Should Online Platforms Have Unions?

In October 2019, two DoorDash drivers — Dave Levy and Nikos Kanelopoulos — launched the #DeclineNow Facebook group.

The duo had discovered that when a DoorDash driver declines a delivery, the app offers that delivery to another driver for higher pay. In the Facebook group, which now numbers more than 30,000 members, they urged peers to reject any delivery that doesn't pay at least $7 — more than double the base rate of $3.

“Every app-based, on-demand company’s objective is to constantly shift profits from the driver back to the company,” Levy explained. “Our objective is the reverse of that.”

On September 1, 2021, many Twitch streamers participated in a coordinated protest — taking the day off streaming in response to the platform’s perceived inaction against harassment of marginalized creators.

Platform viewership dropped by an estimated 5% to 15%. Though the protest lasted only a day, it garnered widespread attention: Many news outlets reported on the harassment problem, and #ADayOffTwitch became one of the top 10 trending hashtags on Twitter.

These two protests reveal an overlooked truth about the platform economy: Despite real differences in their jobs, both gig workers and content creators are reckoning with the fact that their livelihoods depend on the actions and algorithms of platforms that they have little to no ability to sway. Furthermore, they have little recourse to suboptimal policies, dissatisfactory product decisions, and other negative experiences.

In the face of this, a new form of collective labor activism tailored to the gig and creator economies is emerging — what we call decentralized collective action (DCA). This encompasses worker-led movements from insurance pools created by Jakarta rideshare drivers to an informal union of global musicians and music workers.

These efforts tend to be more bottom-up and diffuse than historical labor unions. Workers find and team up with each other in a peer-to-peer fashion; voice opposition on social media, through online forums, and to the media; undermine or challenge a platform’s normal operations; leverage the power of their audience or customers; and at times, leave platforms altogether for more worker-friendly alternatives.

Though individual actions have achieved moderate success, DCA has largely struggled to have a sustained impact. Because platform workers often are not in direct contact with each other, it is hard for them to coordinate, and platforms can induce a competitive dynamic among them.

Moreover, platform workers typically have little leverage because of low barriers to entry and an excess of willing participants on the supply side. These issues heighten the necessity for a more refined approach from workers, and a more sustainable path forward.

This essay outlines the strategies that platform workers are utilizing to voice their concerns through DCA today, and lays out a roadmap for how to achieve outcomes that better serve the interests of all stakeholders: participants, platforms, and end users.

A New Social Contract Between Workers and Platforms

The world of platform labor is in principle a net good: Platform marketplaces can substantially improve social welfare by enabling new or improved transactions.

This leads to new forms of labor activity, expanding workers’ options. Those whose niche interests and skills would have made it challenging to make a living from local customers can now turn to a global marketplace. Someone whose main skill is woodworking, for example, can grow a worldwide audience of fans through YouTube and then sell their creations on Etsy.

But that doesn’t mean platform workers don’t have grievances — or that the contract between platforms and their workers can’t be improved. Platform workers’ liminal status makes them much more vulnerable to exploitation. In the United States, most gig workers and creators are classified as “independent contractors,” a categorization the IRS defines as “people who offer their services to the general public” in an independent trade, business, or profession.

In practical terms, this classification absolves firms from having to provide these workers with benefits, protections, and guarantees that traditional employees enjoy — even as platform workers and creators are dependent on platforms to reach audiences, connect with potential customers, and earn income.

In an ideal world, long-term platform sustainability would supersede short-term profit incentives, motivating platforms to align with workers to help them grow their businesses and — eventually — obtain a meaningful standard of living and success.

Platforms could even set prices and policy to shift more value towards workers and enable them to invest in higher-quality service, which tends to grow engagement and create more value for both workers and the platform in the long run.

Yet instead, most platforms chase short-term gains to boost growth and attract outside investment. As platforms’ network effects have intensified and resulted in significant market power — called monopsony power in the context of labor markets — their workers have struggled to find recourse.

Historically, collective action has emerged to solve for this misalignment of incentives. These movements have centered on workers organizing and negotiating with companies directly or working to promote broader social change. In America, labor unions grew out of the Industrial Revolution of the late 1800s: Unions fought for better wages, shorter hours, and safer working conditions.

At the turn of the 20th century, these labor movements culminated in many of the legal protections workers enjoy today, including the National Labor Relations Act of 1935, which provides the right for workers to unionize, and the Fair Labor Standards Act of 1938, which created the right to minimum wage and overtime pay and established the 40-hour work week.

