1 Introduction

The platform economy has now entered its second decade, and researchers have investigated a wide range of platform outcomes, including inequality, trust, and racial discrimination (for a review see Schor & Vallas, 2021). There are also many studies of workers’ experiences, particularly on lower-paid apps such as ride-hail, shopping, and delivery (Ravenelle, 2019; Griesbach et al., 2019; Robinson, 2017; Cameron, 2018; Ladegaard et al., 2018). These accounts have tended to describe a common platform experience, typically that of highly committed workers. Schor et al. (2020a) have argued that the literature has not sufficiently addressed how unique features of the platform model — low barriers to entry, choice of hours, and the ease of working for multiple platforms at once — produce a heterogeneous labor force. Here, we explore another dimension of a heterogeneous platform workforce, namely the diversity of economic orientations of earners, or to use a term from economics, their “behavioral models.” To sociologists, the presence of multiple behavioral models, especially among professionals or the self-employed, is not a novel finding (Fridman, 2020). Users of these platforms, who tend to be young and highly educated (PEW Research Center, 2016a; 2016b; Schor & Cansoy, 2019), profess a range of motivations for participating (Fitzmaurice et al., 2020; Ravenelle, 2017) and their dependence on platform income also varies (Schor et al., 2020a). Therefore, the existence of multiple behavioral models in this sector may be unsurprising. However, as Beckert (1996; 2003) notes, sociologists have failed to theorize “models of economic action,” i.e., how individuals translate a series of preferences into behavior.

Using in-depth interview data from 102 respondents on three platforms (Airbnb, TaskRabbit, and StocksyUnited) we find that within individual platforms, there are diverse strategies for governing earning behaviors. Iterating between models found in the literature and our data, we have identified three main orientations. Some are maximizers, engaged in the kinds of activities ascribed to homo economicus, economists’ archetypal rational actor. A second group displays a more social orientation, and although these earners are also interested in money, they are not optimizers on the financial margin. They have other goals, including sociability. But they also draw ethical boundaries around their platform work, reject opportunities that don’t reflect their social preferences, or act to gain recognition. We call this type homo socialis. A third group, which we term homo instrumentalis, displays neither maximizing behaviors nor strong social preferences. They merely aim to earn, and do so in casual, habitual, or targeted ways.

We provide an account of the three models and discuss platforms’ tolerance for the heterogeneity we find by discussing how earner diversity aligns with the companies’ imperatives for growth. This allows us to contribute to theorizing questions such as whether platforms represent something different from conventional businesses and the extent to which they are novel forms. The paper proceeds with a discussion of theories of the platform firm, a brief discussion of the three types of earners we discovered, our methods, findings, and a section on platform responses.

2 Theorizing the Platform Firm

The presence of diverse models of economic behavior within a single platform’s labor force raises questions about the nature of this type of firm. Is it more like a market, in which diversity among participants is tolerated, especially if they are willing to take lower returns? Or are these platforms more like bureaucratic firms that attempt to control the behavior of their employees? Scholars have begun to develop understandings of platform firms which shed light on these questions. Vallas and Schor (2020) identified four approaches to theorizing digital platforms — efficiency accounts, “Uberization” and precarity narratives, algorithmic control, and a fourth hybrid conceptualization. A second set of issues concerns the behavior of earners. Is there anything novel about how people use and earn in this context, or are conventional models of economic behavior sufficient? The literature is less explicit on this question, so our discussion addresses implicit assumptions.

Efficiency accounts, which come mainly from economists, focus on the ability of digital technology to reduce transactions costs and increase efficiency (Horton & Zeckhauser, 2016; Sundararajan, 2016). This results in a shift from the firm (a command and control entity) toward the market (a voluntaristic, atomistic structure), as well as toward smaller, peer-to-peer entities (Einav et al., 2016). This view is consistent with an earlier shift to seeing firms as a “nexus-of-contracts” (Jensen & Meckling, 1976) which Davis (2016a) argues helped create “Nikefication”, i.e., the outsourcing of functions, and the “vanishing” of the American corporation (Davis, 2016b). On the question of behavioral models, the economics literature has been silent, assuming that the standard models — either the rational actor or its behavioral economics cousin — prevail.

