1.1 Introduction
We are writing this introduction while watching the horrific suffering in Gaza unfold. More than seventy-five years into the ongoing Nakba,Footnote 2 many aspects of its intensifying violence are being exposed and contested through the traditional channels of international law. The UN Secretary-General and UN General Assembly called (unsuccessfully) for an immediate humanitarian ceasefire. International organisations and expert bodies that have criticised Israel’s besieging, mass bombing, starvation, forced displacement of Palestinians as well as the incitement of plausible genocidal intent towards them include the UN Human Rights Council, the International Committee of the Red Cross (ICRC) and United Nations Office for the Coordination of Humanitarian Affairs in the occupied Palestinian territory (OCHA oPt), as well as numerous UN Special Rapporteurs and NGOs such as Médecins Sans Frontières and the UK- and Canada-based Independent Jewish Voices. The International Criminal Court (ICC) announced its investigation into potential international crimes committed during both the Hamas attacks on 7 October and the humanitarian catastrophe that followed. Resolutions were proposed (and vetoed) at the UN Security Council and adopted by the UN General Assembly. In what is already perceived as a landmark case, South Africa turned to the International Court of Justice (ICJ), which qualified Israel’s actions as plausibly, prima facie genocidal in the earliest jurisdictional phase and imposed provisional measures. Weeks later, the ICJ again served as the stage for Palestinian struggle in the oral proceedings on the UN General Assembly’s request for an advisory opinion on the status of the ongoing occupation of Gaza and the West Bank. The exposition and evaluation of the carnage have thus been channelled through the routes of inter-state diplomacy, institutional action and dispute resolution familiar to the field of international law.
These familiar routes lead to important sites of public authority and raise salient questions on the legal characterisation and consequences of violence. Yet, there are also important aspects to this violence that lie beyond these conventional lines of sight and argumentation. In sustaining the incessant bombing of Gaza, an Intelligence Unit of the Israel Defence Forces (IDF) reportedly deployed an artificial intelligence (AI) system called ‘Habsora’ (or ‘The Gospel’) – a data-driven ‘factory’ for the production and listing of buildings, infrastructure and ‘power targets’ (such as urban residential tower blocks and government offices) to strike ‘at a fast pace’.Footnote 3 Another AI-based system called ‘Lavender’ has reportedly been used to generate kill lists of suspected ‘junior operatives’ of Hamas and Palestinian Islamic Jihad (PIJ) by analysing information collected on Gaza residents and automatically assigning them a score purporting to express ‘how likely it is that they are a militant’ based on algorithmically detected ‘incriminating features’ and suspected Hamas/PIJ affiliations.Footnote 4 By loosening the legal criteria on collateral damage in military targeting and also lowering the algorithmic threshold used by the Lavender AI model for determining what an ‘operative’ is, tens of thousands of Palestinians have been reportedly listed for killing in this way. A geolocational tracking system called ‘Where’s Daddy?’ has also reportedly been used to follow listed people in real-time and send an automatic alert once they enter their homes, so the building can be bombed – often whilst their family members are there. One source familiar with the system called it ‘broad hunting’: ‘You put hundreds [of targets] into the system and wait to see who you can kill … You copy-paste from the lists that the target system produces’.Footnote 5
This war, like the more recent war in Iran, has been a testing ground for an array of AI-enabled technologies – from the use of novel automated target-generation systems to AI-enabled drones and optical sights.Footnote 6 Yet the associational ontology of systems like Lavender builds on and resonates with the increasing use of algorithmic systems – for example, in counterterrorism and border security – to identify and anticipate ‘unknown’ security threats based on the distillation of emergent patterns from large amounts of disaggregated data.Footnote 7 The use of AI technologies in military targeting invites forms of decision-making and action that are troubling the foundational legal principles of precaution and responsibility and prompting new normative standards to try and mitigate their ‘unintended behavior[s]’.Footnote 8 They are reconfiguring human-machinic agency in military targeting, altering the calculus of conventional warfare and eroding opportunities for internal or external scrutiny in ways that the existing regulatory repertoire of international law fails to capture and that are actively reshaping practices of international law and global governance.
