Framing data justice: A global perspective.
A s datafication has gathered pace and evidence of data harm has multiplied, scholars have called for the need to identify and conceptualise issues of data justice. But what does it mean to talk about justice in the context of datafication? How do we ensure data rights and prevent data harm? Are these issues that affect all citizens equally, or are some social groups impacted more than others? Are the solutions to data justice issues–such as data discrimination and bias–technical, regulatory, political or all of the above.
There are many attempts to conceptualise the issue of data justice today and, as wave after wave of technological innovation rolls out, what data justice is and how we might achieve it is far from settled. In this post, I introduce Linnet Taylor’s three pillars of data justice. The next will explore a framework developed in the field of international development studies by Richard Heeks and Satyarupa Shekhar. Finally, I will discuss the social justice perspective of the Data Justice Lab.
Linnet Taylor from the Tilburg Institute of Law has written an oft-cited article on the need for an international data justice framework in which she represents diagrammatically three pillars of data justice (see diagram below). She advocates for a global perspective on data justice because of the exponential adoption of technology globally and the border-crossing nature of data capture and analysis.
Taylor argues that governmental and corporate actors have scrambled quickly to capture citizen and customer data to create new data regimes of intervention and influence, but that alternative perspectives “on the ways in which datafication can serve citizenship, freedom and social justice are minimal” (Taylor, 2017, p.2). Just as the idea of justice is needed to establish the rule of law, Taylor argued that “an idea of data justice is necessary to determine ethical paths through a datafying world” (p.2).
There are three clear reasons for the application of a social justice lens on datafication:
- firstly, the harms of accelerating datafication fall disproportionately on those with low socioeconomic status
- secondly, other forms of social group membership — for example, gender, ethnicity, place of origin and age — intersect with socioeconomic status to further increase the risk of surveillance and data harm (Taylor, 2017)
For example, a teenager from an immigrant family, living in a low-income area, whose parents are poor and who belongs to a minority ethnic group and religion is exponentially more likely to be targeted for surveillance by both protective (social services) and preventive (law enforcement) authorities, and is also likely to have less opportunity to resist that surveillance or intervention than her friend who lives in a high-income area and belongs to the majority ethnic group. (Taylor, 2017, p. 3)
- Finally, while the idea of data discrimination and its disproportionate impact on the already disadvantaged is not new, some aspects of contemporary datafication are new.
For the surveilled teenager in the above example, the problem multiplies when the functions of data collection and analysis are shared between public authorities and the commercial firms that provided her phone, her internet access or the apps she uses.(Taylor, 2017, p. 3)
Data justice problems multiply when we consider the blurring of the interface between public and private data collection, since much “of what we perceive as public-sector functions (counting, categorising and serving our needs as citizens) are in fact performed by the private sector, with corresponding implications for transparency and accountability” (Taylor, 2017, p.3). This complicated ecology of datafication:
suggests that markets are a central factor in establishing and amplifying power asymmetries to do with digital data, and that new strategies and framings may be needed that can address the public–private interface as an important site for determining whether data technologies serve us or control us. (Taylor, 2017, p.3)
This blurring of the boundaries between public and private sector actors, in combination with the imperceptible and global nature of large-scale data collection, can make accountability diffuse, data misuse harder to detect, and the usual legal remedies harder to obtain. Indeed, Taylor argues that the default individual rights-based remedies may no longer be fit for purpose in the global data market for two reasons:
- This liberal individual human rights framing usually requires that harms are transparent and visible so those harmed can make claims, and
- The standard assumption is that redress will be sought at an individual level.
Yet neither of these conditions obtain when there is a lack of transparency and accountability around data collection and use or when the harms fall on individuals as members of social groups.
Data injustice
Taylor goes on to illustrate the global nature of data-driven discrimination using two cases: India’s biometric population database, the world’s largest, known as Aadhar; and a proposal to use a machine learning algorithm to predict how migrants might cross the Mediterranean to enter Europe.
In the case of Aadhar the system was designed to require identity verification through iris and fingerprint scanners. This design choice is ill-suited to the realities of India’s impoverished masses. For example, the system fails to authenticate workers whose work with stone and other rough materials has erased their fingerprints. Nor can it identify people over 60 whose irises have been damaged by malnutrition. As Taylor (2017) points out, “These standards point to a middle-class standard for normality rather than the precarity and unpredictability of the lives of the poor.” (p.5).
In the case of the Mediterranean migrant monitoring system Taylor highlights how its proposed design might preemptively prevent migration and remove the right in international law–expressed in Article 14 of the 1948 Universal Declaration of Human Rights–that “everyone has the right to seek and to enjoy in other countries asylum from persecution”.
The Three Pillars of Data Justice
Taylor recognises and reviews several alternative ways of framing data justice and argues that the three pillars perspective integrates these into a single overarching conceptualisation focusing on three core values for international data justice: (in)visibility, (dis)engagement and non-discrimination.
The three pillars framework is a high-level roadmap designed to help key actors deliberate about data justice around core values. Although the overarching framework is intended to be global, Taylor anticipates that “each legal and social system would work out for itself how the principles of data justice applied” (p.11). The three pillars framework signals where choices need to be made. Still, Taylor acknowledges that there are many dilemmas associated with its operationalisation in different national and international contexts:
…the main challenge in building out this conceptualisation will be in finding how these overarching principles can gain traction in different contexts: some countries or groups will identify benefits of surveillance while others will strongly react against it as oppressive. Some will assert that private sector innovation plays a central role in realising the benefits of data science while others will claim that making the public sector more responsible for controlling data will achieve more just results. (p. 12)
Much of Taylor’s paper seems to assume a deliberative democracy (although in her video she refers to dynamic governance, also know as sociocracy) where citizens and states can use the framework to inform calm and considered debate about the issues before reaching a consensus. However, this approach underplays the role of conflict and the grossly unequal power dynamics associated with the vested interests of social elites, Big Tech and governmental actors. Perhaps acknowledging this issue, she concludes by stating:
Innovation and evolution in technology are constant and desirable, but the ways in which technologies are used to monitor and govern us are negotiable. We should be able to determine our interactions with technology by debating and, if necessary, resisting and proposing different paths. If we cannot imagine how to regain the kind of privacy we would like, how to allow people to opt out of being surveilled through their data — or even of producing those data in the first place — we may have to reinvent as well as renegotiate.(p12)
