Governing digital labor platforms

Context, goals, and strategies

Remarks for the event “The Platform Economy: The Impact on the Labour Market and Practical Solutions” hosted by the S&D Group of the EU Parliament

The Spring House, De Ruyterkade 128, Amsterdam
7 September 2018 19:30

M. Six Silberman, IG Metall

Questions from the event organizers

1. Why is the rise of the platform economy such a new phenomenon, and how does the platform economy function?

A. How new is it?

The so-called platform economy is a continued development of existing trends: outsourcing, digitalization, and globalization.

The first internet-based “digital labor platform” may have been Elance, which was founded in 1999. Elance continues to exist after its merger with Odesk. Together they became Upwork. Upwork may be the largest English-language freelance platform.

iStock, formerly iStockPhoto, was founded in 2000, and might also be described as a digital labor platform, although it works differently from newer websites that are more commonly described as digital labor platforms.

Craigslist was started in 1995 or 1996, depending on how you want to count, and has long had a “gigs” section.

The first platform to be described explicitly as a crowdsourcing platform, in which the workers were largely anonymous and algorithmically managed, may have been Amazon Mechanical Turk, launched in 2005.

But all of these developments are only digital variations of temporary labor arrangements that have existed for hundreds of years, including day labor, gray markets for domestic labor and sex work, and white-collar freelance work.

What is new about digital labor platforms is automated or algorithmic management, in which part or in some cases all of the tasks of matching workers and clients, managing the work, and evaluating work and workers are performed automatically or significantly facilitated by software.

B. How does it work?

We can talk about how digital labor platforms work by talking about the parties, the process or model, and the legal relationships between the parties.

Generally in a transaction that takes place on a digital labor platform there are three parties: the worker, the client, and the platform operator.

There are different models or processes by which these parties interact and work gets allocated and performed. These names are not in any way “official.” I have just made them up.

• Task model: The client posts a task to the platform and does not generally spend a lot of time screening individual potential workers. Generally, the client does not care who does the work as long as it is done to the required specifications. This may mean the client specifies general criteria the worker must meet, or that the platform allocates tasks to workers directly. “Crowdsourcing” platforms such as Amazon Mechanical Turk, Clickworker, and Prolific can be described with this model. So can platforms that provide in-person services, especially for transportation and delivery, such as Uber and Deliveroo, where the client usually simply wants the service to be provided as quickly as possible and does not expect to choose between potential workers. On “crowdsourcing” platforms, a single task may be completed by many workers. Workers may be partially anonymous to clients, known only by an ID number or email address.
  
  
• Freelance model: The client posts a task for a single worker or a small group of workers to complete, and then invites applications or proposals from workers. Workers may seek to win a job by offering a low price and/or by signaling that they will provide high quality work. Workers may post profiles publicly to the platform, and clients may browse and search through these profiles and invite specific workers to apply for a posted job, or even directly offer a worker a job without having first posted it publicly. What distinguishes “freelance platforms” from “crowdsourcing platforms” following a “task model” is the focus on screening and choosing individual workers. Platforms following a freelance model include Upwork and Freelancer.com as well as platforms for in-person work where a higher degree of trust is required than in transportation and delivery, including platforms for domestic work (e.g., Helpling), care work, and sex work.
  
  
• Contest model: The client posts a task and many workers are invited to complete it. The client picks the submissions they like, and the workers who made those submissions are paid. In some cases, only one worker is paid. This model is common in the graphic design sector and is a digitalization and globalization of the existing practice in the graphic design and advertising industry of the practice of “pitching.” 99designs is the most well-known English- and German-language platform to use this model, but other platforms such as DesignCrowd and GoPillar (formerly CoContest) also use it. It is also used in engineering (e.g. Local Motors) and scientific research (e.g. InnoCentive).

iStock uses a different model.

On most (but not all) digital labor platforms, workers are required to agree that they are self-employed persons or “independent contractors” — not employees of the platform operating company or of any client.

As a result, they have none of the protections or rights of employees: no minimum wage; no employer contributions to health insurance or social security; no sick pay or paid vacation, overtime, or parental leave; no protection from dismissal (indeed the worker is “dismissed” by default at the end of every task and can be prevented from accessing future tasks through account “deactivation” at any time and often for any reason); no co-determination rights; and often no right to organize and negotiate collective agreements with the platform operator or clients.

There are some variations on the basic “three-party” structure. For example:

• In delivery platforms, there are four parties: the worker, the “end customer” who receives the delivery, the business customer where the worker picks up the item to be delivered (e.g., a restaurant or logistics retail company), and the platform operator.
  