Outside of economics, the dominant perspective is “Uberization.” In this view, the platform is little more than a “web page” (Davis, 2016a, p 513). Grabher and van Tuijl (2020) formulate this shift as going from employment to a “gig.” Uberization is the culmination of trends such as fissuring (Weil, 2014) and precarious work (Kalleberg, (2013), which describe the breakdown of stable employment and benefits, the shift of costs onto workers, and the increasing insecurity of labor. As with the economic approach, the precarity literature generally does not focus on the heterogeneity of the labor force other than considering variations in levels of precarity. While some accounts do note that earners fall into different categories (Ravenelle, 2019), the focus is on the common worker experience of bearing costs and risks (For an exception, see Manriquez, 2019). Similarly, while ethnographic accounts describe differences among people, this approach generally avoids theorizing behavioral models and has not provided a distinctive analysis of economic action on platforms.

The third approach sees platform firms as novel entities on account of their ability to control labor via algorithms (Aneesh, 2009; Rosenblat & Stark, 2016; Griesbach et al., 2019). Algorithmic control is enhanced by information asymmetries that enhance the power of the platforms over workers. While this view does not deny the precarity of workers, it emphasizes top-down authority, an all-powerful Panopticon collecting user data (van Doorn & Badger, 2020) and a break from previous methods of control. The algorithmic approach does not typically address issues of socio-economic, demographic, or behavioral heterogeneity among the workforce, at least not from the perspective of what it might mean for understanding these firms. Instead, the key variable is the power that technology affords to platform management. A related view focuses less on labor control but on the ways in which technology affords platforms even more power than conventional firms, and allows them to expand their command and control to the markets they operate in (Kenney & Zysman, 2019; Srnicek, 2016). This literature harkens back to the early postwar era with its focus on dominant firms with monopoly power.

The fourth view sees the platform firm as novel because it is a hybrid form, combining organizational and technological factors to create a new entity. It emphasizes the networked and market relations of firms, and their departure from the conventional hierarchical corporation. Yet these accounts also hold that firms retain power over the parties with which they transact, including labor. Two contributions are especially relevant. Watkins and Stark (2018), who studied platforms in a variety of sectors, argue that these entities have evolved into a “Mobius” firm. They have gone beyond the simple network formulation and are able to co-opt resources that are both internal and external to the firm. A related argument by Kornberger et al. (2017) is that platforms are unique because they employ novel accounting methods that move from traditional hierarchical categories to heterarchies. By introducing new evaluative methods such as ratings and reputational systems, platform firms can radically decentralize control and simultaneously centralize power. This hybrid formulation — decentralization with concentrated power — is common to this genre of platform theorization.

We (Schor et al., 2020b) have made a similar argument about the platform economy, arguing that these firms have enacted a “retreat from control” as they allow earners to choose hours and schedules and organize their labor processes relatively autonomously. While autonomy varies across platforms, in comparison to conventional service labor, platform work is generally less scripted or directed by the employer (Leidner, 1993). But while the platform allows the worker the freedom to “be their own boss,” it retains power and certain forms of control, including through the use of market discipline. Market discipline may be especially relevant in low-wage sectors, such as ride-hail and delivery, where requisite skills are generally available throughout the population, which gives firms considerable latitude to recruit new earners, a key lever for affecting supply. Finally, we note that “permissiveness” varies by platform, and lower-wage workers are subject to more control than those with more market power.

Willingness to allow worker freedom of choice over schedules and total hours is a unique feature of platform management. It produces a consequential aspect of platform work — heterogeneity in when and how much earners choose to work on the platform.¹ There has been relatively little attention to how workforce heterogeneity might manifest in other ways. In our research, we discovered that within and across platforms, earners exhibit different modes of earning. This finding adds another distinctive aspect to platform firms, which we explore below.

3 Models of Economic Behavior

How should we understand economic behavior? While economists have historically produced varied answers to this question, by the 1970s, they had coalesced around a single idea for representing and modeling economic behavior — the rational maximizer. However, nearly as soon as they had, “behavioral economists” came along to trouble that fiction, with a wealth of empirical findings that violated the principles of selfishness (“fairness norms”), revealed time inconsistency in preferences, loss aversion, and non-linear probability weighting of alternatives (Kahneman, 2011). These developments revitalized ideas such as income targeting and Simon’s (1957) bounded rationality and satisficing. Among economic sociologists, whose project began as a critique of the neo-classical model, the focus has been on how structures inhibit maximizing behavior. Approaches include Bourdieu’s (1984) habitus, social networks (Granovetter, 1973), Polanyian embeddedness (Block & Somers, 2014), and “relational” economic sociology (Zelizer, 2013). However, given the diverse ways economic sociologists have explained economic outcomes, they have generally not focused explicitly on models of economic behavior. Indeed, Frank Dobbin (2007) has made the point that economic sociologists have generally accepted the view that agents seek profits. Jens Beckert (2003; 1996) has argued that economic sociologists have generally not constructed their own “models of economic action.”

