In the first volume of The Automatic Society, Bernard Stiegler considers the social and philosophical implications of the predicted imminent increase in the proportion of labor undertaken by automata. For Stiegler, the social stakes of this development concern the definition of work and the link between work and remuneration; the political challenge is accordingly to foster the adoption of emergent technical forms in such a way as to provide a beneficial solution to these social questions. The politics of automation at work here has a further dimension, however, which Stiegler does not elaborate, but which this essay seeks to specify: namely, a reconceptualization of effective political agency as distributed across the co-constitutive relation between the form of life we call human and its contemporary technical prostheses.
[End Page 136]
Our present moment is doubtless unusually aware of the questions posed by the irreducible relation between the form of life we call human and its technical prostheses. At the very least, the stories we tell ourselves about this moment certainly suggest as much. If such intensified awareness has plainly always been a defining characteristic of the industrial era, the speed of the technological developments at its root would seem only to have accelerated in recent times—although just such a sense of acceleration has of course also consistently characterized the self-image of this era as such. What such contingent fluctuations in this awareness can miss, however, is precisely that in this relation, the human and the technical emerge as mutually constitutive. From its refunctionalization of body parts, minerals, ambient conditions, and so on, to its distribution of capacities across an extended field, this relation configures what might otherwise be understood as the secondary prosthetic technical extension of native human attributes as in fact operative from the beginning. As Bernard Stiegler puts it, for example, with specific reference to the defining processes of the supposedly exceptional human mind: "Prosthetization impacts what Kant in The Critique of Pure Reason calls the syntheses of apprehension, reproduction, and recognition. But this impact is possible today only because it is originary."1 The following essay explores the politics of this impact in the context of current debates about the automation of labor and indeed, of politics itself. With Stiegler as my guide, I will first unpack the relation between the automation of labor and the originary co-constitution of the human and the technical, before looking at Stiegler's response to this automation and, beyond this, at the impact of our defining contemporary technical forms on the question of political agency.
Where Stiegler typically locates the politics of technics in struggles over the ways in which particular technical forms will be adopted by the collectives with which they have emerged into being, I want to use his insights, along with those of other contemporary thinkers, as prompts towards a model of agency more specifically attuned to the composite modes of existence defining today's key political actors. As I develop it in the final part of this essay, this model aims first to provide a convincing account of how such agency might be thought to operate: to this extent, as will be seen, it implies no particular political orientation. Beyond this initial aim, however, my ambition here is to propose a model of agency which might plausibly and effectively contribute to a contemporary left politics of collective justice and emancipation.
The Politics of Automation
In the first volume of his Technics and Time series, Stiegler draws on Jacques Derrida's logic of the "originary supplement" and André Leroi-Gourhan's account of the development of the cerebral cortex in early hominids to configure "human" capacities and technical prostheses as always already formed in tandem. In the second volume of this series, Stiegler unfolds this paradox of originary prosthetics with specific reference to what he calls the "industrialization of memory." To remedy its finite capacity, human memory comes into being as originarily aided by all manner of technological supports. [End Page 137] "Technics does not aid memory," writes Stiegler: "it is memory, originarily assisted 'retentional finitude' " (TT2, 65). This originary technicity exposes memory to the possibility of industrial appropriation and exploitation, when the supports which are "the very conditions of its e-laboration" (TT2, 8) are integrated into an economy of standardized industrial production for commercial gain. Stiegler accordingly defines the "industrialization of memory" as "the industrial synthesis of retentional finitude subjected, as pre-judgment, to the specific criteriology of calculable credit as the operator of economic development" (TT2, 9); writing in the 1990s, he describes memory itself as having become "the primary matter of industrial activity" (TT2, 61). And with Apple, Google, Microsoft, Facebook, and Amazon topping the Forbes list of the world's most valuable brands as of July 23, 2018, it is hard to disagree.2 As this list implies, the industrialization of memory entails the translation of its contents into "merchandise whose value is correlated with its time and space of diffusion" (TT2, 107).