  
• Many Airbnb hosts use platforms such as TurnoverBnB to outsource cleaning and other property management tasks. In this case there are five parties — the host, Airbnb, the guest, the cleaner, and the cleaning platform (e.g., TurnoverBnB) — although there is no transaction to which they are all party; the host interacts with the guest via Airbnb and with the cleaner via the cleaning platform.

C. Where does the “platform economy” end and what we might call “automated management” as a more general phenomenon begin?

There is a lot of talk in the media and among researchers and policy makers about the “platform economy” and digital labor platforms because they are relatively easily recognized as enabling a “new” and “non-standard” form of work arrangements combining automated management with, usually, self-employment and, sometimes, global real-time competition for work — often on price.

Yet the centrality of automated management to the understanding of “what’s new” about digital labor platforms is often underappreciated. Policy discussions about Uber, for example, often focus on the potential for autonomous cars to put millions of taxi (and Uber) drivers out of work. But the current business model is not, largely, to automate or deskill service provision but rather to centralize and automate management — and, as a result, to establish market power. Uber replaces thousands or tens of thousands of taxi dispatchers working all over the world, mostly for small, local companies, with a few hundred or a few thousand programmers and specialists, working mostly for Uber in San Francisco, tasked with automating management. This principle is not limited to digital labor platforms or to freelance arrangements: it can be, has been, and is being implemented and experimented with within companies as well.

2. What kind of opportunities do platforms bring?

The main opportunity digital labor platforms bring is access to work.

Platforms for remote work provide opportunities for paying work for workers who are not able to work outside the home — because they cannot leave the home or sustain a traditional job due to care obligations, illness, or lack of transportation; because they face discrimination in their local labor markets; or because there are simply no adequate opportunities in their local labor markets.

Platforms for in-person work such as transportation, delivery, and domestic work may provide labor market access for people newly arrived in a new country who may have difficulty securing traditional jobs. In Germany, the United Kingdom, and the United States, for example, many workers on transportation and delivery platforms are immigrants.

Digital labor platforms of all types may offer workers with a low-paying main job as well as students to relatively time-flexibly earn supplemental income.

The time flexibility of platform work, however, should not be overestimated: most workers who rely on platform work as a main income source report working long hours for low and unpredictable wages, and that if one wants to earn a living via platform work, one must be prepared to work whenever the most high-paying work is available.

3. What kind of uncertainties are workers in the platform economy facing?

Some uncertainties faced by workers on many digital labor platforms are:

• Volume of work: With a few exceptions (for example in personal transportation), most platforms have periods when there is no work or only poorly paid work. These periods are not always easy to predict.
  
  
• Pay: Because most platform workers are required to agree that they will be legally classified as independent contractors of self-employed persons, not employees, they are not entitled to minimum wage. Many platform workers earn less than minimum wage in their locations. Especially workers for whom platform work is the main income source, therefore, may have difficulty meeting their basic needs with this income.
  
  
• Risk of unfavorable ratings or other algorithmic disciplinary measures: Many digital labor platforms ask, or even require, clients to rate workers after work is performed. Few platforms, however, provide clear guidance regarding how work or workers should be rated or offer workers the ability to contest ratings they perceive as unfair. Negative ratings, however, can significantly influence a worker’s ability to get future work. A string of negative ratings can even lead a platform to deactivate a worker completely, preventing them from receiving any work via the platform temporarily or even permanently. Some platforms make such decisions algorithmically, without human oversight or without a reliable mechanism through which a worker can present their side of the story.
  
  
• Equipment-related costs and risks; insurance costs: Because most platform workers are self employed, not employees, they must bear equipment purchase and maintenance costs as well as the costs of health insurance, social insurance, and other insurances on their own. Especially for workers in delivery and transportation sectors, equipment costs may be high and equipment failures may present significant economic threats. In deliver and transportation as well as domestic work and some kinds of remote work (e.g., content moderation), the work may be associated with significant physical or psychological hazards which the worker must insure against on their own — if they can.
  
  
• Client non-payment: Some platforms allow clients to refuse payment for completed work, and may not provide workers with adequate mechanisms to contest such non-payment.

4. Who is responsible for bearing the burden of the uncertainties?

Right now, with a few exceptions, the workers. Two exceptions:

• Uber and AXA provide accident insurance for drivers in some European countries
  
  
• The German writing platform content.de automatically transfers a “client contribution” from customers to the German “Artists’ Social Insurance” (“Künstlersozialkasse”) on behalf of its self-employed writers

5. How can these responsibilities be shared?

Two major issues are payment and social insurance. Worthwhile goals could be:

• To ensure that all platform workers earn at least minimum wage in their locations (or, if there is no minimum wage, a living wage or a wage equal to that provided by the relevant collective agreement); and
  
  
• To ensure that clients pay a “client contribution” into appropriate social insurance schemes (e.g., health insurance, retirement or pension schemes, long term care insurance).