But with platforms, collective action is more challenging.

Not only do today’s platform-based workers exist in a nascent labor category, but their participation is decentralized, making it harder for them to connect with each other. As the industry has matured, and consolidation has led to monopsony power, platform labor has reached a turning point.

Platforms are increasingly governing participants’ opportunities and livelihoods, and with that comes the need for mutual accountability and evolving protections and responsibilities. It’s time to revisit the social compact between platforms and their workers — and discover a new form of collective action to do so.

Why Do Workers Need Decentralized Collective Action?

There are several reasons why workers are now turning to decentralized collective action: Right now, tensions between workers and platforms are centered around some familiar labor issues such as the right to unionize, as well as more platform-specific issues such as ownership of data, content moderation, ability to reach customers, harassment in virtual “workplaces,” and monetization policies.

Platform Lock-In and Leverage Over Individual Workers

Platform workers often face substantial lock-in, driven by network effects and lack of data portability. Platforms intermediate transactions, and as a result, they gather market data and often control customer relationships.

This type of capital cannot be easily ported over to other platforms or worker-owned properties, which means that platform labor is reliant on platforms for work — leading to worker dissatisfaction with pay rates and monetization models, income instability, and anxiety and burnout.

In other words: Because a DoorDash driver can’t find delivery opportunities on their own and a Twitter user can’t export followers’ emails, DoorDash and Twitter have market power.

Worker Classification Doesn’t Reflect the Evolving Nature of Work and Independent Contractors are Unable to Unionize

The “independent contractor” classification not only blocks platform workers from benefits and protections, but also means the platform workers are not covered by the National Labor Relations Act, and thus do not have the right to unionize.

Yet platforms nevertheless exercise significant amounts of control over aspects of workers’ jobs, including determining how much workers are paid, what they do, and how they perform their jobs. This has led to contentious debates and regulatory battles around platform worker classification.

Platforms Often Have an Incentive to Commoditize Their Workers as Much as Possible

From the platform’s perspective, commoditization of workers is desirable, as it allows the platform to provide a uniform customer experience and remain the center of the customer relationship. For example, the TikTok app design discounts the value of a follower relationship and instead defaults users to the “For You” page — an algorithmically generated feed of content the platform believes would appeal to users.

On rideshare apps, driver commoditization ensures a consistent level of service, but also means that users come back to the app to request rides, rather than seeking out a particular driver directly. The commoditization of suppliers on a platform erodes the ability of workers to establish their own businesses or to operate outside of a handful of platforms.

Worker Heterogeneity Makes it Challenging to Organize for All

For creator platforms, the power law distribution of success means that top creators have disproportionately more bargaining power with platforms.

Often, this results in top creators receiving special treatment from platforms, in the form of greater access to funding, more favorable take rates, prominent placement in discovery channels, participation in product feedback processes, and inclusion in funding and monetization programs.

Over time, platforms can trend towards catering to a small segment of top creators, who have little incentive to push for improved conditions for all creators because they themselves are benefitting from the platform’s design and policies.

In other words, the heterogeneous nature of creators makes it challenging for the most powerful creators to have the motivation or cohesion to organize. In contrast, labor movements typically harness the collective power of large workforces who have similar shared experiences.

Voice and Exit: Strategies for Decentralized Collective Action

So what options do platform workers have? The Exit-Voice-Loyalty framework, first described by economist Albert O. Hirschman, describes how individuals react when dealing with dissatisfaction in firms, organizations, and states.

In traditional labor environments, dissatisfied employees can voice concerns in an attempt to change their situation, exit to seek new employment opportunities, or passively wait, out of loyalty or neglect, for the situation to resolve:

An early study examining platform labor’s options in the face of work dissatisfaction was a 2015 study of Amazon Mechanical Turk, a marketplace for on-demand tasks. Applying the exit-voice-loyalty framework, researchers found that:

... participants have the possibilities of loyalty and exit (e.g., petitions, boycotts), but not voice. In other words: they may choose to sign a petition or to leave if they don’t agree [with platform policy]. But, there is little place for discourse when the problem, its source, and its solution are not clear.

Since then, we’ve seen numerous examples of platform worker voice emerge — albeit expressed differently than direct feedback to managers or filing a complaint with an HR department.

