The politics of deautomation: Being disaffected by platform capitalism

Introduction

Of all the contemporary technological developments in cities, digital on-demand mobility platforms that move people and goods involving a significant degree of automation have become one of the most captivating objects of analysis over the past decade. A staggering volume of research by geographers and others has explored its manifold dimensions—and for good reason. Labour specialists have been concerned about the impact of digital on-demand platforms on worker conditions, explaining how drivers are forced to develop entrepreneurial skills (Pollio, 2019) within the parameters of an increasingly confined system (Barratt et al., 2020). Researchers have described how workers must navigate poor pay and conditions, in both the global north and global south (Giddy, 2021). Though there have been moments when workers have organised to protest these myriad repressions (Wells et al., 2021), much of this research shows how these platforms are intensifying already-existing inequalities, such as through the exploitation of migrant labour (Altenried, 2021), and ultimately reconfiguring class dynamics (Borowiak, 2019). Political geographers and others have opened our eyes to the corporate interests fuelling platform growth (Stehlin et al., 2020), how platforms accumulate power (Howson et al., 2022), and increasingly drive urban planning dynamics (Söderström and Mermet, 2020). In this regard, on-demand platforms are part of the increasingly automated governance of cities (Wiig and Masucci, 2020), though their relationships with jurisdictions are contingent on a wider flux of forces (Katta et al., 2020). Urban geographers have charted how on-demand digital platforms are centrally involved in changing the experience of everyday life in cities (Barns, 2019; McNeill, 2021) through the way that they increasingly control everyday urban interactions (Graham, 2020).

Yet despite this burgeoning field, much less has been said about the role of bodily feeling in increasingly automated systems of on-demand mobility. Though feeling is certainly implicated in some of the above work on automated mobility, and is slightly more visible in some work than others (such as in research that includes in-depth fieldwork with workers. See, for instance, Popan, 2021), the overwhelmingly political economy preoccupation means that feeling rarely features as a primary object of analysis. Where feeling is implicated, it is often treated as a secondary side effect. This is a very strange absence, given how the modulation of feeling is increasingly recognised to be a vital realm of politics (Thrift, 2004a). As cultural geographers have stressed, far from a personal, interior realm, feeling—understood here in terms of how we are affected—is about how practices, events and encounters register in sensing bodies, shaping dispositions, habits and, ultimately, actions (Anderson and Harrison, 2010). Just as people’s capacities to do things is contingent on a complex host of biographical, circumstantial and structural factors (which also incorporates a host of socioeconomic dimensions), people’s capacities to be affected—to feel—are no less political. In response, this article explores how being affected is a primary site of politics in on-demand mobility. As such, it contributes to a body of work that explores the mutual constitution of the sensorial and the digital in advanced forms of capitalism—work that is being pioneered in the field of digital geographies (e.g. Ash et al., 2018; Lynch et al., 2022) and cultural studies (e.g. Chun, 2016; Karatzogianni and Kuntsman, 2012). It argues that the change in the feeling of using automated on-demand platforms is politically significant. In doing so, the article emphasises the significance of attunement to affect in on-demand platform capitalism (and indeed geographies of automation more broadly), not as an aftereffect that happens once capitalism has had its way, but a primary site of politics—a site where lines of power are being negotiated and redrawn. In short, rather than exploring a body’s variable active capacities to do, as much work in geography on affect does, this article explores how being affected, or a body’s passive capacity to feel, is an underappreciated mode of power (Bissell, 2022; Connor, 2019).

Disciplinarily, the article extends a nascent body of geographical work on on-demand digital platforms, which has drawn on cultural geographies of affect and embodiment (Koch and Miles, 2021; Leszczynski, 2019) to explore the kinds of collective affections that arise using on-demand platforms. Though it is important to acknowledge the diverse, situational and ambivalent ecologies of feeling of individuals who use these platforms (Bissell, 2020), in this article I suggest that feeling is currently being shaped in a more collective manner. Accordingly, it explores why particular structures of feeling are materialising in this present (cf. Williams, 1977), the kinds of material encounters that give rise to them, and it speculates on what these structures might be and, crucially, what these structures of feeling are doing politically. Accordingly, the article takes as its object of analysis the changing structure of collective feelings in relation to on-demand platforms. These are collective passive affections, of being affected in a distinctive way. The specific structure of feeling that this article explores is disaffection, responding to a series of empirical lures that evidence this disaffection. It argues that disaffection—albeit differently felt by both consumers and workers—is compromising the sense of automation that on-demand platforms require. My argument is that disaffection is working to effectively deautomate on-demand platforms through the felt sense that platforms are not working as they did. This is a durational politics where the time of the delay forces these breakdowns into view. Instead of understanding deautomation only through its political economy as a structural process of technological failure or retreat, proper acknowledgement of the affective force of disaffection demands an understanding of deautomation that accommodates its thoroughly embodied dimensions. The article concludes by speculating on the potential of this moment as a kind of opening: a site of struggle, a space of contention, with the possibility for platform urbanism to be otherwise.

