The social lives of network effects: Speculation and risk in Jakarta's platform economy

Introduction

The last decade has heralded a global re-organization of relationships between digital technologies, networked infrastructures, data collection, and the urban environment. On the heels of municipal and private “smart city” interventions (Datta, 2015; Shelton et al., 2015), cities around the world are now grappling with the social and economic implications of platform firms like Airbnb, Amazon, Uber, Deliveroo, OYO, WeWork, and Bird. In particular, ride-hailing platforms such as Uber, Didi Chuxing, Ola, and Grab have upended existing systems of urban mobility. Leveraging proprietary geo-located data, ride-hailing companies algorithmically match real-time passenger demand with a supply of roving contract drivers to provide a market-based digital platform for urban mobility. Within a relatively short span of time, this business model has attracted historically unprecedented amounts of venture capital, the injection of which has engendered new urban mobility regimes, experimental public–private partnerships, and new patterns of transportation across the globe (Jin et al., 2018; Stehlin et al., 2020; Transportation Research Board, 2016). In turn, this has raised popular and academic concerns over how these companies are impacting congestion, carbon emissions, urban governance, public transportation use, and labor conditions for gig workers (Chen, 2017; Henao and Marshall, 2018; Mazumdar, 2021; Rosenblat, 2018).

Notwithstanding these considerable impacts, most ride-hailing platforms remain unprofitable and face increasing scrutiny about their long-term financial sustainability (Horan, 2017). Just a month before its proposed initial public offering (IPO) in May of 2019, Uber Technologies Incorporated disclosed in its mandatory Securities and Exchange Commission (SEC) S-1 filing that it had lost $1.8 billion dollars in 2018 (Uber Technologies Inc., 2019). Though it made big headlines, this in itself was not surprising. Deploying its venture capital war chest liberally, the company has sustained heavy yearly losses since its launch in 2009. What was truly surprising about the report, however, were its frank statements about the company's potential profitability, given Uber's enormous valuation at the time: “We have incurred significant losses since inception, including in the United States and other major markets. We expect our operating expenses to increase significantly in the foreseeable future, and we may not achieve profitability” (Uber Technologies Inc., 2019: 27, my emphasis). The report goes on to state that their ongoing losses are the result of large price subsidies for consumers and cash incentives for divers, designed to pull more of both user-groups into their network and that, if they are unsuccessful in growing and maintaining this network, the company will eventually fail.

That Uber—once valued at $82 billion—would fold is not such an unimaginable prospect. Tech IPOs throughout 2019 revealed the fragility of platform firms propped up by enormous sums of venture capital but that had yet to turn a profit. In March of 2019, Lyft (Uber's major competitor in North America) had its own IPO, quickly followed by Uber itself in May. Both were sobering failures. Share prices slid below initial valuations immediately after opening, and have underperformed since amidst legal controversies and uncertainty around the impact of the COVID-19 pandemic. Combined with the implosion of WeWork after its own IPO announcement, these developments sent shockwaves through the tech and venture capital industries. There is now something of a reckoning in Silicon Valley about the long-term sustainability of “growth-before-profit” platform startups, with investors concerned about another bubble as the share of unprofitable startups headed to IPO grows once again to dot.com bust levels (Ritter, 2020).

Uber's recent history helpfully illuminates the contours of an historical conjuncture in which venture capital speculation on platform firms and the data they collect has become increasingly influential for urban transformations across the globe. Urbanists have shown how financial speculation on land and real estate has ushered in a “speculative urbanism” premised on rendering urban space an object of global investment through neoliberal governance, transnational policy norms advanced by global consulting networks, and new subjectivities (Fields, 2018; Goldman, 2011, this issue; Humphrey, 2020). The case of Uber, however, forces a consideration of different speculative processes, the object of which is not land, but digital sequences of 1s and 0s. Over the last decade, venture capital firms have bankrolled a global transformation of urban transportation systems, working to reorient residents’ mobility towards data-capture by transnational, private platform firms (Stehlin et al., 2020). Realizing this transformation requires enormous capital outlays, leveraged to attract riders and drivers to the platform, scale quickly, collect more data than their competitors, and thereby solidify monopoly position and collect even more data. As Masayoshi Son—CEO of the Softbank Group and architect of its deep, global investments in ride-hailing through the Vision Fund—is fond of saying, “whoever controls data controls the world” (Pfluger, 2019). Yet, as Uber's recent history shows, the limits and risk-exposure endemic to this speculative city-making project have never been more apparent. Profitability in ride-hailing remains elusive.

Bringing attention to these underexamined facets of speculative urbanism, I ask: If ride-hailing platforms are not yet creating a return on investment, how are the networked connections they create in the meantime repurposed by everyday urban residents? And what might this tell us about speculative city-making in the current historical conjuncture of platform capitalism? Drawing on 12 months of ethnographic fieldwork with ride-hailing drivers in Greater Jakarta, Indonesia, I engage with these questions, showing how three different actors—venture capital funds, platform firms, and gig workers—are all brought together through speculation upon “network effects,” a socio-technical phenomenon in which the more users there are in a networked system, the more useful and valuable that network becomes.

This paper reports on findings from fieldwork in Greater Jakarta (Jabodetabek) between January and December 2019, with an additional month of preparatory work done during July of 2018. Over this period, I conducted 71 semi-structured interviews with: ride-hailing drivers (48); platform firm employees and founders (5); and government officials, legal experts, NGO activists, and transportation experts (18). Interviews with drivers were most often done in a group setting with anywhere between 3 and 10 individuals, conducted in English and Indonesian alongside a local research assistant and interpreter. Additionally, I conducted over 100 hours of participant observation with driver communities (described below) at formal events such as social gatherings, community meetings, protests, and fund-raising drives, not counting less structured time spent socializing and observing at a total of 33 driver community “basecamps.” Interviews and fieldnotes were transcribed, translated, and then analyzed using Nvivo software with a combination of emic and etic coding.