While AI is redrawing and extending the lines of enmity in Israel’s violent military campaign, powerful online platforms are using AI-driven systems and algorithmic content moderation infrastructures to suppress and rechannel the public outrage it sparks. ‘Meta’s policies and practices’, Human Rights Watch (HRW) reported, ‘have been silencing voices in support of Palestine and Palestinian human rights on Instagram and Facebook in a wave of heightened censorship of social media’.Footnote 9 According to HRW, this is enabled by Meta’s undue deference to Israel’s Cyber Unit and ‘[h]eavy reliance on automated tools for content removal’.Footnote 10 This builds on earlier asymmetries in the algorithmic moderation of Israeli and Palestinian online content – where Arabic language content classifiers, platform content policies and Terms of Service (ToS) linked to the enforcement of international terrorism lists and the moderation practices of global chains of human reviewers have been assembled and optimised in ways that disproportionately target Palestinian expression as de facto ‘dangerous’.Footnote 11 This mode of algorithmic regulation reveals dynamics of power and profit derived from the content moderation practices of online platforms and the novel global security infrastructures and risk governance configurations that these complex public–private assemblages are enabling.Footnote 12
At the same time, new digital and forensic instruments are being deployed to map, monitor, visualise and publicly contest the lethal violence of this siege.Footnote 13 These emergent forms of forensic activism entail different methods of truth-telling and evidence gathering that intersect and conflict with the conventional narrative styles and evidentiary standards of international adjudication.Footnote 14 The use of algorithmic tools is altering and amplifying the powers of public and private actors – by restructuring the flows of economic, political and social life while widening the scope of surveillance and suspicion. But it is also inspiring alternative modes of resistance and refusal. This signals the emergence of new modes of global ordering and disordering to which current forms of international legal thought and practice are insufficiently attuned – modes of ordering and disordering through data that this volume explores.
1.2 Rule by Entanglement
This volume is interested in how power and authority are asserted and performed in and around these novel routes, sites and scales through the media of digital data. The titular references to global governance ‘by data’ and the advent of ‘algorithmic’ rule reflect our proposition that important contemporary changes in governance are sparked or enabled by the use of advanced digital technologies. This includes technologies for the capture, generation and collection of data as well as algorithmic modes of data analysis, such as those employing AI and machine learning (ML).Footnote 15 It also includes forms of automated decision-making (ADM) and the data-driven assemblages of human-machinic governance resulting from the use of these devices. These changes and associated challenges tend to be framed in multiple, often opposing, ways: optimism about the rationalisation of public governance by technological means co-exists with anxieties about the displacement of modern law and politics by algorithmic governmentality and AI.Footnote 16 Claims about the structural homologies between code and law (‘code is law’) co-exist with arguments about law’s purportedly distinct character as a slow, deliberated, textual, interpretative practice and guardian of human agency and autonomy.Footnote 17 Bracketing these ideal-types, this book instead evokes and empirically confronts more messy and emergent realities where (international) legal practices and digital systems are enmeshed and assembled together through new data infrastructures, sociotechnical relations and forms of data-driven governance.
This volume traces these entanglements of international law with digital technologies by exploring how global problems are enacted, reconfigured and governed through the diverse sociotechnical arrangements and relations that data enable. Our starting point is that data is not a neutral or pre-analytic representation of the external world through numbers or bits of information. Raw data, as many chapters underline, is an oxymoron.Footnote 18 Rather, data alludes to a range of practices of giving and receiving (inputting, processing, abstracting, aggregating and so on)Footnote 19 among human and non-human actors that help bring worlds into being or into focus, while causing other worlds to unravel.Footnote 20 This idea of data actively enacting and shaping relations runs against the grain of much existing legal scholarship and policy concerned with digital technology. In data protection law, for example, data is often posited as an abstract and immaterial phenomenon, with norms dictating how it should be ‘collected’, ‘processed’ or ‘shared’ to ensure legal compliance. Emerging regulatory frameworks for digital innovation aim at unlocking the ‘value of data’ by better controlling conditions for data access and sharing and by promoting interoperability between data-driven systems.Footnote 21 In both variants, data is conceptualised as an extra-legal target of regulation to be captured, harnessed or governed by law. Yet these approaches miss what data does as an active force in global governance: the ways it recomposes and shapes new subjectivities, forms of material agency and social relations; draws people, places and things together into novel relations to be valorised, known and ruled through calculative and representational techniques, and the ways it reconfigures institutional routines and legal and regulatory practices.
In this book we suggest that socio-material relations generated through data are changing how international law and global governance are practised, how power and rights are exercised and how notions of legal agency and jurisdiction are sustained. Across the book’s four intersecting sections – Infrastructure, Imaginaries and Agency; Rationalities, Styles and Logics; Data Ecologies and Jurisdictions; and Violence, Accountability, Critique – contributors from law, geography, politics, computer science and media and data studies unpack and highlight what is at stake – conceptually, methodologically and politically – in these data-driven changes to global life and rule. Data practices associated with automated and AI-informed decision-making are key drivers of legal transformation in the present, yet remain poorly understood and conceptualised within international law, global law and global governance scholarship. As the chapters in this book underscore, the practices that data encapsulates are not merely objects or outcomes of legal control. They are also juridical in the ways they materialise, format and reconfigure legal and governance practices in specific and variegated ways. With insights generated from Large Language Models (LLMs), ML and other advanced ADM systems becoming increasingly key to how global problems are known and governed, we contend that analysing the dynamics of emergent data assemblages and algorithmic infrastructures is crucial for grappling with the normative, epistemic and political problems and inequalities that data are enabling.
