclaims resources and rights currently controlled by corporate platforms, but manages them in the public’s interests and for socially beneficial purposes. The legitimacy of public data trusts hinges on them being designed as institutions that are subjected to democratic oversight, accountability and representation.

This public model poses a crucial counterpoint to existing proposals that are based on privately managed trusts and individual data subject rights. Such approaches leave significant power in the hands of private interests – as in the case of the data trust proposed by Sidewalk Labs (a subsidiary of Google) for the urban regeneration project of Toronto’s Quayside area – to structure the relationship between data subjects and data users and may not provide as many protections for ordinary people. Indeed, they do not even touch on more fundamental issues such as how – and for what purposes – data is created in the first place.

To enact public data trusts, ‘a key enabler would be a legal framework mandating private companies to grant access to data of public interest to public actors under conditions specified in the law’. Such frameworks are being considered by the European Commission and German Parliament. ​​For example, Andrea Nahles, former leader of the Social Democratic Party in Germany, recently championed a ‘data-for-all’ law, as part of new antitrust regulations. To combat anti-competitive behaviour by big tech platforms, the law would require that the data collected by large companies become part of the public domain after a number of years.

Similarly, Hetan Shah suggested the possibility of allowing companies to use the data they gather only for a limited period (e.g. five years), after which it is reverted to a ‘national charitable corporation that provides access to certified researchers, who would both be held to account and be subject to scrutiny [to] ensure the data [is] used for the common good’.

These ideas will have to consider various issues, such as the need to ensure that individuals’ data is not released into the public domain and the fact that commercial competitors might not see any benefit in using ‘old’ data. Nevertheless, we should draw inspiration from these efforts and seek to expand their purview.

To that point, policies aimed at making data held by private companies into a common resource should go further than simply allowing other companies to access data and build their own for-profit products from it. To rein in the largely unaccountable power of data intermediaries who wield enormous, and often black-boxed, influence over people’s lives, these policies must grapple with fundamental issues related to who gets to determine how data is made, what it means, and why it is used.

Furthermore, the same policies should extend their target beyond monopolistic digital platforms. Data created and controlled by, for example, transportation services, energy utilities and credit-rating agencies ought also to be subjected to public scrutiny and democratic decisions about the most socially beneficial ways to use it or discard it.

Ultimately, we need strong legal and institutional interventions that can foundationally transform the existing arrangements of data control and value. Don’t think of public data trusts as an endpoint, but as the beginning of a new political economy of data – one that will allow and nurture the growth of more decentralised models of data stewardship.

Public data trusts would be well positioned to provide alternative decentralised forms of data intermediaries with the critical resources they need – e.g., digital infrastructure, expert managers, financial backing, regulatory protections and political support – to first be feasible and then flourish. Only then can we go beyond rethinking and begin rebuilding a political economy of data that works for everybody.