I proceed in three parts. First, I review the existing literature on digital platforms in geography and media studies, narrowing in on what Barns (2020) calls “platform urbanism.” In particular, I highlight recent interventions arguing that this literature has been preoccupied with macro political economic analyses, overlooking the “technological everyday” (Amin, 2007: 109; Barns, 2019; Leszczynski, 2019; Richardson, 2020). In the second part, I review the history of network effects, arguing that while the platform studies literature identifies network effects as essential to the operations and capitalization of platform firms (Langley and Leyshon, 2017; Parker et al., 2016; Srnicek, 2016; Sundararajan, 2016) current understandings remain narrowly focused on the economic benefits that accrue to platform firms. This conceptualization problematically conforms to platform firms’ interests in “framing” (Callon, 1998) network effects as a technical economic externality that can attract venture capital investment, rather than an embodied product of social relationships created and sustained in everyday urban life.

Third, I show the limits of this conceptualization through focusing on how platform architectures—including network effects—are embedded within a constellation of already-existing social relations in the technological everyday: the social lives of network effects. Drivers in the Greater Jakarta region have repurposed connections derived from a shared platform employer to “autoconstruct” a network of their own, for their own purposes, with its own effects (Holston, 1991; Prouse, 2018). Much like venture capital and platform firms, ride-hailing drivers engage in their own speculations upon the value of these networks and seek to mitigate risk through them, though they are positioned very differently in their ability to accumulate from these efforts. Taking inspiration from Simone's (2008: 197) emphasis on the “pervasiveness of speculation as an urban practice engaged in by all kinds of urban actors,” I foreground these everyday speculations and practices of collective risk management amongst platform workers. In conclusion, I suggest that attending to such practices opens up space to reframe platform urbanism beyond its current preoccupation with macro political economic analyses, while also establishing new lines of inquiry for theories of speculative urbanism beyond a focus on land and real estate.

Platform studies and the urban

Over the last decade, social scientists have sought to understand how digital platforms are reshaping social, economic, spatial, political, and cultural relations (Barns, 2019; Bratton, 2015; Fields and Rogers, 2019; Langley and Leyshon, 2017; Leszczynski, 2019; Maso et al., 2019; Rosenblat, 2018; Srnicek, 2016). Through the provision of a software program, application programming interface or web interface, platform companies facilitate connections between different user-groups, usually buyers and sellers, in order to capture data with the intention of realizing a profit. These connections enable social and economic exchange and the circulation of goods and services, with the platform firm acting as an intermediary and rentier. Uber, for example, uses its proprietary data to algorithmically connect people who want a ride with people willing to provide it, charging both user-groups rent (in the form of service fees) while simultaneously harvesting their data.

The data produced through platform intermediation—about consumption, mobility patterns, viewing habits, worker efficiency, and so on—has become essential to contemporary capital accumulation: “platform capitalism.” For Srnicek (2016: 6), the 2007–2008 financial crisis precipitated a new regime of accumulation in which “capitalism has turned to data as one way to maintain economic growth and vitality in the face of a sluggish production sector.” Such a regime incentivizes all kinds of firms to maximize the data they can extract from users/consumers, which can then be analyzed, inter alia, to enhance firm algorithms, refine production processes, micro-target advertising, coordinate resource distribution, and so on. Scholars across media studies, sociology, and geography have outlined how such data-extractive and intermediary platform logics are reshaping labor markets and work (van Doorn, 2017; Wells et al., 2020), surveillance and privacy (Zuboff, 2018), economic regulation (Ferreri and Sanyal, 2018), and cultural production (Maso et al., 2019).

Nowhere are these transformations more pronounced than in cities. Platform firms rely on concentrations of users and workers, exploiting the spillovers of urban agglomeration to scale quickly and link assets and people in new socio-technical arrangements of urban space (Davidson and Infranca, 2016; Richardson, 2020). Critical scholarship has called for a better understanding of these transforming relationships, a research program Sarah Barns (2015, 2020) coined “platform urbanism.” As an analytical lens, platform urbanism centers “emergent, irreducible, co-generative dynamics between platforms and the urban” (Rodgers and Moore, 2018) through tracing the processes of intermediation that link “physical and digital layers of people, networks and urban infrastructures resulting from real-time ubiquitous technology and platforms” (Barns, 2015).

For Stehlin (2018) and Sadowski (2020a, 2020b) these co-generative dynamics are the product of rentiership under capitalist urbanism. In the same way that locational advantage shapes the extraction of ground rent in physical urban space (cf. Alonso, 1964), platforms seek to become the central, monopolistic intermediary for interactions in digital space (Langley and Leyshon, 2017), their business dependent “on the platform becoming a (necessary) intermediary in the production, circulation, or consumption process” (Sadowski, 2020b: 568). In turn, “platform rentiership” (Christophers, 2020) relations shape the city through, for instance, the conversion of housing into short-term rentals for listing on Airbnb (Wachsmuth and Weisler, 2018) or the reorganization of transportation systems via privatization or public–private partnerships between ride-hailing companies and transportation authorities (Stehlin et al., 2020; Transportation Research Board, 2016). Van Doorn (2019) emphasizes that this increasing proprietary control over urban data production has created a new institutional context in which platforms like Airbnb wield significant influence over housing policies and spatial planning in cities across the globe, creating a variegated regulatory landscape of policies for taxation, data-sharing, safety and security, and so on.