The political dimension of this industrial monetization lies in the question of selection criteria. For Stiegler, "Memory is always the object of a politics, of a criteriology by which it selects the events to be retained" (TT2, 9). In the digital era, with organic retentional finitude supplemented in effectively real time by programs with economic selection criteria, human-technical memory is constituted according to economic priorities on a scale and at a speed sufficiently unprecedented to represent a qualitative shift. In Stiegler's view, this constitutes a technological event of unsurpassed significance. "When the elements of a contemporary technics designed for information processing develop," he writes, "a truly automatic activation of memory will have appeared" (TT2, 79; my emphasis). If it is always the case that, with reference to prosthetically supplemented memory, "the question . . . regards effective means of access, and, through it, of sequencing," and if protocols regulating access and sequencing have of course always served economic interests, the intensity with which this question is posed by the digital industrialization of these prostheses means that "contemporary technical epokhality is radical, equal to the most powerful ruptures that humanity has ever known" (TT2, 74). (The "epokhality" of a technical form is in Stiegler's idiom its capacity to suspend a previous phase and inaugurate a new one.) Although Stiegler may appear in such moments to be aligning himself with a techno-catastrophism, his understanding of technics as "originary prostheticity" in [End Page 138] fact rules out any such apocalyptic bent: for what responds to epokhal rupture is always, one way or another, a political decision concerning the way in which the new technical regime will be inhabited. Progressive or conservative, say, anarcho-libertarian or techno-fascist: the struggle to determine this mode of adoption is for Stiegler the very definition of politics. Though we may be faced with "a truly automatic activation of memory," then, we are nevertheless not looking at the usual fantasy of the "rise of the machines": if "this is clearly a question of the autonomization of tekhnē and its automobility," we nonetheless know "that there has always already been 'autonomization' of the techno-logic automaton," and therefore that "it is not a question of an autonomization relative to a golden age of the mastery of a technics 'closer' to human beings, but rather of a becoming-hegemonic of developmental economic imperatives in accord with a particular interpretation of time qua value" (TT2, 140). This is not destiny: this is a particular mode of the distribution of capacities, which may accordingly be challenged by the composite agents which bear these capacities. This is Stiegler's politics of automation.
In the first volume of Automatic Society, Stiegler develops a particular position within this politics, in his analysis of what has become the pressing question of the automation of paid work, familiar from a wide range of recent discussions.3 Citing reports of various pieces of research, he notes the potentially devastating effect of large-scale computerized automation on the landscape of employment in industrial societies, with approximately 50 percent of jobs highlighted as potentially at risk over the next two decades.4 Responses to this scenario typically maintain a metaphysical opposition between human and machine, in a mode that is either utopian or apocalyptic: a vision of newly rewarding free time (as evoked especially by Marx in the "Fragment on Machines," and latterly promoted by Srnicek and Williams),5 or alternatively of "the machines taking over"—with the machine respectively as the still-loyal servant, or as the rebel servant become master. In either mode, however, the refusal to think the human and the technical as co-constitutive leaves the machine as the more or less reliable instrument of what remains in principle sovereign human agency. In Stiegler's argument, by contrast, this insistence on a relation of instrumental exteriority between the human and the technical can only prove disastrous—can only lead, in fact, to the instrumentalization of both.