The article is underpinned by four broader commitments. First, and ontologically, a commitment to more fully appreciating the significance of affect and embodiment in conceptualisations of automation. Second, and disciplinarily, a commitment to developing geographical conceptualisations of automation that are informed by disciplinary legacies. Third, and ethically, a commitment to appreciating how automation’s futures are fully contingent on evolving circumstances, thereby countering path dependent or technologically determinist conceptualisations. Fourth, and politically, a commitment to moving beyond didactic descriptions of automation towards doing something with our descriptions, to destabilize, to think differently.

Circuits of automation in the gig economy

A fundamental quality of digital on-demand mobility platforms is that they involve a significant degree of automation over previous ways of moving, consuming and working. Automation is centrally implicated in the algorithmic management of drivers (Rosenblat and Stark, 2016) as well as the consumer experience of using an app (Lin et al., 2022). Automation is also at the heart of the future aspirations of these platforms. Some platform companies are exploring possibilities for automating social relations by incorporating social media information (Cox, 2019) and even aspiring to develop fully automated vehicles (Bissell, 2018). These insights contribute to a wider body of work within geography that has charted the socio-spatial transformations implicated in different forms of automation (Attoh et al., 2021). Much of this work has explored how ‘smart’ technologies are producing new spatial relations (Thrift and French, 2002), building on earlier research in the discipline that interrogated the impacts of increasing work mechanisation (Gertler, 1988). A central preoccupation of this geographical work continues to be the impact of automation on workers, by exploring changing labour requirements and opportunities for resistance (Ellem, 2016) as well as the kinds of skills required by workers (Richardson and Bissell, 2019), and ultimately the changed subjectivities of workers (Bear and Holloway, 2015). More recent geographical work has tried to destabilise the very object of automation itself by exploring how different forms of automation connect across multiple urban domains (Macrorie et al., 2021), and is something that is in flux and being reshaped (Lin, 2022), giving rise to new ways of perceiving space (Pink and Sumartojo, 2018).

Yet despite the proliferation and pluralisation of work on automation, zooming back to on-demand platforms, from an ontological perspective automation is conceived of rather narrowly. The dominant way that automation is understood is by how tasks previously undertaken by humans (such as dispatch services in relation to taxis) have now effectively been automated by platforms which displace the human operatives (Lynch et al., 2022). Proprietary black-boxed algorithms automate the assignment of drivers and routings, provide orders for restaurants, and real-time information for consumers. However, ontologically, there are different ways of imagining the automations at play in on-demand mobility platforms involving different material ecologies. Where flow and connectivity have been dominant metaphors used by geographers to imagine automation, I suggest that these different logics of automation can be usefully imagined through the metaphor of the circuit. The metaphor of the circuit is advantageous because it emphasises how automation involves both dynamic feedback loops between different components as well as a temporal dimension of change through repetition (cf. Thompson, 2010). There are two circuits that are significant for this article that highlight two different forms of automation at play.

First, there is technical automation. This is the dominant way of imagining automation in on-demand platforms. The focus here is on the digital computational materialities that undergird these new technologies and how they are implicated in the management of experience. Much of the attention here has been devoted to exploring the algorithmic logistics that organise objects and exploit bodies (Pollio, 2021; Richardson, 2020), and consequently, the kinds of resistance that can push back (Moore and Woodcock, 2021). This dimension of automation can be described in terms of technical circuits. However, even though a consistent refrain from geographers and others is that automation is socio-technical (involving all kinds of situated decision making, managing, negotiating, reckoning etc.), the complexity and richness of bodily experience remains underappreciated. The opening contention of my argument is, therefore, that an enhanced understanding of automation’s socio-technicity requires a better appreciation of a second kind of automation: bodily automation. These are the bodily habits of acting and feeling—affecting and being affected—that are part and parcel of engaging with on-demand platforms for both workers and consumers. As once-novel practices get repeated over time, they become sedimented and fall below the threshold of conscious attention, becoming incorporated as bodily automations (Dewsbury and Bissell, 2015). This form of automation receives significantly less attention from platform researchers and, if acknowledged at all, is treated in a somewhat determinist manner as a site that forces of platform capitalism merely acts on. This dimension of automation can be understood in terms of embodied circuits. These two forms of automation are fundamentally connected: technical circuits give rise to new embodied circuits. However, and crucially, changes in embodied circuits feed back to impact on technical circuits, potentially changing the future of automation. From a disciplinary perspective, a sensitivity to place-based processes means that geographers are well positioned to consider these how technical circuits of automation relate to embodied circuits of automation.