These and other scholars (see also Davidson and Infranca, 2016) make important contributions linking urban political economy with critical platform studies. Recent interventions drawing on feminist and cultural geography argue, however, that platform urbanism cannot be limited to such political economic analyses (Barns, 2019; Leszczynski, 2019; Richardson, 2020). These scholars emphasize that it is equally important to begin with the complex and geographically diverse entanglements of people and platforms in everyday life, taking up the range of epistemologies advanced by digital geographers in recent decades (see Elwood, 2020). Richardson (2020) suggests that platforms are not merely a business model; they should be understood as “flexible spatial arrangements” that hold potential for novel socio-technical organization of cities, even more equitable and ecologically sustainable ones (cf. Gibson-Graham, 2006). Thus, Leszczynski (2019: 13) calls for a “minor platform urbanism” (cf. Katz, 1996), looking beyond the “totalizing analytics” of rentiership, class formation, worker control, and so on to recognize platform urbanism as a contingent phenomenon, “open to opportunities for, tactical maneuvers rooted in everyday digital praxes that remake, unmake, and make differently platform/city interfaces.” Encapsulating this shift, Sarah Barns (2019, 7; 2020) argues for an analytical focus on the “technological everyday” to navigate beyond macro political economy in platform studies (Amin, 2007).

Taken together, these scholars advocate for a theorization of platform urbanism rooted in “everyday life” wherein the quotidian practices of residents make and remake socio-spatial relations in the city (de Certeau, 1984; Lefebvre, 1991). This perspective centers the new, often-mundane connections made possible through digital platforms, recognizing already-existing practices of platform intermediation not solely determined by data-extractive or rentier logics. Attention to such practices—of mutual aid, communing, collective survival, and so on—offers insight into how urban residents reconfigure both platforms and the city in ways counter-hegemonic to platform capitalism (Leszczynski, 2019, see also Gibson-Graham, 2006). In Barns’ (2018) terms, this is “platform urbanism as a mode of ‘everyday’ urban intervention, a site of urban spatial practice, disrupting smartphone media ecologies through collaborative, site-specific media interventions in the everyday spaces of the city, calling into question the valorisation of urban data as a way of knowing the city.” In this paper, taking up these recent calls, I examine how Jakartans tactically repurpose platform intermediation for their own ends by building networks of mutual aid, remaking their lives, relations, and city in the process. I do so through the lens of an especially powerful organizing concept in platform studies: network effects. First, however, a brief history of the concept is needed.

Network effects, venture capital speculation, and risk

While the idea that a network's size is positively related to its value stretches back to early telecommunications systems (Vail, 1909), the modern understanding of network effects emerged out of personal computing. In the mid-1980s, Robert Metcalfe, one of the inventors of Ethernet, developed what he called a “high-concept Ethernet sales tool” (Metcalfe, 2013: 26) for new computer networking hardware that enabled the first Local Area Network (LAN) between personal computers. That sales tool—now called Metcalfe's Law—proposed that the cost of a network is directly proportional to the number of networked devices (N), but that the value of the network was proportional to the square of the number of networked devices (N²). Values only exceed costs once a critical mass of networked devices (or users) is reached. Metcalfe and his sales team used this simple but powerful idea to convince early IBM PC owners to purchase Ethernet adapters so that critical mass could be reached. Metcalfe's Law, still debated empirically in the academic literature (Odlyzko and Tilly, 2005; Zhang et al., 2015), popularized the core proposition of network effects taken up within the tech and venture capital industries: as a network grows so too does its value.

Throughout the late 1980s and 1990s, economists and business scholars distinguished two primary types (Katz and Shapiro, 1985; Liebowitz and Margolis, 1994). Direct network effects result from a network with a single type of user-group, where the addition of a new user benefits existing users equally. For example, a single telephone is useless but more utility is gained with each new connection, creating increasing returns to scale. Indirect network effects, typical of digital platforms, accrue when there is more than one type of user-group in the network, with the addition of a new user in one group (e.g. buyers) increasing utility for users in another (e.g. sellers). A platform company like Uber intermediates a “multi-sided market” in which more drivers attract more users because of lower wait times and cheaper prices, and more users attract more drivers because there is more demand and more opportunity to earn (Parker et al., 2016).

All markets, of course, have two sides, but what distinguishes multi-sided platform markets is how the intermediary company develops a price structure to incentivize the participation of each “side” (Rochet and Tirole, 2003). This is referred to as the “chicken-and-egg problem.” To attract buyers, a platform should have a large, already-existing base of sellers, but sellers will only join a platform with a large, already-existing base of buyers (Caillaud and Jullien, 2003). How a company solves this conundrum depends on its strategy for attracting different user-groups. Google, for instance, offers a suite of “free” services to draw in users and, in turn, this large user base attracts advertisers, its major source of revenue. As Srnicek has identified, this type of cross-subsidization is an essential characteristic of platform companies. Platform firms and investors will accept losses on one side (or, temporarily, even both), betting that they can grow the user base through coordinating network effects to attract the other side, to achieve profitability at scale (Srnicek, 2016).

For early-stage platform companies, then, access to capital is essential to reaching Metcalfe's hypothetical point where the value of the network exceeds cost. Yet the majority of banks and private equity firms see platform companies as simply too risky an investment since most have no physical assets or demonstrable path to profitability in early stages. Venture capital, by contrast, is defined by its exceptionally high risk-tolerance. It is estimated that over half of all venture capital deals lose money, around 20% recoup their investment, with just 6% of all deals producing around 60% of returns in the industry (Dixon, 2015). As such, even more so than other forms of private equity, venture capital is riven through with speculation and risk. Over time, the industry has developed particular institutional and cultural practices designed to mitigate that risk burden across its portfolio of companies, especially since the 1980s when financial deregulation triggered an influx of capital from new sources such as pension funds (Nicholas, 2019; Zook, 2005). From the perspective of the VC firm, these interventions are not geared towards the long-term sustainability of a portfolio company, but rather towards growing it very rapidly so the VC fund can exit its investment profitably through an acquisition or IPO. After this point, whether the portfolio company fails is of no consequence.