As we have seen, the human is constituted for Stiegler in and through its relation to its originary technical prostheses, which themselves are constituted in the cycles of this same relation. A technical regime which replaces this co-constitution with mutual transcendence and instrumental exteriority generates that mode of adoption which Stiegler, borrowing from Gilbert Simondon, calls "proletarianization"; in this mode the originary composite is experienced as split, as the once-sovereign "human" worker becomes the mere operator of a merely "technical" machine itself become the locus of functional knowledge. Crucially, this proletarianization thereby minimizes the worker's participation in those distributed systems in which capacities are generated—specifically, here, the production and circulation of knowledge. The "employee," for Stiegler, the wage laborer, is by definition proletarianized, blocked from meaningful involvement in the generation of individual and collective knowledge: "Employment," he writes, "is characterized by [End Page 139] the fact that the retentions produced by work no longer pass through the brains of the producers, who are themselves no longer individuated by work, and who are therefore no longer the bearers or producers of work-knowledge" (AS1, 160–61). The employee's sole function in this situation is to optimize the performance of the machines in which knowledge is now located. This situation is first exacerbated by the cybernetic management of labor processes via real-time digital feedback, in which the employee is "subject to protocols supported and defined by machines, . . . reporting systems and management control"; with the advent of digital-algorithmic automation, this employee is now literally "without employment, short-circuited by algorithms that outstrip psychic individuals and collective individuals" (AS1, 161). What is more, the harvesting of data made possible by this digital-algorithmic regime sees the un-employed employee mutate into a producer of financially translatable information:
Nowadays, with full and generalized automation, and even more so in the future, there is and will continue to be a decreasing need for employees to serve and set machines that are increasingly becoming completely automatic. It will increasingly be externalities which, producing the traces generated by "connected" psychic individuals, will set the parameters of the systems of production and distribution characteristic of algorithmic governmentality. These are increasingly integrated, increasingly autonomous, and no longer served by producers but by consumers—to the extent that any of these are left amidst the generalized bankruptcy this system fundamentally installs.(AS1, 163; translation modified)6
From worker to machine-operating employee to data farm animal, the trajectory of proletarianization sees participation in capacity-producing circuits reduced to its degree zero. At no point in this process are we dealing with an activity meriting the name of work: for Stiegler, this term is appropriate only where such participation is operative. By contrast, what we see in this process is only employment, up to and including the harvesting of data generated by un-employed employees. Its first two phases comprise full industrialization with "scientific management" à la Taylor, followed by the full extension of the consumerist economy: through these, "proletarianization transforms work in its totality into jobs devoid of all knowledge and requiring only skills defined in terms of 'employability,' that is, 'adaptability'" (AS1, 172; translation modified). Its third phase, through which we are currently living, sees the activities of consumers start to displace labor in a conventional sense as the source of profit: if this is often termed unpaid "work" on the part of these consumers, Stiegler insists again that the term has no place here. Far from "work," this is "the employment of unpaid time—that, by harnessing this employment of the time of the individuals who we try to remain, feeds, reinforces and sets the parameters of the automated and performative collective retentions automatically and performatively produced by totally computational capitalism" (AS1, 163).7 Already in the era of analog broadcasting, "liberated time that does not become free work becomes brain time available for the fabrication of mass consumer markets" (AS1, 179; translation modified); the shift to digital platforms has only found more capillary ways to monetize this "available brain time," now employed to generate the data which drive ever more sharply [End Page 140] targeted forms of commercial address.8 Indeed, so capillary are these ways that they extend beyond the brain, harvesting in new form the output of the originarily technical composite that the human has always been: as sociologist of digital labor Karen Gregory writes, "while 'you' may not ontologically 'be' a gadget such as a camera or a pedometer, as long as those elements are attached to you and producing trails of data, you are a generative source of data aggregation who becomes, often unknowingly, a constituent of data-based populations."9
As we have seen, for Stiegler the political question posed by and to the co-constitution of the human and the technical concerns both the destructive sequestration of knowledge production in circuits which reduce the participation of "workers" to its zero degree and the possible combative reconfiguration of these circuits to include just such participation. Stiegler's position in relation to the progressive replacement of the worker by the un-employed data farm animal is for this reason not solely one of despair. Where the metaphysical opposition between human and machine can see increasing automation only in terms of the reduction of labor time, Stiegler insists—with Marx—that the stakes are in fact "the end of employment and . . . the reinvention of work" (AS1, 166), this latter representing the positive possibility he finds in this scenario. The question, he writes, is whether "productivity, which will escalate in spectacular fashion in the next few years thanks to full and generalized automation, should liberate time or liberate work" (AS1, 170). This is the contemporary political decision as to the meaning of automation, which, as we saw above, forms the overall horizon of Stiegler's Automatic Society: since purchasing power is likely to collapse with the massive replacement of wage labor by automation (AS1, 163), through what economic forms are the resulting productivity gains to be redistributed? (Assuming this redistribution itself is held to be desirable, of course—the alternative plainly being the mass destitution of former unemployed data farm animals whose data have become worthless in the absence of their purchasing power.) In Stiegler's view, this challenge cannot be met solely by the widely touted solution of a Universal Basic Income (UBI), however important this latter step may be: although his position here may resemble Srnicek and Williams's accelerationist celebration of the capacity of automation to "free up time," for example (underpinned economically by the introduction of a UBI), Stiegler's understanding of human-technical co-constitution means, as we saw above, that he cannot embrace the instrumental relation to technology which grounds this celebration. Stiegler's specific proposal is, rather, for an economy which would take seriously this co-constitution and the consequent importance of general participation in circuits of knowledge production: namely, one [End Page 141] founded on what he calls a "contributory income" (AS1, 180), in which the free time produced by massive automation would allow groups and individuals to develop forms of technically constituted knowledge whose social contributions would be remunerated.10 The economics of this model rests on the notion of "intermittent contribution" (according to which French artists have been able to receive state funding even when not actively working);11 its psychosocial potential lies in its integration of those involved into that participation in circuits of knowledge production which for Stiegler defines work in the strong sense. As he writes, accordingly: "reinvented work policies [constitute] the cornerstone of a general politics of recapacitation" (AS1, 181).