Instances of platform worker voice include worker-to-worker, worker-to-algorithm, and worker-to-public communication — all of which happens in a bottom-up, decentralized fashion.

The following represents a sampling of different platform worker DCA voice strategies we have observed:

1. Informal Unionization (worker-to-worker) is when workers engage in union-like behavior, for instance by coordinating and submitting a list of demands to a platform.

Examples include the Twitch protest mentioned above; the Black creators TikTok strike; the Union of Musicians and Allied Workers’ (UMAW), which mobilizes music workers to fight for fairer deals with streaming services and record labels; the Instagram Meme Union, comprised of a group of meme creators demanding greater transparency in communication with Instagram; and the Freelancers Union, a nonprofit advocating on independent workers’ behalf.

2. Mutual Aid (worker-to-worker) occurs when workers engage in reciprocal support among themselves.

Examples include Instagram pods that agree to mutually like, comment on, share, or otherwise engage with each other’s posts; rideshare drivers in Jakarta that have formed physical basecamps (“mutual aid stations”), and informal insurance pools.

3. Third-Party Product Enhancement (worker-to-worker) refers to the design and creation of digital tools that improve the worker experience.

For example, Driver’s Seat Cooperative helps drivers track and optimize their mileage vs. their payouts.

4. Information Leveling (worker-to-worker) occurs when workers pool learnings or unlock new or hidden information to help one another better navigate opaque platform work environments.

Examples include FYPM (for creators) and Turkopticon (for Mechanical Turkers), which are both examples of Glassdoor-like platforms which have been independently spun up by workers to aggregate reviews of employers in the absence of a platform-native alternative.

5. Algoactivism (worker-to-algorithm), a term coined by Stanford and MIT researchers, refers to a growing set of tactics used by workers to resist the managerial control increasingly exercised by algorithms.

Examples include the DoorDash #DeclineNow movement, as well as similar attempts to subvert algorithmic control that have been observed on platforms like TikTok, Uber, Airbnb, Fiverr, and TaskRabbit.

6. Public Media Campaigns (worker-to-public) happen when individual workers share instances of mistreatment or frustration on social media.

Examples include when @deliveryguy100 went viral on TikTok in early June 2021, receiving 1.2M views on his video describing the realities of being a delivery driver. Similarly, popular vlogger Hank Green shared a TikTok video expressing questions about TikTok’s compensation of creators.

While these strategies for decentralized worker voice can be impactful in the short term, the lack of evolution in platform policy in recent years indicates they are ineffective methods for instigating long-lasting, meaningful change.

As Dawn Gearhart, policy coordinator for Teamsters Local 117 and a Seattle labor organizer, explained: “Unions cannot collectively bargain with an algorithm, they can’t appeal to a platform, and they can’t negotiate with an equation.”

Media campaigns have limited lifecycles, and platform responses are often performative rather than substantive. Algoactivism only provides a temporary fix until platforms modify their codebases and close loopholes.

Building out new tools to help workers in the context of their existing platforms requires substantial time and effort, and again is subject to disruption by changes in platform design. And informal unions lack the scale and organization needed to catalyze lasting change.

This doesn’t mean decentralized collective action can’t be effective; rather, the efficacy of action through voice within the existing platform ecosystem is limited.

Returning to the exit-voice-loyalty framework, the alternative to voice is to exit. If voice entails taking actions to improve conditions within the existing environment, exit occurs when individuals or groups resolve that conditions are beyond improvement, that it is better to make a break with existing systems and to seek work elsewhere. The platform labor movement then may be more successful using DCA to build a set of more worker-friendly platforms that disrupt the existing ecosystem — and thus enable exit from it.

The Future: Worker Ownership and Enfranchisement

It is ultimately through a combination of regulation and organizational restructuring that workers shift the balance of power in a sustainable way. Here are some emerging efforts in that direction:

Platform cooperatives that institutionalize worker voice

Platform cooperatives are the tech-native version of cooperative organizations: platforms that rely on democratic decision-making and are owned by their workers and users. The Driver’s Cooperative, a New York City-based ride-hailing app that is entirely worker-owned and governed, is one such example.

Founded in 2020, it has more than 3,000 drivers and 30,000 users. Stocksy is a cooperatively-owned marketplace for stock photography and video, which pays out more than half of its revenue as royalties to its contributors.