This systemic prioritization of rapid growth makes venture capital a volatile and highly speculative industry. Failure is the norm and valuations can be radically out of proportion to a startup's existing revenue and assets, or even path to profitably. While these characteristics stretch back to venture capital's origins (Nicholas, 2019), platform capitalism has accelerated the industry's speculative tendencies. As the case of Uber demonstrates, platform firms losing billions of dollars a year still can attract venture capital firms betting that the platform will coordinate network effects to scale quickly, capture more data than their competitors and monopolize more of the market, which then creates more value and draws in even more users—a virtuous circle that culminates in monopoly rents (Parker et al., 2016).

From this light, it becomes clear that network effects are part and parcel of what Tsing (2005) calls the “economy of appearances.” As she writes, “In speculative enterprises, profit must be imagined before it can be extracted; the possibility of economic performance must be conjured like a spirit to draw an audience of potential investors” (Tsing, 2005: 57). For this reason, Langley and Leyshon (2017: 14–15) conclude that it is through network effects that platform firms become “a legitimate object of capitalisation,” their shares constructed as asset class that can generate returns on investment. While all financial investment is speculative in that a return is never guaranteed, network effects legitimize speculation on platform firms, computationally and discursively constructing them as a viable business model and object of investment. Despite this importance, however, the concept remains underexamined in the platform urbanism literature, particularly from the perspective of the technological everyday—a lacuna I seek to remedy in the remainder of this paper.

The social lives of network effects

The above political economic analysis of network effects is a limited conceptualization that should be expanded by attending to its social “overflows” (Callon, 1998). Any attempt to “frame” a phenomenon in purely economic terms results in inevitable social seepages that require material and discursive work to manage (Appel, 2012; Polanyi, 1944). In economics, the framing process of modeling economic interactions results in constant attempts to identify and price externalities—viewed as anomalies that reflect the failure to adequately frame the phenomenon in the first place. From this perspective, network effects are such an externality; users gain consumption benefits each time another new user joins the network that may not be reflected in their price of entry. Indeed, much of the platform economics literature seeks to internalize network externalities into the price structure of a platform so as to avoid market failure (e.g. Katz and Shapiro, 1985). Epistemologically, the assumption is that network effects created by the platform can be enclosed and measured, enabling an appropriate price to be placed on their value.

Michel Callon reminds us, however, that this distinction between what is internal to the network and what is external is only ever a temporary achievement. Technologies and people are already situated within dense social relations that “overflow” any framing that attempts to delineate boundaries of interaction (Callon, 1998; Latour, 2005). As he puts it: “[A]ll framing thus represents a violent effort to extricate the agents concerned from this network of interactions and push them onto a clearly demarcated ‘stage’ which has been specially prepared and fitted out” (Callon, 1998: 253). Platform firms frame network effects as an object of speculative investment, seeking to extricate the phenomenon from its social basis. In this performative process, platform workers are reduced to data points, atomized into market-actors, severed from collective norms and existing social relations, transformed into data trails that represent growth-potential, and stripped of their agency to build networks of their own. Even the critical platform studies literature tends to conform to this narrow economistic understanding by remaining preoccupied with macro political economic analyses of how network effects reinforce platform power: network effects enable the accumulation of more users and data (Srnicek, 2016), legitimize capitalization (Langley and Leyshon, 2017), and create strong monopoly tendencies (Christophers, 2020; Sadowski, 2020a).

Ultimately, though, network effects are a socio-technical phenomenon that—despite the aligned interests of venture capital and platform firms to disembed, enclose, and commodify—creates new relationships between people in the technological everyday. ¹ There are always social “overflows” to the framing of network effects as purely an object of speculative investment and future rent-extraction. Digital platforms not only rely upon pre-existing social relations between urban residents, but also engender new networks of interaction that are not fully captured by within their architectures. Drawing on Callon and responding to recent calls by Barns and others, I seek to advance platform urbanism by attending to the social lives of network effects—how platform architectures fundamentally depend on the socio-spatial relationality of those situated at the intersection of multiple forces: cultural arrangements, institutional pressures, racial hierarchies, religious beliefs, knowledge systems, gender norms, political economies, and so on. (Barns, 2019; Hecht et al., 2014). In Greater Jakarta, ride-hailing drivers have drawn on their social relations to repurpose platform intermediation, autoconstructing their own networks derived from those developed via a shared platform employer.

The autoconstructed driver network in Greater Jakarta

Since 2015, the digital platforms Grab (based in Singapore) and its domestic Indonesian rival Gojek have disrupted urban transportation systems throughout Indonesia. Unlike ride-hailing platforms in Europe or North America, Grab and Gojek's success relies primarily on motorbike taxis, known locally as ojek, which are significantly faster in Jakarta's congested streets because they can cut through automobile traffic. Drawing on this pool of online motorbike taxi drivers (or ojol, derived from ojek online), both companies offer not only rides but also an extensive array of delivery services (Grab Food/Express and GoFood/GoSend, respectively), supplemented by their now-ubiquitous mobile payment systems (OVO and GoPay). ² These companies have grown at incredible rates: Gojek expanded from completing around 5000 orders per day in 2015 to over 3 million by 2018, roughly 35 orders per second (Noormega, 2018).