The Automation of Politics and Political Data Agents
We might think of the first phase of such recapacitation as developing the kinds of literacy necessary for meaningful participation in contemporary cultural production, broadly understood. Recalling the eighteenth-century context of Immanuel Kant's "An Answer to the Question: 'What Is Enlightenment?,' " we might emphasize, as Stiegler does, that the accession to mature reason Kant champions in this text is linked to just such literacy: Kant defines "the public use of reason" as "that use which anyone may make of it as a man of learning addressing the entire reading public."12 If in a print culture, reading and writing are functionally conjoined, the same is no longer true in a digital culture, where content and code are no longer identical: and the resulting illiteracy (or "proletarianization") opens a gap through which humans and technical forms are separated out into a relation of mutual exteriority, and both are reduced to the status of mere instruments. I will now look briefly at the operation of this gap in the context of contemporary digital media, and the critical response this may be thought to require, in order to open the question of the kind of specifically political agency this context might afford.
In the second volume of Technics and Time (written twenty years before notions of "fake news," "impression management," and so on became common currency), Stiegler suggests that the effective elimination of delay in "real time" technologies—that is, the reduction of delay to a level experienced as negligible—produces a "temporal rapture" which itself leads to an "inability to distinguish facts from fabricated facticality" (TT2, 116). In this state of affairs, "the contemporary politician is increasingly a manager of opinion," and "public life . . . is a function of real-time procedures, . . . the goal of which is audience-control through quantified knowledge of current trends and, through direct, 'strategic' 'no-delay' interventions, on their evolution" (TT2, 122–23; translation modified). It may even be, he suggests, that politics in the strict sense can only be articulated over the long term: to the extent that all politics entails deliberation and decision, its rhythm is scored along technologies of relay, and indeed delay. When that rhythm sees this temporal gap narrowed to the minimum of real-time circulation, the result is "the end of strictly political programs coordinated with coherent and integrated choices made over the long term as functions of ideas and collective actions, which are replaced by objectives and strategies of communication" (TT2, 123). [End Page 142]
Stiegler's distinction here between the "strictly" political and whatever it is that replaces this under a regime of "temporal rapture" is not especially helpful: the pre- or anti-deliberative practices that replace (or better, overlay) his ideal order of debate and long-term projection are still plainly political in other terms (in Leninist terms, for example: as pure power play); and indeed, such alternative perspectives might observe that these other dynamics have always also been at work within politics "in the strict sense." What we should take from it is not this distinction, then, but rather the key point that any political order has at its core the rhythm of its defining technologies of circulation. On this basis, Stiegler's analysis emerges as strikingly prescient, given the challenge posed recently to the post-Enlightenment understanding of the public sphere by the "filter bubbles" created by social media and the associated monetization and partisan political deployment of data extractivism.13
The technological horizon here may be identified as what Antoinette Rouvroy and Thomas Berns call "algorithmic governmentality." This concept—a major reference for Stiegler in Automatic Society—is defined by its authors as: "the automated collection, aggregation and analysis of big data so as to model, anticipate and preemptively affect possible behaviours."14 In their account, where statistical governmentality (as analyzed by Michel Foucault, and persisting alongside its algorithmic successor) entails reference to an "average" citizen, consumer, etc, algorithmic governmentality entails "apparent individualization," "the idea of one becoming one's own profile, automatically attributed and evolving in real time" ("AG," x). But crucially, for Rouvroy and Berns, "algorithmic governmentality produces no subjectification, it circumvents and avoids reflexive human subjects, feeding on infra-individual data which are meaningless on their own, to build supra-individual models of behaviours or profiles without ever involving the individual, and without ever asking them to themselves describe what they are or what they could become"; consequently, "the moment of reflexivity, critique and recalcitrance necessary for subjectification to form seems to constantly become more complicated or to be postponed" ("AG," x). It is the absence of such a reflexive, critical moment that allows Rouvroy and Berns to make the connection to the public sphere: just as Wendy Brown analyzes the endgame of "governance" as the dissolution of the political in the economic-managerial, so they describe the "colonization