And what of platform actors? Which of these conceptualizations do they conform to? We found three distinct orientations — economicus, socialis, and instrumentalis. Before discussing each, however, we want to stress that all of our respondents are active on the platforms in order to earn money. If they were not, they would be more likely to be participating on gift exchange sites such as Couchsurfing (a free alternative to Airbnb) or time banks (multi-lateral barter service exchanges). Therefore, our analysis does not replicate well-worn tropes of altruism versus self-interest or money versus love (Folbre, 2001). What we find is that among a financially-motivated group, there are major differences in how people think, act and transact. That said, we note that we have not included low-wage platforms, where we would expect more conformity among earners. However, our ongoing interviews with workers in ride-hail, food delivery and shopping also suggest some diversity of orientation.

The first group we call homo economicus. These are classic rational, self-interested actors who pursue optimal outcomes. While they focus on maximizing earnings, what’s distinctive is how they do it. They develop individualized strategies to maximize the prices they can command, often by seeking out market information. They pursue actionable information about costs, earnings, and how to improve their margins. They are engaged in continuous calculations, keep spreadsheets, pay careful attention to costs, and sometimes experiment with pricing. They strategize about ways to increase their earnings, by improving or expanding their real estate assets, sub-contracting tasks or services, or investing in their platform activities, which they think of in largely commercial terms. Ultimately, they understand themselves, and others, through a lens of idealized rational action.

Noting the shortcomings of the homo economicus model, the “new economic sociology” of the early 1980s questioned how social networks might clarify seemingly illogical behaviors not otherwise explained by the rational actor model (Granovetter, 1973). At economic sociology’s core is the claim that economic activity has a social dimension that is integral to understanding why actors make fiscal decisions. Building on this insight, Zelizer argued that economic relations were not merely embedded in social context but that they were “continuously negotiated” and “meaningfully interpersonal” (Zelizer, 2012, p. 146). Zelizer defined this as a relational package in which actors balance four unique elements: distinctive social ties, a set of economic transactions, media, and negotiated meanings. This framing suggests a robust social actor who will weigh social incentives in their economic decisions. We term this category homo socialis.

Homines sociales have varied motivations and behaviors, such as meeting people and socializing, building community, or avoiding status threats. They are unified by strategies that are guided, first and foremost, by relational incentives and social considerations. They value income, but instead of spending energy tinkering with the bottom line, homines sociales turn their attention towards maintaining personal ethics, seeking validation, and fostering social connections. While a good number of homines sociales are prosocial, this category also includes individuals who draw strong boundaries to avoid particular interactions, such as tasks which involve status insults or hassles they would rather avoid. Some engage in discriminatory behaviors, even at the expense of making money. In short, homines sociales do not prioritize income maximization. Nor are they particularly oriented to calculative completeness or searching for market information. They are happy to participate in economic transactions that can coexist with their social specifications, but abstain from economic opportunities that violate their larger “social” orientations.

The third type we have identified, the homo instrumentalis, is less discussed in the literature, either in economics or sociology. Like homo economicus, this type is strongly oriented to making money, rather than to social relationships or social goals. However, their relationship with money is largely instrumental. Some earn for a specific purpose, such as for rent, debt payments, vacations, or even beer. Others operate with a target income — when it is reached, they reduce economic activity.² Our homo instrumentalis lacks a coherent strategy, uses simple heuristics and resists pressures to do more, or to optimize participation. They often settle for the “good enough” outcomes described by Simon (1957). They are satisficing agents, who are not compelled to spend further effort searching for marginally better outcomes (Caplin et al., 2011).