In mapping these dynamics, this volume brings insights from critical data studies, Science and Technology Studies (STS), feminist theory, critical black studies and political economy into dialogue with international law and global governance debates about the transformative effects of AI and ADM. Social science scholarship highlighting the agency of matter (i.e., the operative, manifest effects of physical substances and artefacts and their attendant practices) in compositions of political and legal power provides one important entry point for our collective investigation.Footnote 22 Empirical studies analysing data and digital technologies through their situated practices and material-semiotic effects is another. While we explore different approaches to how data enables and mediates global governance, these are tied together by a shared attempt to think relationally about how the juridical and digital as well as the human and non-human elements of data assemblages are made and sustained in practice, beyond framings of dichotomy or displacement. Our aim in doing so is twofold. First, we seek to trace how emergent forms of global law and governance are co-produced by and entangled with specific sociotechnical arrangements. Second, we aim to explore new ways of conceptualising and engaging with these sociotechnical assemblages and forms of legal ordering that can guide legal action and provide critical analytical and political purchase on issues generated by global governance through data, AI and algorithmic infrastructures.
This practice of looking across sociotechnical systems and mapping the effects of data-driven governance diverges significantly from dominant approaches to AI and digital technology in legal scholarship. Rather than trying to open the ‘black-box’ of digital governance to make its inner-workings transparent (‘looking inside’),Footnote 23 or invoking higher-order normative principles to subject those workings to legal checks and balances (‘looking from above’),Footnote 24 we aim to experiment with immanent strategies of critique and analysis that follow (without subordinance to) the emergent relations and practices that global governance by data are enacting.Footnote 25 If data is entangled with the material-discursive practices and sociotechnical relations that bring it into being, then these relations require examination. As Ruppert, Law and Savage argue, grasping how data governs requires ‘attending to the lives and specificities of devices and data themselves: where and how they happen, who and what they are attached to and the relations they forge [and] how they get assembled’, juridically and otherwise.Footnote 26
1.3 Infrastructures in/of/as International Law and Global Governance
One way in which we attend to the lives and governance effects of data throughout this book is by analysing the information infrastructures that give it shape. This approach draws inspiration from the interdisciplinary field of infrastructure studies (spanning critical security studies, anthropology, sociology, STS and media studies),Footnote 27 and builds on emerging scholarship on how socio-material infrastructures enable and condition legal and governance practices.Footnote 28 Data infrastructures include digital devices and data, but also all the wider technical, social and organisational processes, equipment and techniques through which data is produced, circulated and made ‘algorithm ready’Footnote 29 – from the capture, formatting and transmission of data across multiple sites for analysis, to the curation of training datasets and models for producing insights from ML algorithms.Footnote 30 Reframed this way, data infrastructures are more than invisible systems of cables and wires or layers of inert technological substrate. They are emergent articulations that configure and hold together the human-machinic relations that sustain the crafting and governance of global problems through data. Such relations also leverage, compound and/or alter other modes of relation and power, including racialised, class-based and gendered inequalities and forms of domination, as detailed below. And because data infrastructures are imbricated with the practices of those who design, maintain, use and those that are targeted by them, an infrastructural approach allows their transformative effects to be empirically mapped and analysed in motion.
We suggest that thinking law and (digital) infrastructure together is an analytically generative and politically important move in an environment marked by massive expansion in the use of AI-informed driven and ADM algorithmic systems for global governance. It shows that seemingly intractable characteristics of global order – including prevailing distributions of power and resources – require continual assemblage, repair, legitimation, use and organisational work in order to be (re)produced and maintained. It also foregrounds mundane data practices of collection, cleaning and curation that tend to be disregarded in law and technology literature, as regulatory processes that are politically, ethically and normatively charged and empirical problems to be examined.Footnote 31 This suggests multiple strategic points of entry for questioning data infrastructures’ ordering effects and possibly transforming them.
Infrastructural approaches also challenge conventional human-centred assumptions of legal agency and the idea of law as a system of norms that operates principally though reasoned deliberation and persuasion to govern information technology and data-driven problems ‘out there’ in the world. If human actors and machinic processes co-produce how global problems are known and governed through data, for example, then both must be part of how such problems are diagnosed, addressed and critiqued in legal scholarship and practice.Footnote 32 As the contributions in this volume variously show, legal and regulatory practices are moving parts of fluid yet friction-laden sociotechnical environments, put into motion through data and the data infrastructure practices that they purport to govern.Footnote 33 An ‘infra-legalities’ approach to law and technology problems – which follows how legal and infrastructural practices work together to co-constitute emergent data governance ecologies – offers a useful frame for exploring these reconfigurations.Footnote 34 It also carries potential for opening exciting pathways for their critical engagement at a moment when new approaches, questions and methods are urgently needed.