of public space by a hypertrophied private sphere," and "the systematic capturing [End Page 143] of any available human attention for the benefit of private interests . . ., rather than to foster democratic debate and serve the general interest" ("AG," v).15 Loyal to a Foucauldian tradition, Rouvroy and Berns insist that individuals are not simply in the grip of this mode of power: rather, the significant point is that "irrespective of their capacity for understanding, willpower and expression, 'power' approaches them no longer on the basis of these capacities, but rather on that of their 'profiles'" ("AG," xii). Nevertheless, it remains their argument that "algorithmic governance . . . neither produces nor provides an affordance for any active, consistent and reflexive statistical subject likely to lend it legitimacy or resist it"; and so they ask, in conclusion: "Does the regime of digital truth (or digital behaviourism) not threaten, today, to undermine the very underpinnings of emancipation by eliminating notions of critique and of project . . . ?" ("AG," xxvii, xxviii). With this reference to the project recalling Stiegler's emphasis on the rhythms of political regimes, we might ask: can there be such a thing as emancipatory politics at the speed of the digital algorithm?
To anticipate a little, my answer to this question will be "certainly not," as long as we insist on a metaphysical opposition between the human and the technical. Although they come close to it with their opposition between critique and digital interpellation, Rouvroy and Berns largely avoid this trap through their emphasis on processes of subjectification and individuation; elsewhere, we can see this question of just how to conceptualize distributed digital capacities playing out in current literature on the politics of our algorithmic culture. As David Beer writes of the field of critical data studies, "A primary concern here might be the meshing of human and machine agency."16 Three tendencies may perhaps be identified in relation to this concern. The first two respectively entail a continued reliance on the sovereignty of human judgment and a Latourian description of the assemblages through which algorithmic data processing acquires its agency: if the former is unequivocal in its appeal to human exceptionalism, the latter can be politically insubstantial, foreclosing a sense of the assemblages in question as sites of the decisive, confrontational agency required by a politics of emancipation.17 More promisingly for our purposes, a third tendency considers digital algorithms as actors in processes of sociopolitical antagonism. In the work of Kate Crawford, digital platforms are pictured as riven by forms of antagonism between humans, between humans and algorithms, and even between algorithms themselves. According to Crawford, "Such a logic emphasizes that algorithmic decision making is always a contest"; it would, she writes, "assume perpetual conflict [and] recognize contestation."18 Drawing on the political philosophy of Chantal Mouffe, Crawford is looking to "find purchase for an understanding of the work of algorithms using agonistic pluralism":19 while agreeing with such authors as Neyland and Möllers that digital algorithms "are rarely stable and always in relation with people: that is, both in flux and embedded in hybrid spaces," she adds to their emphasis on uncertain assemblages a welcome note of political strife, while mostly managing to steer clear of the human exceptionalism on which this usually relies.20 A similar approach is taken by Ezekiel Dixon-Román, who examines how "sociopolitical relations of 'difference' . . . materialize in the digital acts that performatively discipline humanity."21 For [End Page 144] Dixon-Román, the racial politics of online profiling, for example, are configured not by sovereign human actors alone, but in the "more-than-human ontologies of digital architectures." If this means that "racializing assemblages are beyond the ontological sovereignty of the human," it also enables us to understand the digital algorithm as a decisive actor within forms of social conflict, whose ordering protocols, with their unexpected gaps and overdeterminations, take its actions beyond the mere expression of its human author's design—and thereby to grasp the decisive role played here by the composite assemblage (economic, institutional, social) of which it forms a part.22
Building on these approaches, and with Stiegler's account of human-technical co-constitution very much in mind, I now propose a model of the political agency at work across such assemblages with enough descriptive traction to prove plausible and effective within our digital algorithmic culture. As with Stiegler's contributive economy, and as is often called for in the literature on our algorithmic culture, this will first require the development of enhanced functional and strategic literacy, allowing competent participation in the manipulation of the codes that sustain digital platforms and general understanding of the ecology of their operation and ownership, including the forms of subjectivity they make possible and the commercial structures by which they are sustained. As Beer writes, then, "This will require us to understand the technicalities