Automation’s seductions through structures of (un)feeling

Unlike some other disciplinary approaches, geographers insist that automation happens somewhere (Macrorie et al., 2021). It is this place-based focus that gives geographers a distinctive insight into how automation generates specific structures of feeling that arise through the experiential textures of urban life. Raymond Williams’s (1977) concept of structures of feeling is useful here for two key reasons. First, it moves beyond individualised accounts of affectivity by helping us to understand how our bodily capacities to sense and feel are shared between people. In this regard, though individuals will respond to situations in ways that are contingent on their own biographical context, Williams’s account encourages us to think about the general epochal undercurrents of feeling that are collectively experienced. Second, it draws attention to how our affections—our bodily capacities to sense and feel—are not universal and unchanging. Rather, our capacities to sense and feel are contingent on our historical and cultural milieu. In this regard, it helps for understanding how particular technologies, such as developments in technologies of transport, can shape our affections. Though geographical distinctiveness is not stressed in Williams’s account, his foregrounding of the contingency of milieu also heavily implies that structures of feeling have a geographical specificity. As such, there is a uniqueness to how these feelings have emerged in fieldwork in Melbourne and we cannot unproblematically assume that they would be similar in other urban contexts (cf. Cirolia et al., 2023). Following from this appreciation of geographical context, in the wider project that this article is based on, I was interested in getting a better understanding about how using on demand mobile platforms makes the city of Melbourne feel differently for a range of different people. For both consumers and workers, I wanted to learn how using on-demand platforms might redraw people’s felt sense of place—in terms of distance and proximity, changing accessibility, possibly opening opportunities to do things differently (Rosenblat, 2018)—as well as felt sense of time. I was also keen to understand how, in turn, repeated engagement with on-demand rideshare for consumers, for instance, might change the felt sense of other modes of transport.

To sense these transformations, between 2017 and 2020 I conducted in-person interviews with 30 consumers and 30 workers drawn from two surveys advertised on social media in Melbourne—a city that has experienced significant growth in digital on demand platforms over the past decade. A research assistant, Elizabeth Straughan, undertook a further 10 interviews virtually with workers during the pandemic lockdown in 2021. I also did interviews with 20 people involved in the governance of on-demand mobility. Over the past seven years, I have also kept abreast of the wider background ‘hum’ (Thrift, 2004a: 87) of on-demand mobility in Melbourne through viewing online message boards and news media. Aligning with the conceptual framing of this project, the choice of people to interview was not based on getting a representative sample, rather it was about maximising a diversity of experiences in terms of the demographic and geographic indicators as well as intriguing long-form text responses in people’s survey responses. Rather than imagining interviews as uncomplicated windows onto the lifeworlds of others, these interviews were understood as encounters in their own right, with all the challenges of comprehending expression that come with that. As such, rather than excising a series of exemplary quotes, analysis and presentation in this article reflects on these interview encounters (see Bissell, 2023). This involves reflecting on how the sensory dimensions of on demand became refracted through different forms of spoken expression and how these expressions affected me in different ways.

I argue that beyond just the spatial reconfigurations that have begun to be explored in terms of people using different services, there is also structure of feeling integral to automation itself—the feeling of an automated infrastructure that has emerged in this historical moment in Melbourne, to draw on Williams’s emphasis on the situated nature of affectivity. Throughout the interviews, especially the earlier ones, ease was a recurring motif in various incarnations. What perplexed me initially was how many consumers found it challenging to describe their use of rideshare precisely because it was so effortless. People talked about when during the day they would use it, and where they would travel to and from, and they reflected on the financial ramifications of using it. But when prompted to convey the actual bodily experience itself in terms of what they would do and how they felt, responses were often truncated and unremarkable, leaving me feeling deflated that I had somehow not managed to get to the bottom of things. Tap, wait, move, get to destination. What more is there to say? For some, I sensed how ease was expressed in terms of dampening undesirable sensations. Using rideshare became a spacetime wormhole: a way to get somewhere quicker when you are in a rush, a personal valet, a way of avoiding the discomfort of public transport on rainy days, a way of avoiding drunks on trains at the end of an evening, a way of avoiding turning up to meetings sweaty, a way of still getting to work on time if you slept in by mistake. For others, this ease made other previously effortless things feel more difficult, like waiting for a bus, or even going through a drive through and having to place a food order verbally. Such simplicity sometimes had other difficult side-effects. I sensed anxiety from some about how the ease of ordering food on a mobile phone meant racking up huge bills, for university students studying all-hours where food preparation is a time suck and an inconvenience, where pressing a few buttons to sate an appetite was a far easier option. I sensed how the ease of food delivery could easily spiral out of control, prompted by tantalising app notifications, and I learned about the kinds of self-discipline that had to be mustered to rein in these habits. For workers too, ease manifested through surprise at just how simple it was to sign-up to become a delivery driver relative to other jobs they had done previously, and how straightforward the app was to use.