4 Methods

Our goal is to explain a variety of participant orientations in the platform economy. We therefore focused on platforms with different business models, barriers to entry, and remuneration structures. We discuss three platforms — Airbnb, TaskRabbit, and StocksyUnited. Airbnb is a platform where hosts rent rooms or entire homes on a short-term basis, at prices that are significantly higher per night than long-term rentals. Hosts set the price, manage booking requests, and clean and prepare the home for stays. Depending on the characteristics of the home, hosts might spend time with their guests, for instance if they share a kitchen or a living room. TaskRabbit is a platform for a wide range of tasks, but most are either delivery, cleaning or manual labor tasks such as moving (Cullen & Farronato, 2018). In the first version of the platform, workers used an auction model to bid on posted tasks. In 2014, auctions were replaced with a model where customers search an inventory of Taskers and select one based on hourly prices and profile descriptions. The third platform, StocksyUnited, is an artist-owned and governed stock photography co-operative. It has a competitive, limited membership of over 1000, and members make major decisions. Members’ work is included on the company’s website and sold at a flat rate determined by the management. In contrast to Airbnb and TaskRabbit providers, Stocksy artists do not set prices. However, they can boost their sales and increase earnings by investing in photo shoots and tailoring their work to the market. They also control how many photographs they submit.

We conducted 102 semi-structured interviews, scheduled for 60 minutes, although some lasted longer. Most Airbnb and TaskRabbit respondents were located in Greater Boston and were interviewed in person, but a few lived in other US cities and were interviewed by videoconference. Stocksy interviews were done online, as members span the globe. Airbnb and TaskRabbit interviews began in 2013, and continued until 2017; Stocksy interviews were conducted in 2017 and 2018. We recruited Airbnb earners initially by messaging them on the platform, but switched to snowball sampling and social media as the platform deactivated the accounts we were using. For TaskRabbit, we hired interviewees through the platform with the interview as the “task.” We recruited Stocksy members with the help of management, who provided a list of names and emails. We contacted members and asked them to schedule an interview.³ Compensation was initially $30 per interview, and was subsequently increased to $40. Demographic details of our sample can be found in the appendix.

All interviews were transcribed. For the qualitative analysis, we sorted our interview transcripts by platform category and assigned a principal coder to each group, to make it easier to compare different logics within the respective platforms. We began with a round of open coding, in which we developed basic descriptive categories around narratives of self, platform goals, and economic practices, which we discussed as a team in subsequent memos and meetings. For example, we noted that some Airbnb hosts used demand-estimation strategies in order to optimize their prices. Quotes from such participants were distributed to all team members and in discussions we agreed that such a practice is characteristic of homo economicus, because it contributes to the larger goal of maximizing platform earnings. After agreeing on ideal-types for economicus, socialis, and instrumentalis, which we detailed in a shared codebook, we read all transcripts and coded all participants in a round of systematic and theoretical coding. Throughout the process, we frequently returned to the codebook, made adjustments when necessary, and continued to write and talk about our participants and their practices, until we had formed full coding consensus on our participants (Table 1). We had a small number of hybrid cases which displayed features of more than one model. This process meant that while we began with three ideal models from the literature, we developed and elaborated on their features through analysis of our data.

5 Findings

6 Platform Behaviors

An obvious question raised by our findings is whether platforms will continue to tolerate diverse earnings strategies. Given that platform investors and operators generally prioritize growth and subsequently profitability, will they continue to accommodate providers who do not maximize? If they do, it lends credibility to the view that they represent a new kind of hybrid firm. To find out how platforms adapt to provider strategies and to what extent they attempt to modify these behaviors, we draw on our interview data, personal experiences with the platforms, and archived website data from our three cases. Although none of the platforms has eliminated lower-performing earners, all three have made policy changes that nudge providers towards more lucrative earning strategies.