of the systems as well as their social ordering potentials."23 This can be only a first stage, however: for it risks playing down the full role of the technologies in question, letting us imagine that we are understanding and manipulating them from the outside, however much we acknowledge that these objects also act on us. No adequate response here can forget the existence of what Göran Bolin and Jonas Andersson Schwarz call "the metricated mindset":24 the very thought processes through which we are deciding our behavior on the basis of our new understanding of technical forms are no longer conceivable as transcendent in relation to these forms. A subsequent stage would accordingly require us to conceive of a form of decisive political agency exercised by constitutively composite agents. If we certainly need to become a new critical "reading public" (and more genuinely inclusive than the first iteration of this ever was), then, we also need to unhook notions of critique and intervention from our fantasies of human transcendence and learn to think even political agency as at once distributed and conflictual.
Were something like this to become possible, we might be able to conceptualize the mobilization of what we could call political data agents, composites exacerbating [End Page 145] algorithmic agonism into decisive political intervention. Such agents already exist—we will encounter some of them below—but we still need to understand the kind of agency they deploy. Stiegler will take us a long way with this: key here especially is the fact that for him, the political decision on the mode of adoption of a technical form is structured recursively, in that it is taken by agents constituted within that form's regime. With this insight, Stiegler allows us to think critical adoption without transcendence. Stiegler will only take us so far, however, for two reasons. First, his attachment to a concept of work defined as an authentic source of value (in contrast to its alienation in labor) threads through his thinking traces of human "species being" which inhibit a conceptualization of composite agency—including the production of value—that might break decisively with this schema.25 An understanding of agency as composite all the way down and at every level can only be compromised by the human exceptionalist baggage of attempts to determine value by criteria of authenticity. Secondly, the model of specifically political agency we need to develop will have to exacerbate Stiegler's emphasis on the strife characterizing arguments over the adoption of technical forms into a firmer and wider embrace of conflict. If the "contributive economy" represents a significant social policy proposal in the era of increasingly automated labor, modelling political agency requires us to add some sharply antagonistic grit to its essential emphasis on participation. We will need to look beyond Stiegler, then, to develop a fuller sense of what the political data agent might look like. We will need to find our resources in something less proper, something more messily composite, something with more edge. We will need, in fact, something weirder.
We might get an intimation of this from the "weird solidarities" that digital sociologist Karen Gregory imagines at work in the contemporary assemblages across which agency is distributed. As Gregory notes, the value currently extracted from data "is already predicated on a social body and the generative connections that can be forged among its constituent elements"; and these elements, she specifies, "do not necessarily have to reduce to 'the human.'"26 When Gregory writes that "the data economy extends and opens the human body to the pre-personal, to what we do not necessarily have conscious access to," for example, she has in mind the genetics by which we are traversed; but she might equally well be describing the impersonal memory which for Stiegler is embodied in technical forms. Crucial here is Gregory's insistence that we go beyond the dystopian visions called up by our disappointed metaphysics: that "we must also take up the notion of 'weird solidarity' to explore how the 'more-than-us-but-not-us' is giving rise to new forms of sociality and social relations"—and, I would add, new forms of agency. In Gregory's words, "A map of this weird terrain is necessary to foster a politics [End Page 146] of solidarity that understands how and where value is being produced": just as Marx identifies the conditions of possibility of revolutionary action precisely in the industrial concentration of the working class, so Gregory stresses that the already existing fact of intimate, saturated coexistence with digital technical forms opens the possibility of a new kind of agency. We might describe this as composite agency reinflected as a solidarity through which to contest the political-commercial reduction of its defining relationality to mutual instrumentalization. Gregory writes: "A weird solidarity is already being created among us, in and through the ubiquitous project of building algorithms into every facet of day-to-day life. That solidarity is an essential aspect of the aggregation of the data. That solidarity is there, and therefore it is there for us to see, to experiment with, and to build from."