This rapid expansion has been enabled by a parallel growth of global investment in Indonesia's digital economy, quadrupling to $40 billion between 2015 and 2019 (Davis et al., 2019). As the two largest private tech companies in the country, Grab and Gojek lead in attracting investment from prominent venture capitalists such as the SoftBank Group, tech companies like Google, and even other ride-hailing companies like Didi Chuxing and Uber (which sold its Southeast Asian assets to Grab in 2018 for a 27.5% equity stake). At time of writing, Gojek has raised approximately $5 billion dollars, and Grab has nearly $12 billion dollars ³ , with significantly higher valuations for each. Nevertheless, like Uber and Lyft, neither company currently operates at a profit in ride-hailing. Nadiem Makarim, founder and former CEO of Gojek, has openly admitted that “we built the business with the assumption that ride hailing is only at a break even,” speculating that the popularity of motorbike ride-hailing will cross-subsize their more profitable food delivery and digital payments services (Suzuki, 2019). In the eyes of both venture capital investors and the platform firms, drawing in and maintaining a large pool of drivers is thus essential to coordinating speculative network effects in the Indonesian market. This has led to mass recruitment events, bonuses, and promotions to draw in more drivers: Gojek alone claims to have 1.7 million across the country (Samboh, 2021).

Despite these numbers, ojol drivers face substantial legal, material, and economic risks and uncertainties. First, the ojek—online or otherwise—is not a legally recognized form of transportation. Under Indonesian Law 22 of 2009, governing road transportation throughout the archipelago nation, the motorbike taxi cannot be considered public transportation in the same way as taxis (online or conventional), and thus never has been regulated at the national or municipal level in Jakarta. ⁴ Ojol drivers therefore operate in what one prominent NGO activist calls a “legal grey zone” that leaves them open to arbitrary state intervention: “If the law remains grey, the life of the ojol will also be grey.” ⁵ Second, ojol confront opaque algorithms and platform rules that govern their everyday lives and wages. Following the playbook deployed by ride-hailing platforms across the globe, drivers were initially drawn into the Grab and Gojek platforms through relatively high wage rates and large bonuses, incentives that have steadily been cut back in recent years as more users join the network. Drivers can be suspended or terminated from the application at any time with little recourse, and they report that the company's rationale for doing so is often unclear to them. Lastly, while the motorbike is extremely popular it is also deadly; the vast majority of all traffic accidents in the country involve a motorbike. Ride-hailing drivers therefore put themselves at significant physical risk to deliver passengers, food, and packages for others.

To manage these uncertainties and risks, drivers have built grassroots communities of mutual aid. Ojol driver communities (komunitas) usually consist of around 20–30 drivers who band together and establish a “basecamp” or “shelter” where they can rest between orders. Most start with a handful of drivers who wait for orders in the same area (mangkal), but these groups frequently grow into sophisticated organizations with their own internal structure, strict hierarchies, operating procedures, and elected or appointed positions: leader, field coordinator, secretary, treasurer, first responders, public relations, and so on. Emerging in South Jakarta by 2015, driver communities have mushroomed and evolved to take on significant responsibility for the social reproduction of drivers, informal worker training, and the regulation of the ride-hailing industry in the city-region. Based on my conversations with leaders in these communities, I estimate that there are approximately 3000 online ojek communities throughout Greater Jakarta, each with their own unique name and logo that drivers proudly represent as they move throughout the city-region.

Driver communities themselves are remarkably networked, regularly gathering in-person with other communities, but also online in local, district-wide, city-wide, and even nation-wide online communities via social media, particularly WhatsApp. Individual driver communities coordinate internally via WhatsApp groups, while also splitting off to form new, online organizations (wadah, or “container,” and lintas, “crossing”) dedicated to a shared purpose (e.g. emergency response, discussed below), shared territory (e.g. East Jakarta), or even a shared make and model of motorbike. WhatsApp allows for easy forwarding of messages simultaneously to many groups, rapidly spreading information about road accidents, protests, the latest app update, and so on amongst communities. Most drivers are a part of at least 20 such WhatsApp groups and it is not uncommon for drivers, especially community leaders tapped into more online groups, to receive hundreds—sometimes thousands—of WhatsApp messages in an hour. ⁶

In this way, drivers engage in what Prouse (2018) calls digital autoconstruction, expanding upon Holston’s (1991) analysis of self-built housing in peripheral Brazilian cities. Prouse shows how everyday residents and journalists in Complexo do Alemão, Brazil, create online spaces and collectives through social media applications, actively stepping into material and discursive vacancies left by the state and reshaping racialized state violence in the process. The prefix “auto” conveys a conceptual lineage: lacking state support, residents may come together to autoconstruct their own infrastructure, but Holston and Prouse emphasize that the consequences of this process are socially, spatially, and politically complex. Autoconstruction can engender new political subjectivities and possibilities, but also can re-entrench hegemonic relationships and norms.

In Greater Jakarta, as throughout Indonesia, Gojek and Grab have compiled an enormous pool of laborers—most already engaged in piecemeal work—with a low-barrier-to-entry job, and an extremely popular and affordable form of transportation (the ojek) to coordinate platform network effects, successfully packaging this business model for global investors. Yet there are unintended social overflows in bringing drivers together as a flexible labor pool. Online ojek drivers step into gaps left by platform firms and the state, autoconstructing socio-technical networks of their own to mitigate risks associated with their lack of legal status, economic and physical uncertainties, and the responsibility for social reproduction shouldered upon them. Though they are initially brought together by Grab and Gojek under a shared identity, drivers exceed platform architectures of data-capture and rent-extraction by building grassroots networks via the messaging platform WhatsApp.

This autoconstructed driver network provides value for its participants, which grows with the number of drivers. Each new driver added to this ecosystem of online and offline communities benefits those already connected: more resources for collective social insurance, improved response times for first responders, faster information dissemination, improved access to potential patrons or customers, and more protection against violence from conventional transportation drivers. I argue that these too are network effects, whose social lives are excluded from existing conceptualizations in platform studies because they are coordinated, maintained, used, and speculated upon by gig workers, not by platform firms or venture capital.