TaskRabbit’s model has evolved in a way that disadvantages homines sociales. Its original setup was a bidding system in which clients posted a task and interested Taskers responded by quoting a rate. Clients then chose among those bids. Taskers had agency in this system and none of the three earner groups were particularly privileged. Moreover, clients and Taskers could discuss prices and details, even outside of the platform. Although highly-reviewed Taskers benefited from the bidding system, less-experienced Taskers could use their profiles and private communications to advocate and establish rapport with clients. In 2014, the company overhauled the system in order to increase the volume of transactions. The new version — an app for smartphones and tablets — lists task categories and blocks off-platform communications. Pricing also went from per task to hourly. Though Taskers set their own hourly wage rates, the platform suggests ranges. Taskers are free to decline requests, but this lowers their acceptance rating and results in lower algorithmic priority. The new system also added a “quick assign” feature in which the algorithm suggests matches. Quick assign simplifies selection but it caters to instrumentales at the expense of sociales, who no longer have access to their established clientele. In addition, the system began to prioritize “Elite Taskers,” a designation which requires completing a significant number of monthly tasks and being in the top five percent of earners on the platform. Given their maximizing tendencies, this created an advantage for economici. Further signs of Tasker professionalization and nudges toward optimizing behavior include new features of the app, such as a dashboard titled “Your Performance,” which highlights “Needs Improvement,” categories such as measures of “Task Acceptance,” “Task Completion,” and “Response Rate.”⁴

Stocksy has also made efforts to monetize the work of its contributors. Originally, Stocksy recruited well-established photographers to join the platform. More recently, it is expanding its membership to increase and diversify the collection. The staff prioritizes selecting members who mesh well with the existing brand and elevate its aesthetic. In order to minimize competition among members, the staff tries to find “niche” photographers who can add to the collection without cutting into existing members’ profits. While staff attempts to accommodate individual situations, they did hold a vote to institute a cut-off for inactive members. The staff holds weekly editorial meetings and selects pictures to promote in the curated feeds, to highlight more marketable work. Archived data from the website suggests that management has steered the co-op towards more artistic or alternative works, to develop its market niche. In order to protect Stocksy’s brand, staff regularly reject submissions that fail to meet their standards. On the site, photographers have their “Assets” and “Followers” listed — metrics that serve as proxies for the quality or popularity of their work, similar to ratings on TaskRabbit and Airbnb. Photographers are given incentivizes to add large numbers of photos to the archive, but management still tolerates low activity, presumably because low earners do not incur a cost for other co-op members. When photographers experience high rejection rates or low sales, staff reach out to provide guidance and brainstorm solutions. Because Stocksy is a co-op, there are community-wide forums to voice issues, celebrate successes, and discuss the future of the platform. Homines economici reach out to other earners to encourage profit-maximizing and increase shared payouts. When these discussions become heated, staff may step in and “put a pin in them” to de-escalate tensions.

Airbnb has also made changes that are relevant to earner behaviors. In 2015, the platform introduced “smart pricing,” an optional tool that set rates automatically, in order to boost occupancy and earnings. The tool draws on more than 70 different variables to predict “shifts in the market,” such as adjusting for seasonal pricing.⁵ The platform also introduced “Instant Book,” a feature that lets guests book accommodations without waiting for host approval, not unlike TaskRabbit’s “Quick Assign” feature. Airbnb also started promoting a new insurance policy, thereby reducing guest screening and the need for rapport-building. Reviews have become more detailed, and now include six categories (location, cleanliness, etc.). The online system promises confirmation within 24 hours and measures host response rates to encourage this. Hosts who fail to reply within 24 hours have their response rates reduced, and the pending reservation request is automatically declined. Failure to reply also affects a listing’s search placement. Moreover, the response policy covers replies to messages during a stay, such as requests for information and help. While bookings can still be cancelled, doing so now incurs financial penalties. The company also sanctions hosts who fail to meet review standards. One of our participants — a homo socialis who reduces rates to lower guest expectations — received a warning from Airbnb that cited her ratings of four or less stars. Responsive and highly rated hosts can now earn the title of “Superhost,” which prioritizes their listings. Airbnb is also taking an increasingly interventionist approach as a mediator of exchange: sending frequent email reminders to hosts, prompting responsiveness to bookings, offering tips for increasing earnings, and sanctioning hosts who fail to meet standards. Hosts who wish to be favorably placed in search results now need to read the platform’s fine print and be diligent in their hosting behavior, a modus operandi that suits homines economici.

Notwithstanding the push towards professional standards on Airbnb, there is no meaningful discrimination against low-activity hosts. One can still be a Superhost with infrequent bookings. While designing the smart pricing tool, the company interviewed both hosts who depend on their Airbnb income and supplemental earners, to gauge pricing needs in relation to income goals. Airbnb’s tolerance of differing hosting needs is notable, because it means that homines sociales, the largest group in our Airbnb sample, can continue to choose guests based on personal preferences so long as responsiveness and ratings remain high. On the other hand, the push towards professional hosting standards may be altering guests’ expectations. Many of our participants complained that compared to early adopters, recent guests are less interested in social connection, and the platform’s actions may be exacerbating this trend. We also find that Airbnb’s more stringent demands are less compatible with a homo instrumentalis approach. However, automatic pricing and instant booking features do support this group’s hands-off approach to hosting. Economici in our sample took advantage of some streamlining features like instant booking, but continued to make individual calculations.