If Gregory mostly inflects her weird solidarities in the direction of more attentive modes of coexistence, I want to expand her proposal by drawing out another aspect, implied but not developed in the phrases I have been quoting, and towards which I have been moving here: namely, that these new solidarities across the biological, the cognitive, the social, and the technical might present the conditions for the emergence of a specifically political form of distributed agency. To the extent that we have overwhelmingly conceived of politics as the realm of deliberation and decision, and of these as exceptionally human capacities, we have rarely been able to imagine something like this agency. But the recursive-reflexive decision on adoption which, for Stiegler, is a defining capacity of the human as co-constituted with the technical, means that we have no need to elevate, say, human will to a fantasy level of transcendence in order to secure the possibility of trenchant political decision: the self-consciousness that this possibility requires is already immanent within this process. It thus becomes entirely possible to conceive of decisive agency as emergent within processes of human-technical co-constitution—and so, of this decisive, distributed agency in political form. Digital algorithms are already making decisions, of course, faster than we can think, and competing with each other to do so (on the price of derivatives, say, or the cost of life insurance); but as Rouvroy and Berns point out, their imperative is invariably to avoid any kind of drastic breach. Can we on the contrary imagine forms of interruptive and even emancipatory political action taken by assemblages composed of ontologically plural participants?
I think we can. From the labors of Cambridge Analytica to the hacktivism of Anonymous or alt-right social media persecution, from #MeToo to #BlackLivesMatter to the Fight White Genocide movement or, in South Africa, #RhodesMustFall, varied and antagonistic forms of interventionist digital solidarity are already in evidence. If this reminds us that there is nothing necessarily progressive about digital politics, it also indicates the possibility and the reality of political data agents, issuing from the very structures of digital domination. Against the increasingly smooth integration of finance and security that is the nexus of contemporary algorithmic governmentality, we might set the fact that, as Rouvroy and Berns write, it is in fact difficult "to produce algorithmic subjects who conceive of or think about themselves as such" ("AG," xvi). Rather than seeing in this difficulty a possible survival route for critical consciousness conceptualized via [End Page 147] Enlightenment-modernist human exceptionalism (as Rouvroy and Berns tend to do), we might, however, see it as opening the sites of digital composition as zones of possibility and contestation. If the digital algorithm does tend inexorably to minimize disruption, integrating lessons from any ruptures into its evolving functionality, it is not therefore by nature technocratically anti-political. This may overwhelmingly be the capacity it contributes to its current assemblages, but it need not exhaust its possible modes of participation.
What I am proposing, then, is that we understand the algorithmic assemblages currently engaged in many of the signal struggles of our time as already existing political data agents, characterized by the impossibility of attributing their key capacities definitively to one or other side of the metaphysical division between the human and the technical. Not only are their interventions inconceivable without the massification and granular reclassification of data on a scale and at speeds which are the hallmark of the digital algorithm; these processes also and thereby configure novel modes of existence in which avatars and profile management, say, intertwine with the exercise of or exclusion from political and economic power, and affects of rage, elation, or despair both individual and collective flow through and are indexed to digital platforms. The massification and reclassification which organize these platforms give rise to new collectivities whose effective identity is inseparable from the distinctive topology of their virtual milieu: solicited not as subjects but as profiles, as Rouvroy and Berns correctly observe, participants intervene and are addressed as a site within a collective, where digital meets cognitive meets affective, as a singular-plural nexus within the field of economic, social, and political forces. (The political social media troll, say, is an unprecedented social-existential form, whose specific psychosocial configuration is digital-cognitive-visceral.) No action in this area, from the most anodyne to the most drastic, emanates from anything other than an assemblage.