Drawing on principles from offline cooperatives such as Mondragon (the largest worker-owned cooperative in the world, with 81,000 workers and €12 billion revenue in 2015) and REI (a consumer cooperative with 20 million members and $2.75 billion revenue in 2020), platform cooperatives are often funded through a combination of their members and outside debt financing, with profit distributed according to the wishes of the member-owners.

However, this ownership structure and member-centric governance has made it difficult for them to attract outside capital — hindering the ability of platform co-ops, and co-ops in general, to be competitive with traditional corporations.

Cryptonetworks and Decentralized Autonomous Organizations (DAOs) as the Future of Internet-Native Co-Ops

Cryptonetworks provide a promising alternative: Decentralized organizations that no single entity controls, which facilitate trust among participants through hard-coded rules. In these networks, ownership is distributed to all stakeholders via a native cryptocurrency or “tokens,” that reward actions that contribute to the network’s success.

Such networks address two major issues that have prevented traditional cooperative businesses from becoming pervasive: access to capital and complexity of governance. With the issuance of a token, cryptonetworks can benefit from market speculation and raise capital that enables them to be competitive vs. traditional corporations.

Experimentation around internet-native governance endeavors also enables diverse member bases to coordinate decision-making at scale: Token-based voting, for example, lets members vote in proportion to their ownership of the platform, similar to the way in which traditional shareholder voting rights exist in corporations today.

Reputation voting, meanwhile, allows users to participate in governance based on their perceived value as community members rather than their economic holdings

The digital art marketplace SuperRare recently launched a token to decentralize curation and oversee a treasury that is collected from platform commissions and fees.

Decentraland is a virtual world owned by its users, who can make collective decisions on the future of the virtual world. Yield Guild Games — a gaming guild that trains and onboards players into play-to-earn video games — can be seen as the crypto-native version of a workers’ union; its large contingent of gamers allows it to negotiate for better platform policies and design.

Labor Laws to Protect Platform Worker Rights

In parallel with technology-based solutions like cryptonetworks, there is the potential for labor laws to evolve to address the unique needs of platform workers. Regulation, historically, has codified and advanced the rights of workers.

Analogous to existing regulations around minimum wage and overtime pay, there could be regulation of earnings and revenue share rates for creators and gig workers in the platform economy.

There has been much debate around the employment classification of gig workers and whether they are entitled to a minimum wage. California’s Proposition 22, which carved out rideshare workers from being classified as employees, was recently ruled unconstitutional for limiting the ability of workers to organize and have access to workers’ compensation.

There could also be regulation to promote data portability and ownership: Creators’ and users’ ability to port data would mean they could migrate between platforms more easily or set up their own independent properties.

As one of us (Jin) explained in a recent blog post: “Creator and user ownership of data, relationships, content, identities, and interactions would weaken platforms’ lock-in and entail a shift in power from platforms to their participants, enabling them to operate outside of a handful of platforms.”

We are aware, however, that regulation can and often does have unintended consequences, potentially strengthening incumbents’ market position. Regulations around payment rates and creator compensation could favor giant tech platforms that have deep pockets vs. newer startups.

Data protection regulation such as General Data Protection Regulation (GDPR) has been criticized as hindering innovation and competition by making it more challenging for new entrants to collect and share data that would enable workers and consumers to have more options.

Building Worker Power

In the aforementioned examples, users — in a decentralized manner — take action to shift their usage to new networks entirely of their own volition, motivated by the benefits over centralized alternatives (social media platforms, traditional financial institutions).

Users’ participation in cryptonetworks is not only driven by the desire for more governance and enfranchisement, but also by financial self-interest, given the potential upside of tokens.

That points to a basic truth: New alternatives will succeed at scale only when they can be genuinely — and holistically — better for workers. And that, in turn, sets into motion a positive macro flywheel, pressuring incumbents to evolve their policies and products to favor platform participants.

In the long run, platforms have a lot to gain from becoming more worker-friendly. As one of us (Kominers) recently argued in the context of both delivery and short-term housing marketplaces, by investing in improving their suppliers' operations and outcomes, platforms improve the quality of their networks in the long run, and increase the overall opportunity for marketplace-based transactions.

Decentralized collective action helps us move in the right direction — both by influencing current platforms, and by forging the next generation of disruptive networks that are more aligned with their participants.

In the coming months and years, as creators and workers realize their collective power, these movements will grow in number. And as demonstrated throughout history — from the French Revolution to the growth of Wikipedia — the power of the distributed many can greatly outperform the power of the hierarchical few.