These expressions that dominated earlier pre-pandemic interviews, I think, hint at a specific structure of feeling that on demand promises. Though this collation of reflections inevitably conceals a diversity of experiences, what is common to each is a feeling of ease and convenience. Yet such an ostensibly simple term does not sufficiently grasp the reconfiguration of sensation that automation is doing here. Where the historical contingency of Williams’s concept of structure of feeling certainly appreciates that capacities to feel can be transformed, his account is less forthcoming about exactly how this might happen. In response, we can turn to habit literatures in geography that shed light on embodied automation to help imagine how this feeling of ease comes about materially. Geographers have explained how ease, rather than being thought of in an essentialised manner, can be understood much more contingently in terms of the removal of felt resistances relative to other experiences of mobility and consumption (Sharpe, 2013), where a reduction of effort through using on demand apps translates into a feeling of grace (Ravaisson, 2008). As indicated in the stories I heard, platform apps are central to creating these experiences. For rideshare services, for instance, a streamlined, pared-down interface is designed such that users can hail cars, select their destination, and pay for their ride, removing the need for finding a local taxi number, phoning a dispatch office for a cab, or negotiating a financial transaction with the driver. The interface produces a bodily sense of ‘on-demandness’ which attempts to minimise having to wait, and a sense of seamlessness that opens up possibilities to travel anywhere in any city hooked into the app at the touch of a button. For workers, sign up is easy, intuitive even (Van Doorn et al., 2020).

Reflecting on the interview encounters, what is so interesting about this structure of feeling is its non-intensity. Ease is not contingent on especially joyful affects associated with feeling good. Few participants talked in excitable ways about the joys of using rideshare or ordering a cooked meal, for instance. In this respect, the affectivity of convenience seems subtractive. Rather than producing new positive feelings, positive here in the sense of there being something present, the snippets above overwhelmingly point to how engaging on-demand apps works to reduce a range of sensations associated with more laborious ways of moving, consuming or working. So rather than producing ‘intensities of feeling’ (Thrift, 2004b), automated convenience appears to be desensitising. Indeed, to push Williams’s concept further in this regard, we might call this a ‘structure of unfeeling’. The temporal dimension of convenience is important here because it is a structure of (un)feeling that is contingent on the absent presence of other (relatively) less comfortable ways of moving that would register more strongly in sensation. The idea of automated convenience being subtractive at the level of feeling has various antecedents through histories of time-saving and labour-saving technologies (Jackson and Viehoff, 2016; Shove and Southerton, 2000). At the level of practical actions, the affectivity of convenience can be traced back to Taylorist imperatives developed in the earlier part of the twentieth century for cutting out superfluous movements (and therefore resistances) in practical tasks (Cresswell, 2006). Though a central argument for geographers of mobility has been to consistently point to how movement is imbued with meaning and significance rather than a displacement from A to B, the non-intense affects associated with ease and convenience take us back to the ideal of displacement and the removal of any other experiential textures that might get in the way.

Though many of my interviewees talked about ease in broadly affirmative ways indicating the desirability of removing resistances, curiously, this structure of (un)feeling has often been viewed by critical social theorists in a much more pejorative manner. Lin (2022), for instance, explicitly politicises the affectivity of ease of automated on demand platforms through the notion of being ‘lulled’. Diagnosing this structure of (un)feeling, Lin writes

By changing the rhythm of people’s routines, and infiltrating the spaces of everyday consciousness through apps and social media, an automated infrastructure composed of platforms has arguably begun a work of socializing subjects into a new lull of self-service transacting. (2022: 474)

Here, to be lulled captures precisely this sensory quietude, a sense of subjects who are not in possession of themselves, potentially sleepwalking into trouble. Certainly, this fear of being carried away by the ease of automated infrastructures is reflected in interview reflections above, where people told me about having to learn self-discipline to moderate their on-demand habits. Other thinkers are more explicit in their concern that this ‘lulled’ structure of (un)feeling is giving rise to worryingly impotent forms of subjectivity. For thinkers like Stiegler (2018), automation’s lulling represents a decline in active affections—capacities to act—through deskilling. An example of this from the vignette above might be the participant who told me how ordering something through a drive-through suddenly felt challenging after becoming used to the automation of the app. For others, automation overly insulates subjects from the myriad encounters that would give rise to a more ethical sensibility, removing opportunities for passive affections—capacities to feel (cf. Sennett, 1996). Yet in denouncing ease and convenience, these thinkers could in turn be roundly critiqued for an evaluation that is ignorant of the immense burdens of social and economic reproduction that many of my participants clearly shouldered, where ease and convenience provide a small crack of respite.

A more compelling line of critique aimed at the lulling or seductive dimensions of the structure of (un)feeling of automation involves the power geometries that exist between consumers and workers at the level of feeling. As Baudrillard (1990) reminds us, seduction involves an interplay of revelation and, crucially, concealment. Sustaining these feelings of an automated logistics infrastructure requires a ‘standing reserve’ of low paid, insecure, ununionized labour that is concealed when engaging with apps. As a business, the generation of profit for rideshare platforms such as Uber is reliant on a vast insecure ‘reserve army of labour’, to use Marx’s (1976) term. As others have consistently pointed out regarding other automated technologies, automation often requires a human presence, but this is variously offshored, backgrounded, or concealed from view—and it is this concealment that sustains the feeling of automation for consumers, such that it becomes a generalised background ‘structure’, in Williams’s sense (Irani, 2015; Suchman and Stacey, 2012).