Whether we consider electoral manipulation via fine-grained user profiling, quasi-legal funding routes, and legions of bots, or social movements of every stripe, we find modes of antagonistic intervention whose agents crystallize where the exercise of power meets the kinds of group formation operated by their platforms. Working in tandem with the forms of a pre-digital idiom (demonstrations, speeches, lawsuits) and the affordances of other technologies (the body camera, the voice memo, the offshore bank account), these composite, collective agents are constituted in the platform's dynamics: their [End Page 148] existence is defined by the visibility and anonymity, the speed and social penetration, the deliberative plasticity of their particular assemblage. What is more, if their actions are politically decisive, interruptive, and—in some cases—emancipatory, the mutating "realtime" rhythm of this action sees authorship of even its key moments distributed across the assemblage as a whole, as such key moments emerge in the rapid cycles of feedback, anticipation, and modification. There is no need here to be nostalgic for a golden age of deliberative democracy (in the agora or the town hall): if such earlier versions were themselves also organized by the confluence of economics and technics, conversely these more recent forms plainly sustain sites and practices of deliberation and contestation. If the rhythm of these practices can no longer be thought in terms of exceptional human qualities, this might itself produce a finer-grained exploration of political agency as the differentially—indeed often discriminatorily—configured meeting place of the socioeconomic, the existential-affective, and the technical.
From within the automation of politics, then, there emerges a possible politics of automation: as a Stieglerian policy of adoption concerning our defining technologies, answering the shock of their impact with the new cognitive and organizational possibilities offered by critical intimacy with their forms, yes; but beyond this, political data agents whose agency is distributed across manifold assemblages, and whose critical decisions and conflictual interventions are fully immanent to this composite world. And—in their progressive form—whose improper, disparately networked collectives are fighting throughout the digital public sphere for an algorithmic culture not of financialization and racialized securitization, but of common emancipation. [End Page 149]
Martin Crowley is Reader in Modern French Thought and Culture at the University of Cambridge, where he is also Anthony J. Lyster Fellow and Director of Studies in Modern and Medieval Languages at Queens' College. His most recent book is L'Homme sans: Politiques de la finitude (Lignes, 2009; with an afterword by Jean-Luc Nancy). He serves as General Editor of the journal French Studies.
. Thanks to Emily Apter, Michel Feher, Karen Gregory, and Mara Polgovsky for discussions which have greatly helped me to refine the model I am seeking to develop here.
6. Stiegler takes the term "algorithmic governmentality" from the work of Antoinette Rouvroy and Thomas Berns: I will return to this below.
8. Stiegler refers with this phrase to the comments of Patrick Le Lay, who when in charge of French TV network TF1, observed in 2004 that for advertising to work, "the brain of the viewer needs to be available. . . . What we sell to Coca-Cola is the time of this available brain" (Stiegler, Symbolic Misery, 2, 46).
11. See the French employment website Pôle emploi, "Les allocations versées aux intermittents du spectacle," https://www.pole-emploi.fr/informations/les-allocations-versees-aux-intermittents-du-spectacle-@/article.jspz?id=60567.
14. Antoinette Rouvroy and Thomas Berns, "Algorithmic Governmentality and Prospects of Emancipation," 10; hereafter "AG." Rouvroy and Berns use "algorithm" in its usual contemporary sense, as shorthand for "digital algorithm." For convenience, I will mostly preserve that usage here.
17. For the first tendency, see Karen Yeung, " 'Hypernudge': Big Data as a Mode of Regulation by Design"; for the second, see Rob Kitchin, "Thinking Critically about and Researching Algorithms," and Daniel Neyland and Norma Möllers, "Algorithmic IF . . . THEN Rules and the Conditions and Consequences of Power."
19. Crawford, 87.
22. Dixon-Román, 489.
26. All quotations from Gregory are from her online essay "Weird Solidarities," http://dismagazine.com/discussion/72958/karen-gregory-weird-solidarities/.