While still taking seriously the mutual imbrication of speculation and network effects highlighted above, I shift attention to the social lives of network effects in the technological everyday. As I will show in the following sections, drivers—and indeed a whole range of actors throughout Greater Jakarta—speculate on these autoconstructed driver network effects and seek to mitigate risk through them, although not all are equally able to do so. Speculation always exists in relation with risk, and the following sections explore this relationship. First, I examine the economic and physical risks downloaded to drivers by platform firms, and how drivers have cultivated socio-technical structures and practices of mutual aid in order to collectively manage those uncertainties. Second, I explore how various groups and institutions—including drivers, firms, civil society groups, political parties, and the Indonesian state—seek to tap into grassroots driver networks, speculating that they can advance their political, social, or economic interests by so doing.

Mutual aid and collective risk management

The platform business model relies on the ability to externalize costs, risks, and responsibility for social reproduction onto workers (van Doorn, 2017). This is primarily achieved through the legal, political, and discursive work these firms put into positioning themselves as merely passive intermediaries connecting different users, despite the significant control they maintain over conditions of work (Gillespie, 2010; Rosenblat, 2018). Ride-hailing firms classify drivers as independent contractors, or “mitra” (partners) in Indonesian labor law, making them ineligible for employer-paid insurance, pension funds, collective bargaining, and other employee benefits. Furthermore, drivers must pay for their own gas, insurance, vehicle maintenance, and so on, forcing “workers to shoulder the risks and responsibilities of social reproduction” (van Doorn, 2017: 902). This downward redistribution is itself tied to the valuations of platform firms, and venture capital speculations upon them, positioning them as a “lean” business with low overhead (Srnicek, 2016).

The risks displaced onto the hundreds of thousands of ojol drivers in Greater Jakarta are significant. Drivers are regularly on the road for 12–14 hours a day, and deaths of Grab and Gojek drivers are a common occurrence in Jabodetabek according to my informants, especially in the industrial areas of North Jakarta where a driver can easily be crushed by a lorry. Even a minor accident with no injury can mean devastating lost wages for the driver if their motorbike is damaged beyond immediate repair. Falling wage rates and bonuses in recent years compound risk: drivers must spend even more hours on the road, exposing themselves to not just more accidents but also Jakarta's extreme temperatures and chronic air pollution, regularly among the worst in the world.

Drivers’ autoconstructed networks and systems are designed to collectively manage such hazards. One primary means is the collection and redistribution of dues. ⁷ Nearly all of the komunitas that I visited during my fieldwork require members to pay community dues, which averaged around 20,000 rupiah per month, per person ($1.46 USD). Dues “are collective in nature, their purpose is to further collective interests.” ⁸ These funds, managed by the community treasurer, are distributed based on need: helping to pay for motorbike repairs, parental leave after the arrival of a new baby, a stipend if a driver cannot work due to illness, and so on. In effect, community dues function as a mode of social insurance in Jakarta's platform economy, filling in responsibilities for social reproduction of the platform labor force.

Dues can also be redistributed to the larger population of ojol beyond the komunitas. Many communities that I encountered donated a monthly percentage to emergent social and religious organizations dedicated to improving ojol drivers’ lives. One such organization is GAS (Garasi Amal Sholeh, or “Good Deeds Garage”), which provides a sort of life insurance for the children and families of deceased ojol. GAS supports several hundred orphans throughout Jabodetabek, around 70% of whom had parents who were ojol before they passed away. ⁹ At a weekly distribution of donations to orphans in North Jakarta, one leader shared with me that the organization has actually been around for 19 years, but was recently reinvigorated by participation and contributions from ojol communities. ¹⁰ The increased number of children orphaned because their parents were killed while driving for the platform companies has meant that there is more need, he somberly explained.

Driver communities have also formed grassroots emergency response networks. All driver communities have an internal “unit reaksi cepat” (URC) or “quick reaction unit” that is responsible for responding to emergency situations and other types of “trouble,” the adopted English term used by drivers to refer to problems such as mechanical failures, flat tires, conflict with conventional ojek drivers, or other emergencies. The leaders of these community-level URC teams participate in larger-scale regional umbrella organizations (e.g. URC South Jakarta, URC Bekasi), facilitating scalar coordination and information-sharing in the event of an incident. URCs make extensive use of WhatsApp features, especially voice messages, group chats, and the live location sharing feature that drivers use to track one another in real time to monitor safety. These techniques, developed and shared throughout the URC communities, have created a sophisticated lattice network of coordinated emergency response throughout the city-region.

A vignette from my fieldwork illustrates this point. On an especially muggy afternoon in April 2019, I visited a community basecamp in the North Jakarta region of Tanjung Priok. In the middle of our conversation there was a sudden flurry of activity as drivers started rapidly listening to and sending audio messages through the URC North Jakarta WhatsApp group, reporting that there had been an accident about 10 km away in which a motorbike driver had been hit by a Grab Car driver. The “field coordinator” (korlap, kordinator lapangan) of this basecamp dispatched a URC member to the accident, the exact location of which was shared by the Grab Car driver via WhatsApp. The injured man eventually was taken to a nearby hospital, while URC members tracked their progress on the road through the live location feature. Once he arrived, a URC member sent a selfie to the URC WhatsApp group to confirm that the patient has been successfully admitted, and that the hospital would accept his state-provided insurance and had the proper equipment to handle his injuries. In the end, it turned out the victim was not even an ojol driver.

Across the city-region (indeed, the country), URC units respond to these types of events hundreds of times daily, rivaling existing social services and far outstripping the companies’ efforts to protect their “partners” (mitra). Under pressure from driver protests, in 2019 Gojek provided three ambulances for the entire Jabodetabek region. According to the URC members I talked to, however, it can take up to 3 or 4 hours for an ambulance to arrive. “By that time” one member told me, “the driver will already be dead.” ¹¹ In contrast, URC response times are within minutes because of their wide dispersal throughout the city, ever-growing numbers, and innovative standards and practices.