Sites of disaffection

Lulling has often been used as a paranoid diagnosis for our times. It has often been a way for critical theorists to rise above the melee and reprimand others for being led astray by untoward powers against their will or better judgement. Though the notion of being lulled into the use of automated on-demand services certainly bears out at times in fieldwork as described in the previous section, to diagnose this as a dominant structure of feeling at play would be unfaithful to the breadth of experiences documented. There is a wealth of ethnographic research emerging with consumers and workers that paints a much more variegated affective terrain. Far from just unthinkingly lulled, pacified and desensitised, this research highlights how consumers wrestle with the ethics of on-demand services (Bissell, 2020), actively deliberating on how, when and why they use these services. The breadth of work on workers is wider, bringing to light all kinds of bodily intensities that are endured by delivery riders and rideshare drivers, for instance (Badger, 2021; Popan, 2021). Appreciating the range of intensities that are bound up with moving and consuming has been especially guided by work on mobility that directs our attention to the practical, embodied aspects of daily life. Attention to shifting atmospheres (Simpson, 2017) and transitions of bodily (dis)comfort (Kent, 2015) helps to fragment blanket diagnoses of lulling by opening our eyes to a much more patterned terrain of experience.

Yet, conversely, to highlight only this variegation and diversity can obscure some of the commonalities that thread through experiences. This is what is so powerful about Williams’s (1977) concept of a structure of feeling and why it is useful for our analysis here. Though experiences are of course being felt differently, not just in terms of the differences between workers and consumers, but also in time, the analytical strength of thinking in terms of structure of feelings is that they can accommodate all manner of ambivalences (cf. Ruez and Cockayne, 2021), yet retain distinctive patterns. My argument here is that disaffection is currently a dominant structure of feeling being experienced in the on-demand platform economy in Melbourne. Though there are multiple ways that disaffection has been understood, common to each is a sense of separation. Disciplinarily, thinking about separation is a timely rebuttal to the radically connectivist nature of so much geographical thinking over the past couple of decades that foregrounds different forms of relationality at the expense of non-relations of distance, detachment and rupture (Rose et al., 2021). In this section, I outline how disaffection has been understood by geographers and others in three distinct ways, before reflecting on the experiences of disaffection that thread my fieldwork.

A first way that disaffection has been understood is as a diagnosis of alienation. This is perhaps the most conventional and historical understanding of the term and has its roots in Marx’s writings on how the capitalist system produces a sense of alienation in workers by separating them from the means of production and the products of their labour. This sense of separation then creates feelings of powerlessness and dissatisfaction. Within geography, this understanding of disaffection has been broadened to diagnose particular social groups as being left out or left behind (Loughenbury, 2009), or lacking voice or participation in political processes (Kraftl, 2013). These understandings aggregate disaffection as a collective experience faced by young people, for instance. Of relevance for the themes of this article, this aggregate understanding of disaffection as alienation has also been deployed to consider the effects of new digital technologies. For instance, Stiegler (2012) argues that disaffection is the inevitable end point of the loss of expectation in hyper-industrial society. His argument hinges on the idea that digital technologies have exacerbated a ‘proletarianization of the senses’ for workers and consumers alike, separating them from real opportunities to sense and participate in the making of society . Some of the problems of an aggregate understanding of disaffection are highlighted through work on algorithmic governance where particular kinds of classed and racialized subjects or specific working class, ethnic-minoritized urban areas are diagnosed as disaffected and have been subjected to intensified securitisation (Leszczynski, 2016).

A second way that disaffection has been understood is as a separation from feeling. This has mostly been developed through thinkers who draw on theories of affect to consider experiences where a capacity to be affected by an event, person or situation is closed off in some way. It is therefore a more situational and practical use of the term than the epochal and aggregate definition developed through Marxist interventions. This understanding of disaffection has often been drawn on in pejorative ways, for instance to point to how repeated encounters of images can dampen their capacity to shock (Pedwell, 2014) or kindle sympathy (Moeller, 2002). However, more recently, disaffection as incapacity to be affected has been developed more affirmatively as a tactic of empowerment, especially for marginalised or traumatised people, through a sense of being unavailable (Berlant, 2011). For Yao, disaffection understood as unfeeling offers a ‘break from affectability’ (2021: 5) as a tactic ‘from below’ with insurgent potentials, such that the Other does not yet again become conscripted into (colonial) regimes of legibility and recognition—and therefore subordination. In the context of the on-demand economy, disaffection as unfeeling might be a practical tactic to get by, a way of anaesthetising the pain that comes from being forced to inhabit a painful situation (Bissell, 2022).