To that end, contributors to this volume explore the following kinds of questions: how is the exercise of authority in global governance materially enacted, directed, maintained and enforced through digital data and associated ‘datastructuring’ practices?Footnote 35 How do the promises of digital sensing and algorithmic sorting impact the ideals and sociotechnical relations, styles and scripts of governance on the global plane – from the realm of counterterrorism and military technology to tech-based dreamscapes in environmental policy and mobility control? How is agency enacted and judgment exercised or resisted in systems shaped by AI and ADM where ‘humans are lodged within algorithms and algorithms within humans’?Footnote 36 How are practices of global governance ‘by data’ nested in longer lineages of governmentality that shape the problems and possibilities of data-driven governance projects?Footnote 37 Which social divisions and signifiers of racial difference dwell within the algorithmic calculus and how do people and machines mess with those?Footnote 38 What are the analytical advantages, methodological challenges and conceptual stakes of foregrounding infrastructural operations to understand these shifts in global authority, law, power and knowledge production? In tackling these and adjacent questions, contributors to this volume explore legal agency and authority not as attributes of specific actors, sites or entities. Rather, legal agency is examined relationally, as emergent effects of processes of entanglement, struggle and enactment across multiple sites that are remaking legal techniques and governance practices through data and its related media practices.Footnote 39
1.4 Intersections: Common Themes, Concerns and Interventions
The relational, sociotechnical and infrastructural approach to global governance by data and algorithmic rule adopted throughout this book allows us to revisit key units of analysis and points of controversy in international law and social theory.
1.4.1 Scales
Our titular reference to the ‘global’ quality of ‘governance by data’ does not refer to a fixed or stable space or scale. We do not ask that readers take the world as they find it nor that they treat the globe as undivided under human authority – far from it. Rather than foregrounding the pre-given, all-encompassing scale of the ‘global’ – as traditionally invoked in literature on ‘global governance’ – we are interested in how varying spatial (and temporal) scales of governance are relationally and materially composed. This structuring of scale – be it through the assemblage of data-driven supply chains or the planetary mapping of oceanic conditions and atmospheric changes – is precisely what is at stake in many of the digital infrastructures that contributors explore in this volume. Emanating from such practices of spatial composition, the globes of the ‘global’ that we encounter in this volume are not one but many.
In the chapters by Angelina Fisher, Benedict Kingsbury and Thomas Streinz (Chapter 7) and Roxana Vatanparast (Chapter 9), devices and modelling tools for ocean sensing and climate monitoring are shown producing a (digitally mediated) planetary scale of environmental governance. This aligns with Kingsbury’s invitation to ‘think infrastructurally’ with ‘planetary time and systems’,Footnote 40 in resonance with Dipesh Chakrabarty’s claim that ‘[w]e are all living … at the cusp of the global and the planetary’ – at the verge of spatial terrain that is the ‘condition of human existence and yet profoundly indifferent to that existence’.Footnote 41 While these chapters explore current efforts to attune governance practices to this planetary scale, they also testify to how the transition from global governance to (digitally enabled) planetary politics is defined by friction and looming risks of capture and control. Focusing on the floating ocean observation infrastructure of Argo, Fisher, Kingsbury and Streinz show how scientists’ ambition ‘to generate ocean data on a planetary scale’ has been confronted by legal, material and corporate conditions yielding an uneven ‘spatial distribution’ of sensory data – a skewed planetary picture. The ambition to measure, map and modulate environmental conditions on a planetary scale, Vatanparast adds, also risks producing a ‘totalizing and universalizing view’ that ‘overlook[s] the spatially heterogeneous effects of climate change and unequal embodied experiences of it’. As Jennifer Gabrys argues, the planetary perspective is often linked with lineages of ‘imperial control’ – ‘suggest[ing] total dominion: the rolling out of behemoth systems that hold the planet and all of its entities in a space of complete capture’.Footnote 42 This signals a need to study how these scales are made and contested and how AI and digital technologies alter the materialities and politics of scale-making and modify the ‘globes’ of ‘global governance’.
These themes of temporal and spatial scale return in the chapters by Julie Cohen (Chapter 2) and Jennifer Cobbe (Chapter 8). ‘Data infrastructures’, Cohen argues, ‘are remaking economic, social and even physical geographies’ in ‘their capacity to scale vertically and laterally’.Footnote 43 This geographical recomposition has important regulatory and political effects: ‘the emergence of data infrastructures’, she notes, ‘has both enabled control and coordination of formerly distinct activities and afforded points of entry for new assertions of political and geopolitical power’. The ‘massive datasets’ amassed by dominant platforms thereby perform infrastructural work of ‘scale-spanning’ which catalyses changes in commercial activity, alters labour conditions, amplifies state surveillance and reconfigures physical spaces. Fostering a ‘potent imaginary of geographically limitless commercial reach’, platform infrastructures produce new scales of governance ‘organized around logics of modularisation, algorithmic intermediation, and control’. Tracing the infrastructural composition of digital supply chains, Cobbe further shows how platforms obtain and exercise ‘regulatory power’ by leveraging ‘asymmetrical interdependencies’. These asymmetries – resulting from ‘legal and technical protocols’ that shape ‘patterns of operation and control’ – ‘afford certain providers regulatory power in their relations with others: the capacity to durably affect or constrain their behaviour’. Importantly, these levers of power and control are diffuse, multipolar and contingent: they do not pertain to specific actors or stem from formal legal ascriptions of authority but continuously emerge (or disappear) as the result of the relational placement of platforms in networks of interdependence between providers, vendors, suppliers and customers. Law – be it ‘international’, ‘transnational’ or ‘global’ – is interlaced in these chains, but not as a determining factor – there is no central sovereign point where regulatory power resides. This points to forms of governance by data that are attuned to the ever-shifting conduits and scales of digital supply chains – a distinctively ‘global’ formation. A Hayekian dreamscape this is not. Frictionlessness (for those who experience it) is the outcome of vast public and private investments in such chains’ defence and the relentless suppression of those who threaten to divert or disrupt them.