Surveying platform actions over the first decade, we find that platform earners are managed from afar, sometimes with a firm hand, but more often in subtle ways that sustain autonomy. For instance, Airbnb tells hosts how to increase their margins by using its pricing and booking tools, rather than helping them use the platform’s affordances for sociability. These efforts arguably nudge participants towards adopting a double-entry-bookkeeping perspective on hosting, but do not mandate it. Opportunities for maximizing are further enhanced by a growing list of auxiliary services associated with platforms. These include Taskers subcontracting out work, Airbnb hosts hiring professional cleaners, or Stocksy artists employing assistants. The sociales and instrumentales in our sample might resist pushes towards maximizing behavior especially on Stocksy and Airbnb, where occasional participation is still acceptable. For economici, there may be a growing tension between agency and efficiency as more processes become automatic.

Larger, external factors are also affecting platform changes, such as regulatory policies and competition from other companies. On-demand services have seen increased pressure from labor activists and politicians to classify workers as employees, in order to grant them essential rights that independent contractors lack, such as a minimum wage and unemployment benefits. This is not an immediate threat to any of our platforms, but competitive pressures and market conditions are relevant to all three. TaskRabbit faces strong competition from other on-demand labor sites, and has moved away from deliveries, at least in part due to the emergence of major courier and food delivery apps. Though Stocksy’s “boutique” aesthetic initially set the co-op apart from Getty and Alamy, those industry giants have begun to imitate the Stocksy brand. In an attempt to counter these moves and gain economies of scale, Stocksy partnered with Adobe Stock in 2017 and increased membership in order to meet the demands of an expanding clientele. However, this has resulted in unintended competition among Stocksy photographers as more members enter existing niches. Airbnb, by far the largest of our three platforms, grew in part because it resided for years in a gray area between work, subletting, and “sharing.” This ambiguous classification helped market actors escape taxes on their income, making it more profitable and attractive, which in turn boosted the company’s capacity for expansion. Moreover, the company’s dominant position in the home-sharing market gave it ample time to experiment with how to operate a platform with a plurality of participant motivations. Increased regulatory pressure might change this. In 2019, Airbnb commenced automatic collection of State Sales Taxes and Local Occupancy Taxes in Massachusetts, where our participants reside, and similar initiatives have been implemented elsewhere. The platform’s success has also encouraged a number of resourceful challengers, including Marriott International.

7 Conclusion

How can we theorize these digital platforms? We have argued for seeing them as hybrid entities that give earners more control over their actions than conventional firms, but which also exercise distinct mechanisms of control. We found that multiple approaches to earning co-exist within a platform: homo economicus, homo socialis, and homo instrumentalis. We then asked, is the continued presence of the latter two groups, who typically work and earn less, sustainable for the companies? To answer that question, we reviewed relevant policies and platform affordances. We found that all three companies have instituted changes that nudge providers in the direction of a maximizing orientation, with TaskRabbit being the most aggressive in this regard. However, their behaviors are nudges, rather than directives. None of the three have made it impossible for lower-activity participants, or homines sociales and instrumentales to continue on the platforms. The platforms appear to be tolerating the diversity we have identified. This suggests that the hybrid designation, as theorized by Watkins and Stark, Kornberger et al, and Vallas and Schor, is robust to the presence of multiple earning orientations.

That said, there is some movement in the direction of promoting the modus operandi of homo economicus, via platform design, management practices, and the broader ecosystem of auxiliary services. Many of the market devices that modify behavior in our three cases — response rates, popularity metrics, elite status and curated ordering of entities for sale — are standard in the gig sector. A central question for further research is how will earners respond to these converging trends? In particular, what happens to those who feel that the platform economy no longer works for their original goals? Do they assimilate; do they seek out alternative platforms; or do they give up on platform work altogether? And as these and related platforms grow into their second decade, will regulatory and market forces make it difficult for them to host the heterogeneous earners that we have found in the first decade?