A third way that disaffection has been understood is through the idea of bad feeling that separates a body from its powers. Though there are elements of this in Marxist notions of alienation, geographers and others have developed a more nuanced and dynamic understanding of this kind of disaffection. Much of this work draws from theories of affect via Deleuze’s reading of Spinoza where a body is understood in terms of its power to act (active affections) and feel (passive affections). Passive affections are either joyful or sad, and it is an ethical priority to avoid or transform the bad feelings of sad affects as they ultimately diminish a body’s power (Dekeyser et al., 2023). Yet geographers have been keen to expand this affective calculus beyond categorical polarities of good and bad (more evident in Marxist thought), through an appreciation of the range of more ‘minor’ feelings and situational ordinary affects that do not easily align with a Spinozist evaluation. Here, the ‘bad feelings’ of disaffection might bring to the foreground an ‘oscillation between forms of attachment and detachment’ (Wilson and Anderson, 2020: 593), or even indicate a host of more enabling attachments, such as the restlessness of boredom (Anderson, 2021).

Here, I am keen to develop this third way of understanding disaffection in terms of bad feeling and what it does in the context of on-demand platform mobilities. In my fieldwork, there was no shortage of bad feeling being relayed to me. I witnessed expressions of bad feeling attributable to a host of different triggers, from uncomfortable experiences to more generalised feelings of shame and despondency caused by underpayment for workers. Indeed, this wider hinterland of bad feeling is beginning to be mapped out by others in the field (Petriglieri et al., 2019). In this article, however, I am interested in thinking about one specific kind of trigger for disaffection: the delay. My choice of the delay is because this trigger runs counter to the optimisation promised by the automated structures of (un)feeling associated with seamlessness and efficiency outlined in the previous section. Rather than dampening intensities through lulling, I am interested in the heightened intensities of bad feeling that are prompted by the delay for differently situated people in the on-demand economy, and I am interested in what these do to politics in this space understood in terms of differential enablement and constraint, and the kinds of conflicts and contestations that result.

Deautomation

In light of this diagnosis of disaffection, in this section I return to the concept of automation to ask: what does disaffection tell us about the changing nature of automation in the context of digital on-demand mobilities, especially in post-lockdown Melbourne? To what extent might disaffection change the future of automated digital on-demand mobilities in this city? Does it hold incipient potentials for socially progressive change? Or, alternatively, is it just a peripheral ‘side affect’ that will become accommodated in urban life? Previously, I explained how automation in the context of on-demand mobility can be understood in terms of two interlocking circuits. Technical circuits involve the digital materialities of algorithms, interfaces and digital devices that automation works through. Embodied circuits involve the repetitious practices of engaging with these on-demand platforms, and the formation of habits. Where significant attention has been given to figuring out what is going on in terms of technical circuits, much less has been said about embodied circuits. Yet, crucially, both are involved in automation, since it is the embodied circuits that help both consumers and workers to engage with these platforms such that they can occur below the threshold of conscious attention. Attention to these embodied circuits help us to understand the affective dimensions of automation, in terms of how interacting with these platforms gives rise to a particular ‘automated’ structure of feeling. Through reflections on fieldwork from consumers and workers, especially in the early days of my fieldwork, I indicated how the non-intense affect of ease becomes the dominant structure of (un)feeling of automation. This structure of feeling is also baked into the promotional logic of platforms for both consumers and workers, promising the ease and convenience that contrasts with other more laborious ways of travelling and working.

My argument is that disaffection is the feeling of circuits of automation coming apart. Disaffection is how the breakdown of automation registers in bodily sensation. Reflecting on the vignettes in the previous section, if we were to focus on just the technical circuits of automation alone, we might surmise that these automated systems are still working exactly as intended. Cars are still being dispatched, and riders are still being sent to collect meals from restaurants. However, if automation involves both technical and embodied circuits, then adding in the other side of automation—the embodied circuits—this is where breakdowns in automation are being felt. Though there are myriad moment-to-moment experiences of disaffection that punctuate the everyday life of on demand mobility, my choice to focus on the delay is because the delay introduces a gap in these bodily circuits of automation. Rather than being ‘lulled’ (Lin, 2022), where a practice such as ordering a rideshare mostly takes below the threshold of conscious attention, the delay shakes bodies out of this lull, forcing previously backgrounded habits back into conscious attention. The disaffection of the delay can be understood as a negative affect, understood here in terms of the diminution of capacities to act. The delay introduces a feeling of ‘obstructed agency’ (Ngai, 2005: 3), where a sense of not being able to do something is felt. Rather than the negation of affect (Yao, 2021), through my fieldwork I learned how disaffection is being felt through an array of more intense affects united by their unpleasurability. For many consumers in Melbourne, digital on-demand platforms are no longer being felt as pleasurable, graceful, and easy; rather they are being felt as depletion, impediment, and a resistance. Where the feeling of automation may have previously obscured the presence of workers in conscious attention, deautomation thrusts these workers back into the spotlight: negative affects are felt because of this driver cancelling, or the felt absence of drivers nearby. For workers, the deautomation of the delay might make the depleting dimensions of their jobs register more sharply in bodily sensation. In short, where on-demand might previously have been felt through an ‘automated’ sense of ease, my suggestion is that these resistances are being felt as deautomation—circuits of automation coming apart. Where a political economic definition of deautomation would view it more narrowly as a structural process of technological failure or retreat, thereby implicating only technical circuits of automation, proper acknowledgement of the affective force of disaffection demands an understanding of deautomation that accommodates its thoroughly embodied dimensions, through disruptions to embodied circuits of automation.