Throughout the volume, we thereby encounter different ways in which the temporal and spatial scales of power and authority are relationally (re)composed through the mediation of digital data and the legal fortification and extension of data infrastructures. In foregrounding these shifting ‘globes’ of global governance and the socio-material practices through which these scales are fabricated and sustained, the territorial scale of the state is no longer privileged as it customarily is in international legal work. Following processes of global governance by data resituates international law amid an array of regulatory agents, forces, techniques, institutions, infrastructures and emergent ordering practices not fully encompassed by its standard disciplinary repertoire. Global phenomena are heterogeneous, as this volume shows and best understood in relation to the forms of placement, connection, priority and attachment they claim to surpass, oppose or make relatively small by comparison.Footnote 44 Studying global governance by data relationally is to maintain curiosity about these rivalries and asymmetries in power and the sociotechnical politics of scale-making that creates, links and sustains them in practice.
1.4.2 Subjectivities
Another strand of inquiry explored in this volume traces how legal and political subjectivity are constructed and managed – including how international law draws boundaries between agential subjects and passive objects and centres around the image of ‘the human as a stable and accepted point of reference’.Footnote 45 Against these stable notions of human autonomy and agency, practices of global governance by data are shown redrawing human-machine boundaries in knowledge production and decision-making in high-stakes ways. In focusing on such ‘distributed and composite form[s] of being’, contributors to this volume productively recast questions of legal subjectivity and responsibility across various domains – from automated border control to welfare fraud detection and from military targeting to online content moderation.Footnote 46 The questions canvassed by this volume’s contributors in this connection are vital and include: where might we situate moments of human discretion and decision-making amid the endless feedback loops of AI and algorithmic governance?Footnote 47 Who may be held to account – and how – when legal agency is dispersed across digital decision-making chains and expressed through tentative indicators of algorithmic inference – ‘triaging’, ‘flagging’ or ‘risk scoring’?Footnote 48
This interest in the reconfiguration of legal agency and subjectivity is reflected in the chapters by Jennifer Raso (Chapter 3) and Jake Goldenfein (Chapter 4). Investigating practices of border control, Raso encounters a network of institutions, state officials, non-state actors, databases and technologies (ML algorithms, risk models and interoperability mechanisms) that are tied together in a diffuse socio-legal-technical infrastructure of decision-making. In this web of relations, Raso observes, ‘we find agency and responsibility to be widely distributed’. This raises clear legal concerns for those affected by the algorithmic violence of such systems: ‘the spreading out of agency occurs at a spatiotemporal scale that makes it nearly impossible to formally “name” or pinpoint the many actors who materially contribute to a border decision using available legal tools’. By confronting the entanglement of agency in algorithmic systems, Raso shows how ‘administrative law and its vertical and temporally limited accountability pathways are … rendered ineffective’. Rather than centring accountability strategy on ‘human actions’ as a ‘single node within a decision-making process’, Raso argues for new ‘horizontal accountability strategies’ that can target the ‘interwoven nature of infrastructural relations’.
Raso’s call to take algorithmic ‘decision-making infrastructures seriously’ resonates with Goldenfein’s intervention which focuses on the ‘human’ of the ‘human in the loop’ ideal – one of the most recurrent tropes in efforts to regulate and rationalise practices of governance by data.Footnote 49 Excavating the ‘broader political work performed by the human in the loop as a regulatory strategy’, Goldenfein describes this idealised figure of agency and autonomy as an unreliable talisman the prevalence of which leads to obfuscation and misdirection. The ‘human in the loop’, Goldenfein claims, deflects political responsibility and ‘strategically elides questions around the different ways humans are threaded into decision systems and attendant attributions of responsibility. It separates the allocation of accountability from the complex distributions of decisional authority that come with the deployment of automated systems’. By qualifying only the human ‘decision’ as legally meaningful, he adds, the ‘human in the loop’ ideal ‘obscures the legal significance of automated processes that select individuals for evaluation, marshall data for analysis, and impose obligations on decision subjects to prove or disprove entitlements or obligations’. This ‘misassigns distributions of responsibility between human and technological processes … limits which elements of automated decisions are open to interrogation and contestation’.