URC's ubiquitous presence helps keep drivers on the road when their motorbike breaks down, gets a flat tire, or other “trouble.” Delays or incomplete orders can trigger negative ratings from customers that can lead to suspensions or terminations, putting strong structural pressures on drivers to take matters into their own hands. In effect, this relieves pressure from the platform companies for handling these types of issues, further downloading responsibility onto driver communities and networks. As one field coordinator put it to me, “It's the risk we take as drivers. The office doesn’t wanna know about it. Their attitude is that they’re just here to sort us out with orders, how we fulfill them isn’t their business.” ¹²

While significant scholarship shows how platform intermediaries retreat from social reproduction by claiming that they merely link users, my findings reveal that gig workers in Greater Jakarta step into these gaps themselves to reshape the material conditions of their work. For platform firms, a downward redistribution of risk and responsibility reduces operations costs, performs their “lean” overhead to VC firms, and allows them to redirect capital towards coordinating network effects at scale rather than towards labor costs. For drivers, on the other hand, it motivates digital autoconstruction of their own networks in order to collectively manage that risk, scaling as more and more drivers become connected. These are not merely different instantiations of physical or financial risk; rather “profitable risk and exploitative risk are mutually dependent” and relationally constructed in the platform economy (Appel, 2012: 703). And yet drivers in Greater Jakarta have found creative ways to collectively manage that risk through their online and offline networks, keeping them on the road to feed their families and insure themselves in case of injury or death.

Speculative network effects in the technological everyday

While speculation is often relegated to the realms of financial markets, geographers and economic anthropologists have shown how speculation is a social practice through which all kinds of actors and groups attempt to deal with uncertainty, improve their life chances, and plan for the future (Bear, 2020; Humphrey, 2020; Simone, 2008; Leitner et al., this issue; Upadhya et al., this issue). Throughout his work on African and Asian cities, including Jakarta, AbdouMaliq Simone articulates the ways in which urban residents engage in speculation through investing time, energy, and money in new ventures and social relationships that may bring unforeseen prospects, patrons, new access to housing or credit, or other opportunities. For Simone, this is a modality of speculation that allows Jakarta's urban majority to manage the risks associated with urban life, where one's access to housing, work, water, and so on may only be temporary. Such uncertainty provokes “doing something out of the ordinary” (Simone, 2008: 60)—a side hustle, a new relationship that might pay off at a later date, money down on an acquaintance's nascent business, a bribe with an uncertain payoff, participation in a multi-level marketing scheme, a move to a more advantageous location. These “everyday speculations” may or may not be financial (Leitner et al., this issue), but nonetheless are geared towards improving urban residents’ lives and livelihoods, even if that outcome is uncertain or ventured at great risk.

Adopting this lens shows how many different actors in Greater Jakarta speculate upon socio-technical driver networks, if not always with the same power to accumulate from them. For drivers, everyday speculation could mean simply utilizing driver networks to advance their other economic ventures. At driver community meetings, on WhatsApp groups, through other social media, and even while on the road, drivers market individual side-businesses—selling Lebaran cakes from the back of their motorbike, offering a promotion for their auto-repair shop, bakery, or clothing business. Indeed, whole new cottage industries have cropped up to cater to online drivers and their communities. Whenever I attended social gatherings for drivers I would run into Joko ¹³ , who sells customized pins, buttons, and stickers representing the logos of different driver communities. Driving part time for Gojek, he spends weekends selling his wares at driver community “anniversaries,” large celebrations that can draw hundreds of ojol but also entrepreneurs like himself and others catering to the ojol market, such as telecoms companies, motor oil companies, or motorbike manufacturers. Joko spends most of his Gojek earnings on making his pins and other wares, betting that he can sell them for more than he can earn from Gojek.

Others speculate that access to driver networks might offer social capital or patronage relations. During my fieldwork, I regularly visited the basecamp of a driver community called Go-Venture ¹⁴ , whose leader explained to me that he expected his members to drive only part time for Grab or Gojek so that they could focus on their other entrepreneurial activities. But he also emphasized that the two could be mutually beneficial:

We don’t support entrepreneurship financially, but we can help by marketing the goods. We can also help by registering the [member's] business with GoFood. We have contacts with the [Gojek] office, those guys can cover us. ¹⁵

As this leader suggests, higher-ups in driver communities often have privileged access to Grab and Gojek employees through dedicated driver-management WhatsApp groups or personal connections. These “informal” online and offline spaces enable drivers to receive other benefits from associating with a well-connected community: getting their side-business registered with GoFood, sorting out a technical problem with the app, and especially help with being re-instated if temporarily suspended for an offence (such as a 1-star review). ¹⁶ As Ari puts it, “The management befriends us because there's something that we can give to them, and of course it's the same for us. There's always give and take […] so, if our friends have trouble with their account, we can just ask for help from the management.” ¹⁷ Notably, these company patronage relations exist outside of formal channels for re-instating drivers, a motivating factor for drivers to join a komunitas under the speculation of this privileged access to management.