Of course, a foil to this argument is that circuits of automation never work perfectly—though this is the conceit that much writing on automation’s failures responds to. As Sterne (2021: 32) says, ‘nothing works exactly like it is supposed to—machines, media, models, conversations.’ However, my understanding of deautomation differs from other scholarship that has explored automation breakdown. Though it might share some of its political capacities, it’s not a glitch—a ‘small-scale disorientation’ (Leszczynski and Elwood, 2022: 1)—but rather a more fundamental undoing of technical and embodied circuits. Where I have previously explored automation breakdown as a result of plenitude—excessively complex material environments (Bissell, 2020), here, automation breakdown is about delay and insufficiency. Here, delay, waiting and non-performance is about something not happening. The important point is that it is the feeling of this non-performance that matters. Though infrastructural breakdown has received lots of attention by geographers and others (Graham, 2010), bodily feelings are often overlooked in favour of focusing on the pragmatics of what happened and why, and the macropolitical forces that such breakdowns reveal. Where feeling is invoked, it is often a way of highlighting the magnitude of a situation and how it impacted on people, rather than a primary force capable of transforming situations (Massumi, 2015). So: what kinds of transformations might these feelings of deautomation be giving rise to? What alternative futures might be discernible that were previously unthinkable? And, importantly, under what conditions might deautomation be desirable and for whom?

I frame my closing speculations around two contentions: first, that deautomation is doing things, but that its effects are being differently experienced by workers and consumers; and second, in evaluating these differential experiences, there are both repressive and recuperative dimensions of deautomation. For workers, it might be tempting to question whether these transitions in feeling of deautomation even matter. Geographers have written about how other workers who are at the mercy of repressive systems of management and control are likely experiencing related forms of disaffection (e.g. Kanngieser, 2013), but these feelings might do very little to effect change to working conditions. If automation is indifferent to disaffection, then these transitions in feeling experienced by workers become just another symptom of this work that must be managed. Furthermore, declining service standards are not a surprise given how hyper-capitalist digital platforms such as Uber are based on the construction of a huge two-sided market and its associated network effect, creating an oligopoly that can be sold on to public sharemarkets (Srnicek, 2017). Even more darkly, deautomation as an outcome of insufficient numbers of workers to keep platforms operable could potentially exacerbate the onset of urban futures where platforms double down on reautomating platforms by progressively removing their reliance on human labour, such as through drone delivery and fully automated vehicles. Of course, the counter argument is whether this would ultimately be a good thing from the point of view of workers, minimising what are ultimately poor-quality jobs in the eyes of many (Myhill et al., 2021; Bastani, 2019; Straughan and Bissell, 2022).

On the other hand, however, the transitions in feeling of deautomation also have the potential to be recuperative for workers. Though the reading of negative affects in geography popularised through Deleuze’s reading of Spinoza firmly denies that negative affects—such as disaffection—have any positive role to play in the constitution of subjects (cf. Dekeyser et al., 2023), other thinkers invite us to think about the operative role of negative affects in more affirmative ways. Though hailing from a very different context, Malatino (2022) describes the impossibilities of expunging ‘bad feeling’ for trans people, even within moments of joy or euphoria. However, they also point to how this bad feeling might work in more productive ways, to create collectives and intensify shared bonds, making life more bearable. Malatino’s thesis follows others who have argued for the potentially recuperative aspects of disenchantment (Berardi, 2017), and the potential productivity of bad feelings (Ahmed, 2010). This channelling of bad feeling into affirmative action might, in this context, manifest in worker strikes and organised forms of protest. Though bad feeling could just as easily tip people into situations of negative solidarity, where workers turn against other workers who they perceive as a threat (cf. Bissell, 2022: 98–99). Indeed, as Crary (2022: 23) argues, ‘gig economy workers have little to share with each other but their destitution and exhaustion’. In more minor ways, we might speculate on the existence of small kinds of pleasure derived from knowing that drivers are more in demand. In a kind of inversion of the power geometries between the powerful consumer and subordinate driver, consumers feel disaffection as a result of workers having more choice about which rides to take. Indeed, an extension of this logic could be that deautomation of this part of the digital on-demand platform economy might mean that platform workers finally get increased pay and conditions in a bid to retain them, to ‘reautomate’ the platform, to reduce the delay. Even in the absence of such a breakthrough, bad feeling for workers might just make their investment in on-demand platforms even more temporary than they might otherwise have intended, opening other horizons with more promise.