Beyond this problematic of human agency, the volume also explores how emerging modes of governance by data disrupt practices and ideals of community and collectivity in international law.Footnote 50 Again, the questions that contributors to this volume pose in this connection are several, including: which forms of sociality and subject-making are enacted and foreclosed in the collection and analysis of digital data?Footnote 51 Which ties of similarity and dynamics of difference animate and are activated by the clusters and collectives distilled from algorithmic patterns – by the ‘fleeting associations foregrounded in data mining’?Footnote 52 Can these fluid, opaque computational categories sustain collective political claims to rights, accountability and social justice?Footnote 53 How can the emergent disjunction between infrastructural publics of governance by data and existing legal and political publics be theorised and evaluated?Footnote 54 How and where are politics of contestation or reuse/disuse/misuse against algorithmic rule articulated and put into action? Can digital and infrastructural encounters spark new relations of sociality and collectivity that resist reifying liberal ideals of legal subjectivity and the regulatory force of in/exclusion in the public sphere or forge an ‘undercommons’?Footnote 55 Can they help in reassembling communities in ways that extend to the more-than-human or that enact new material publics?Footnote 56
These questions surface in several chapters. Focusing on the logic of ‘proxification’, Thao Phan and Scott Wark (Chapter 6) show how technologies of ML both amplify and alter dynamics of racial difference – a construction of ‘race, by proxy’.Footnote 57 The placement of subjects in relational and algorithmically inferred ‘clusters’, Phan and Wark observe, redefines criteria of ‘similarity and belonging’ as measures of ‘proximity or distance’ in a mathematical vector space.Footnote 58 In this sense, they argue, ‘[d]istance differentiates’ becoming a ‘measure of dissimilarity’. The use of clusters thereby produces ‘new forms of grouping’ that are not based on ‘shared characteristics’ but on the ‘alignment and association’ between data points, thereby eroding the (always already insufficient) safeguards of anti-discrimination law. If processes of racialisation are no longer (exclusively) based on constructed or attributed social ‘identity’ but on abstract measures of ‘proximity’, Phan and Wark conclude, identity politics might miss its target. Grappling with ‘proxification’ requires new expressions of collective resistance – a ‘new set of political techniques equal to a politics of proximity’.
This attentiveness to the violent excision of difference and a prospective opening towards other practices of becoming-common resonate in the contribution by Matilda Arvidsson, Daniela Gandorfer and Dan McQuillan (Chapter 11). Adopting a novel style of writing ‘in conversation’, the chapter asks ‘how to make non-fascist modes of resistance, alliance, and community-building sense-able, thinkable, and do-able?’ Confronting the representational logic of new digital technologies by which difference is ‘legitimate only in the negative’, it strives ‘to make sense-able desires, dynamisms, and operations preceding the generation of patterns (of thought, perception, action, data) and to render thinkable alternatives grounded in an ethics of non-binary and equivocal relationality’. It is argued that this cannot be a political demand to ‘restore[] the “rights” of the individual’, because ‘[t]he individual is the product of power’ – the demand is not to restore the collective group as an ‘organic bond uniting hierarchized individuals’ but to compose the collective as a ‘constant generator of de-individualization’.Footnote 59 The digital can be an opening to such (re)compositions, the chapter contends, whether in the form of ‘blockchain-informed modes of commoning, people’s councils, or posthuman alliances’ extending to a transversal, ‘posthuman feminist “we”’ that transgresses thresholds of difference across the human and more-than-human divide.Footnote 60
It is important to note that the contributors to this volume do not, by and large, approach the question of subjectivity through the lamentation of loss – a loss of human autonomy, reason and responsibility – but rather as an opening towards alternative configurations of becoming-common.Footnote 61 There are signs of such an opening in the Argo project, described by Fisher, Kingsbury and Streinz (Chapter 7), which ‘engages non-human publics’. As Argo’s sensory infrastructure is currently conceived, however, this public only performs a passive role: a subject of data extraction (or pollution). Yet, in line with the alliances envisaged by Arvidsson, Ganforder and McQuillan, Jennifer Gabrys is among those outside the pages of this volume who have pointed to alternative planetary perspectives attuned to more-than-human agencies, locations and rhythms. ‘[B]eing planetary as praxis’, Gabrys argues, might allow us to encounter ‘the difference, distance, and duration with, within, and against which it might be possible to think differently about being human and becoming collective’.Footnote 62 In scholarship working into this opening, including in this volume, questions of scale and subjectivity appear as inherently intertwined.
1.4.3 Governmentalities
A third focal point recurrent throughout the pages of this volume relates to changes in governmentality and political technology that are enabled by and related to AI. Contributors explore the impact of such changes – which have been variously conceptualised as ‘algorithmic governmentality’,Footnote 63 ‘global ungovernance’,Footnote 64 ‘algorithmic regulation’,Footnote 65 the ‘ontopolitics of the Anthropocene’,Footnote 66 or the shift from ‘planning to prototypes’Footnote 67 – on key tenets of (international) legal ordering. At the heart of these changes in the routines of global governance lie a number of recurrent features: the displacement of cause-and-effect thinking by inferential and correlational techniques; the reorientation from stable socio-political categories towards non-representational relational associations detected or constituted in data; and the ambition to move towards adaptive and speculative modes of governance attuned to conditions of uncertainty and contingency.Footnote 68 This volume aims to map the impact of these changes in rationalities of rule on key domains in international law and global governance.