As Ari hints, the platform firms themselves speculate on autoconstructed network effects, though not just in the purely economistic sense described above. With the growth of komunitas in Jabodetabek, Grab and Gojek have sought to tap into their leadership because their WhatsApp groups give them access to—and influence over—large numbers of drivers outside of the more rigid architecture of the application. The bigger the driver community, the more pressure there is for Grab and Gojek management to engage. According to the Gojek VP for Operations for Jabodetabek, “This [driver] engagement really helps us do many things in our process…it helps the scalability…Even if we do on-boarding [registration], what do we do? We just tell the community leader….” ¹⁸ Management's access is never guaranteed, however; the risk of being cut off is always present. While some drivers describe these efforts as “symbiotic,” many saw them simply as worker control: “The company needs the komunitas because they want a good relationship with drivers…amicable communication ensures drivers will strike less because they have a channel to voice their opinion.” ¹⁹ Regardless of a driver's individual position, the autoconstructed driver network and its scaling effects becomes an object of speculation that cannot be reduced to data alone, but is suffused with social relations of patronage, trust, risk, and control.

Everyday speculation on these networks is not limited to drivers or the companies, however: “The more the movement of the ojol as a group [grows], the more irresponsible people take advantage of our solidarity, our social spirit.” ²⁰ The number of driver communities, their networks, and the depth of their organizing have led to political power that, in turn, has garnered overtures by all manner of actors well-removed from the ojol lifeworld. Throughout my fieldwork, my informants reported that their communities were approached by local officials (RT or RW), political parties or individual candidates, civil society organizations (ormas), labor unions, and the Indonesian state. All speculate that connection to the driver network can advance their political agenda, whether that be conscripting neighborhood “eyes on the street,” patronage politics, or worker solidarity and dues collection. These actors sometimes even become ojol themselves, sparking frequent rumors and suspicion of spies amongst communities: “They [join] because of…another motivation, not an economic interest to make a living, no…I mean political power, mobilizing the ojol for political interests.” ²¹ These concerns were especially prominent during my fieldwork, as the 2019 general election unfolded, and various political parties and groups approached communities due to their large numbers and dense networks.

In short, the autoconstructed driver network has engendered practices of financial and non-financial speculation by a diverse range of actors in Greater Jakarta. Following Simone, I understand these practices as speculative in that Jakartans—under uncertain conditions—invest their time and energy into networked social relationships that may or may not realize other opportunities. From this vantage, it is clear that venture capitalists and the platform firms are not the only ones who speculate on network effects.

Conclusion

For those whose lives have been reshaped by Grab and Gojek in Greater Jakarta, there is deep-seated uncertainty about the online ojek's future. Drivers are wary after watching many ride-hailing competitors rise and fall over the last 5 years; even the apparent global titan Uber was only a short-lived opportunity here. Investors and platform executives alike eye consolidation, and rumors of Grab/Gojek mergers are always on the winds. The regulatory landscape itself lacks solid foundation, as regulators consider designating gig workers employees and revising Law 22 of 2009 to recognize the ojek as public transit, both of which would have profound but unclear consequences. In the meantime, the COVID-19 pandemic has laid bare the underlying, day-to-day precarity in which drivers operate. Despite its near ubiquity, the industry dwells in the ambiguity of long-term viability, even if the short-term is flooded with venture capital.

Amidst this uncertainty, ojol drivers speculate while they can on the opportunities, cottage industries, and social networks engendered by platformization. Drivers have reforged their connections to one another to manage risk, deal with uncertainty, and piece together a living. In doing so, they have autoconstructed networks that now permeate ride-hailing operations in the city-region, with sophisticated systems of mutual aid, insurance, emergency response, and social reproduction filling in gaps left by the retreat of platform firms and the state. And yet, Prouse reminds us that the outcomes of digital autoconstruction are never straightforward. Even as the driver network is a response to platform firms downloading risk and responsibility, it also paradoxically re-entrenches downward redistribution by allowing the firms to withdraw further. Moreover, driver network effects stimulate practices of speculation by a range of actors who seek to exploit driver networks for the own political economic interests.

Such an analysis offers a more multifaceted view of network effects than the current platform studies literature might suggest. On the one hand, the mainstream platform economics literature understands network effects as a technical externality, to be internalized into a platform's price structure if correctly measured and modeled. On the other, the critical literature largely assumes that network effects reinforce the economic value and power of the platform firm. Notwithstanding significant epistemological differences, the analytical focus and shared assumptions are the same: it is platforms that create and harness network effects. Taking up calls by Barns and others arguing for a more expansive theorization of platform urbanism, I have sought to refocus attention on urban residents and the complex socio-technical constellations of which they are a part, showing how other actors repurpose, recreate, and re-coordinate platform network effects in the technological everyday. As recent interventions have argued, attention to these types of tactics illuminates already-existing, counter-hegemonic modes of platform intermediation beyond those determined solely by data-extractive or rentier logics. Conceptually, this allows space for retheorizing platform urbanism as contingent, overdetermined, and always, already open to reformulation.

These findings suggest unrealized overlaps between platform studies and speculative urbanism. Currently, the literature engaging with this concept focuses almost exclusively on land and real estate. Yet, given the extent to which the urban transformations caused by digital platforms are materialized primarily through speculative risk capital, there are productive intersections between these two literatures. As I have shown, network effects link different practices of speculation and risk management at multiple spatial scales, intermediating how differently positioned actors shape, and are themselves shaped by, digital platforms. From the desk of a venture capitalist, network effects are an object of speculation, an abstract, calculated risk inherent to the cost of doing business. From the seat of a motorbike in Jakarta's streets, however, network effects are a lived reality, a critical—but never guaranteed—resource to mitigate risk, secure a living, and speculate on a better one. This is not to say these are simply different experiences of speculation and risk, although that is true. The point is their relationality. Tying them together are co-constitutive, inter-scalar relationships of speculation and risk exposure that, if paid attention to, highlight the ways in which urban mobility is increasingly underpinned by speculative city-making in the historical conjuncture of platform capitalism. In unearthing these relations, however, it is critical to not lose sight of how everyday users inevitably overflow the economic interests in bringing them together—to, in other words, keep in view the social lives of platform architectures and their network effects.