For consumers, there may be a similar double-sidedness to these transitions of deautomation. More pejoratively, just like investments in other areas of consumerism, such as clothing, where the registration of bad feelings (about provenance, for example) might not override the good feelings derived from the item itself, it might be that such bad feelings just become incorporated as part of the warp and weft of what it is to move and consume in the city. Indeed, these bad feelings that come from the breaking up of bodily habits integral to automation might themselves become habitual, over time registering less in sensation and so becoming easier to bear. In this regard, Williams’s (1977) differentiation between residual and emergent structures of feeling might help to not only appreciate such transitions in disaffection over time, but also how structures of feeling such as disaffection might coexist with other less dominant structures of feeling. Furthermore, disaffection might represent an intensification of the ugliest aspects of consumerism, where the feeling of disaffection serves to reveal an entitlement to feel automation in a particular way, to be serviced on demand. Even more cynically, Ngai (2005: 4) reminds us how negative affects like disaffection might represent less a systemic breakdown—in this case, an automated system—and much more an opportunity for capitalism to mutate further. She recognises how ‘functional such affective attitudes and dispositions have become, as the very lubricants of the economic system which they originally came into being to oppose.’

However, the feeling of deautomation for some consumers might also prompt more progressive responses. The gap introduced by deautomation might open up a space to more consciously question how to move and to consume differently. At the very least, deautomation might prompt a variety of smaller creative workarounds where people ask themselves: how else can we do this? But in the longer term, deautomation might also prompt consumers to abandon these lean hyper-capitalist platforms, making them even more unprofitable, perhaps even hastening their demise. Indeed, it might be possible that deautomation could force new habits of urban inhabitation that sees on-demand platform consumers calling the local taxi firm or returning to other modes of (public) transport.

Conclusion

It is hard not to revel in automation’s failures. As academics, we tend to rejoice in those deautomations where the dreams of techno-boosterism wobble or turn to dust—not because we harbour some atavistic yearning for analogue nostalgia, but because we are highly attuned to the myriad oppressions that contemporary automated systems induce. We take pleasure in that the disruptive platforms are now themselves being disrupted. Disaffection points to the limits of platform companies’ power to modulate these logistical systems (Richardson, 2020)—for instance through the management of supply and demand through dynamic pricing—now that their limitless standing reserve of labour is much less assured. Intriguingly, and perhaps more significantly, this diagnosis contrasts with Lin’s (2022) argument that COVID-19 has intensified both the reach and efficacy of automated systems. Disaffection as a public feeling matters here because it does something—it is not a residual aftereffect.

Yet Appadurai and Alexander (2020) warn us that failure no longer has the subversive potential it might once have done. In a searing analysis of how failure has become increasingly financialised, they contend that to celebrate technological breakdown buys directly into the capitalist logic of creative destruction. They remind us that innovation is utterly contingent on failure, and in turn, failure is turned into a commodity. We can see this in the way that the time of the delay is increasingly monetised by companies such as Uber, where if you pay more for a more premium service, you minimise the delay (see also Graham, 2005), so that waiting becomes a condition of subordination. They also remind us that the success of apps such as Uber are also contingent on a more bodily kind of failure, since the apps are at their most successful when users ‘fail to control themselves’, spending money they don’t have (2020: 58).

Connecting with broader diagnoses about the ‘disabling of memory’ that is brought about by the capitalist internet complex (Crary, 2022: 8), for Appadurai and Alexander (2020), critical to the sustenance of the on-demand automated economy is a collective forgetting of these failures. Experiences like internet buffering represent what they call a ‘habitual delay’, but such experiences could easily extend to include the kinds of waiting traversed in this article. Crucially, the problem with the habitual delay is at the level of affect. They claim that we are left unaffected and unchanged by the habitual delay. Where a ‘habitual failure’ is experienced, we might feel momentary frustration, but this is quickly forgotten. It is this forgetting that is pivotal– in fact, Appadurai and Alexander claim we’re habituated to forgetting the memory of the last harm ‘so as to be innocent and hopeful when we experience the next one’ (2020: 123).

However, my argument is that these failures do stick. If we take an understanding of affect that appreciates how all experiences accrete in sensing bodies that can build over time to tipping points, then we should not so easily dismiss the operative role of disaffection in changing the course of (de)automation. Where Appadurai and Alexander’s plea is that our salvation relies on us remembering failure, my argument is that we do remember failure, and that the disaffection is doing things in the here and now at the level of affect to unravel the gig economy. My hope is that the kinds of deautomation explored in this article present an opportunity for governments and consumers to pause and think more seriously about the kinds of more socially progressive alternatives that could counter these deleterious forms of platform capitalism.