Engaging with these shifts in governmentality, Karen Yeung (Chapter 5) focuses on efforts to ‘replicate the success of ML in commercial contexts into the public sector with no regard for the fundamental differences between the two contexts’. Foregrounding AI tools for practices of ‘citizen scoring’, Yeung identifies a mode of governmentality that deploys AI to ‘predict the future behaviour of individuals’ and enable ‘early interventions’.Footnote 69 The chapter situates these technologies in a long lineage of ‘prediction tools’ in public governance and canvasses the legal and political issues with the logic of proxification on which they rely – echoing insights of Phan and Wark (Chapter 6). These issues are not merely related to recurrent computational errors, Yeung argues, but to a more foundational problem of using ‘patterns at population level’ to inform decisions targeting specific individuals. Yeung takes issue with the temporality of pre-emption: the uncertainty and natality marking the human condition, she argues, should be shielded from technologies of algorithmic anticipation and the ‘contextual detachment’ that their deployment in complex social settings inevitably entails.
These shifts impact existing legal safeguards as well as modalities of critical intervention. This becomes apparent in the questions that arise in their wake. How does global governance by data impact or remake the legal protections afforded by modern international law?Footnote 70 How potent is the standard repertoire of critique that targets reductive, causal schemes of liberal and nonliberal reform – so prevalent in international legal theoryFootnote 71 – in confronting the forms of algorithmic governance on which this book focuses?Footnote 72 How are critical interventions being reshaped in line with these shifting styles, scripts and practices of governance – by focusing, for example, on infrastructural modulation and reassembly, forensic activism, heterodox design and ownership structures, or novel forms of algorithmically-enabled non-human witnessing?Footnote 73
There are productive avenues for critical reorientation expressed throughout this volume. The invitation to ‘think infrastructurally’ by Fisher, Kingsbury and Streinz (Chapter 7) is taken on by Cobbe (Chapter 8) who argues that the concentration of platform power – as a result of supply chain interdependencies – can be countered by ‘[m]itigating asymmetries in operation’ at an ‘infrastructural level’, for example by ‘requiring open sourcing of proprietary systems or interoperability between platforms’. This argument for critical reorientation also resonates with Raso (Chapter 3) who observes that we can ‘use infrastructural knowledge to identify new sites [for] strategic action … and to trace how such action affects an infrastructural web’. Goldenfein (Chapter 4) issues a related call to ‘focus on the complex distributions of decisional authority that come with the deployment of automated systems’. Cohen (Chapter 2) raises the prospect of enlivening a wider-than-usual concept of the ‘public interest’ understood as ‘extending to all of the mechanisms through which the data-driven infrastructuring of mapping, mobility, and commerce is performed’, encompassing both ‘matters of configuration’ and ‘matters of knowledge production’. Phan and Wark (Chapter 6) propose to complement a ‘politics of identity’ with a ‘politics of proximity’ that is directly attuned to the pattern- and proxy-based affordances of global governance by data.
1.4.4 Forces
Finally, this volume maintains an explicit focus on questions of algorithmic violence – that is, it tackles how force, prejudice, exploitation, domination and surveillance are lodged and amplified in practices of global governance by data and the infrastructural assemblages through which these materialise.Footnote 74 These interlocking forms of violence are manifested across many different sites, taking novel or emergent forms and expanding, as the contributors to this volume show. Biometric data collection in international humanitarianism and refugee processing, for example, enables new extractive and exclusionary practices, the reconfiguration of (neo)colonial administrative techniques and the development of novel forms of digital inscription and control.Footnote 75 The use of ML in border security, in turn, is extending the reach of surveillance regimes, enacting new forms of inequality and resignifying the dynamics of racial difference.Footnote 76
In the sphere of platform governance, the use of algorithmic tools (such as hash-filtering in content moderation or online copyright enforcement) is generating powerful ‘architectures of control’ and forms of global regulatory power that are reshaping and redirecting established modes of data governance, creating forms of norm-making shaped by very different conceptions of agency and weakening legal accountability constraints.Footnote 77 The algorithmic moderation of ‘online harms’ has precipitated the rise of a precarious ‘ghost’ economy of data cleaning and labelling that is deepening socio-economic inequalities and embedding new modes of extractive informational capitalism around the globe.Footnote 78 New content moderation capacities are also reshaping the powers of platforms and states and fostering novel public-private governance assemblages. As Claudia Aradau (Chapter 10) argues in relation to the asylum process, ‘platform power’ is blurring the existing binaries of international legal ordering ‘by being neither fully opaque nor fully transparent, neither private nor public, neither open nor secret, but by being both’. In other words, the use of advanced digital technologies in global governance is generating new forms control and immiseration and contested geographies of inequality and data-driven violence. Whether these digital technologies can also facilitate modes of counter-conduct remains to be seen, although it is possible – as this volume aspires – to trace some emergent prospects.
Different modalities of algorithmic violence can be distilled from the contributions to this book. First, a range of contributions highlight how practices of global governance by data reproduce and amplify existing injustices and inequalities. This is reflected, for example, in Raso’s claim (Chapter 3) that new ‘socio-legal-technical infrastructures tend to target the same marginalized people and communities for unfair, even violent, treatment’ as previous generations of technology did. Arvidsson, Gandorfer and McQuillan (Chapter 11) argue that the ‘pervasiveness [of AI] and the vast seas of data on which it feeds extend the reach of carceral effects, while the virtual redlining that occurs inside the algorithms reiterates the historical form by fencing people off from services and opportunities’. The violence analysed in these chapters is one of reproduction and extension. Martina Tazzioli and Ana Valdivia (Chapter 12) observe likewise how ‘the use of algorithmic systems in migration governmentality invisibilises, fosters and recrafts racialised bordering mechanisms’. Tracing ‘entanglements between emerging and sedimented modes of violence’, Tazzioli and Valdivia focus on the productive, performative power of digital systems: how they enact (and act upon) racialized subjectivities through the distillation of patterns and correlations – in tune with the insights of Phan and Wark (Chapter 6) on the (re)composition of ‘race, by proxy’.Footnote 79 To grasp the violence inflicted at the digital border, Tazzioli and Valdivia argue, we should look beyond the problems of surveillance and repression (‘the repressive hypothesis’) and focus on productive practices of subject-making that amplify existing forms of racialized exclusion.Footnote 80
A second and related manifestation of algorithmic violence is manifest in the novel fractures, injustices and dynamics of difference that practices of governance by data generate. Tazzioli and Valdivia contend that digital technologies not only reproduce past and present injustices but also ‘re-craft social injustice along new lines’. This is reflected in how lines of racialisation are multiplied and redrawn – as Phan and Wark explore – through a post-visual logic of proxification. Underlining emergent injustices of governance by data, Claudia Aradau (Chapter 10) similarly focuses on ‘new forms of epistemic violence’ faced by asylum seekers which do not flow exclusively from ‘knowledge erasure and disbelief’ but also from digitally mediated ‘demand[s] to know the unknowable’. These demands for ‘more and more data’, she observes, engender an ‘experience[] of data wrangling’ that alters and amplifies the vulnerability of those seeking international protection. Aradau discerns this form of epistemic violence in decisions of the UK Upper Tribunal (Immigration and Asylum Chamber) (UTIAC), which, in its demands for ever more (meta)data, intensifies ‘asymmetries of power’ and requires asylum seekers ‘to make themselves surveillable in ways that harbour future risks of harm’. This violence in turn fosters novel forms of ‘platform power’ that are reconfiguring and obscuring the boundaries between data extraction and surveillance.Footnote 81
A third concern of this volume, following the argument by Tazzioli and Valdivia, is how racialised and other forms of violence are not only ‘enhanced’ and ‘re-structured’ but also obscured. In relation to the epistemic violence exposed in her chapter, for example, Aradau argues that data provided by asylum seekers to evidence the risk of surveillance ‘is never sufficient, not just because social media platforms produce many opacities about the data they collect and share but because their operations are dispersed and can only be held together through machine learning algorithms’. The traces of data thereby become intractable. This concern with the diffusion, multiplication and obfuscation of power resonates with Raso’s observation that ‘algorithmic violence become[s] shielded from recognition and scrutiny when performed by an infrastructural network that seems infinitely expansive’. Thinking with Arvidsson, Gandorfer and McQuillan, this political dislocation takes on a broader onto-epistemological dimension. In encounters with algorithmic systems that draw performative lines between ‘who or what matters and is excluded from mattering’, they ask urgent questions about how to navigate these forms of algorithmic force and violence?
1.5 Conclusion
This volume explores these algorithmic power relations and the practices of resistance they are engendering, exploring possibilities for justice, critique and accountability in relation to global governance by data. In mapping these dynamics, contributors trace recurrent entanglements between AI models and ML algorithms, ADM practices and histories of racial hierarchy and exclusion.Footnote 82 As they trace how longer legacies of extractivism, inequality and subjection dwell within and become amplified by data-driven regimes of rule, contributors to this volume reiterate some central themes and concerns of critical international legal theory. Yet, in mapping the infrastructural routes of algorithmic governance, certain tropes of this critical repertoire are also called into question. For example, the material and relational networks of governance traced in this volume display neither the structural indeterminacy nor champion the political promise of contingency at the core of international law’s critical repertoire – at least not in the forms in which they have been analysed to date.Footnote 83 The collective attentiveness of this volume’s writers to pattern and formula rather than irregularity and insight offers different entry points into international law’s politics and how it might be leveraged and contested. In doing so, we aspire to approach patterns of violence, exclusion, inequality and domination infrastructurally to explore possibilities for critique that are immanent to practices of global governance by data yet not beholden to them. And in the world we currently inhabit, that is being recomposed by AI and the sociotechnical practices of algorithmic governance, new forms of mapping, organisation and (legal) critique are urgently needed.
Open access