Additionally, it could be desirable to establish various rights for workers, including for example the right to protection from arbitrary account suspension or deactivation; the right to protection from arbitrary nonpayment; the right to contest unfavorable reviews; and so on (see more below).

Platform operators could provide clearer guidance to clients rating workers and enforce stronger oversight regarding unfavorable ratings.

How exactly these goals should be achieved however is an open question, given the uncertainty of some platform workers’ legal status (i.e., self employed or employee) and the importance of that status for legal entitlement to protections such as minimum wage and social insurance as well as workers’ right to negotiate collective agreements on matters such as pay or rights regarding account deactivation and ratings.

Worker groups, unions, policy makers, legal scholars, and other worker advocates — in some cases in cooperation with platform clients and operators — are developing a variety of strategies, some of which are mentioned below.

The EU General Data Protection Regulation may provide relevant rights for platform workers based in the EU. If the proposed Regulation on Promoting Fairness and Transparency for Business of Online Intermediation Services is passed, that too may provide EU-based platform workers relevant rights.

6. How can we tackle the downsides in the platform economy?

It might be desirable for all parties in the “platform economy” for policy makers to clarify the boundaries between self employment and false self employment or to otherwise attempt to resolve the legal uncertainty regarding platform workers’ rights — and platform operators’, or clients’, responsibilities — resulting from their sometimes unclear legal status.

This may however present both legal and political challenges. It may therefore also, simultaneously, be desirable to seek to secure rights for all platform workers, regardless of their employment status. A list of possibly desirable rights follows.

A. Possibly desirable rights for all platform workers, regardless of employment status

1. Right to minimum wage in the worker’s location (or, if there is no minimum wage, the wage provided by the relevant collective agreement; or, if there is no relevant collective agreement, a living wage)
2. Right to explanation for declined account application
3. Right to proection from arbitrary account suspension, closure, or deletion
4. Right to advance notification and explanation of account suspension, closure, or deletion
5. Right to clarify or correct alleged violation of platform terms of service leading to account suspension, closure, or deletion
6. Right to contest account suspension, closure, or deletion; right to mediation
7. Right to prompt payout of funds in the event of account suspension, closure, or deletion
8. Right to complete work history in the event of account suspension, closure, or deletion
9. Right to information regarding nonpayment conditions; right to protection from arbitrary nonpayment
10. Right to explanation for nonpayment
11. Right to contest nonpayment and request review from platform operator
12. Right to further contest nonpayment; right to mediation
13. Right to “redo” rejected (i.e., nonpaid) work
14. Right to relevant client information
15. Right to accurate, fair, transparent, fit for purpose evaluation, rating, and qualification
16. Right to comprehensive work history
17. Rights regarding potentially psychologically harmful tasks:
    • Right to receive clear information, prior to accepting a task, that a task may be psychologically harmful, and why; and
    • Right to receive qualified psychological support and counseling services at the cost of the client and/or platform after having completed such tasks.
18. Right to organize and negotiate collective agreements (with platform operators and/or clients, as appropriate)
19. Right to discuss pay, work processes, clients, tasks, and working conditions (i.e., restriction of nondisclosure agreements)
20. Right to participate in platform governance according to national labor law (i.e., co-determination)

B. Establishing the rights

• Minimum wage: Absent effective policy action to guarantee all platform workers minimum wage, progress could be made on this front by initiating a voluntary client pledge. Clients could pledge to pay all workers completing their tasks (satisfactorily, if the platform in question allows nonpayment) at least minimum wage in the worker’s location. Tasks from clients who have signed the pledge could be displayed more prominently, either through direct participation of the platform operator or through the development of third party software. Disputes could be resolved via a mechanism similar to the existing “Ombuds Office” of the “Crowdsourcing Code of Conduct.”
  
  It is not clear what percentage of clients would be willing, freely or under public pressure, to make such a commitment, but preliminary discussions suggest that at least in some important areas — for example, research tasks posted to crowdsourcing platforms such as Amazon Mechanical Turk and Prolific — a significant number of clients would be prepared to do so. Once the pledge was in place, “mid-level” mechanisms and institutions such as industry collective agreements and university ethics regulations could encourage or even require companies and universities to commit to participating.
  
  An example of such a pledge in the domestic work sector can already be seen in the “Living Wage Pledge” for self employed cleaners of Airbnb properties, administered jointly by Airbnb and the US National Domestic Workers Alliance.
  
  
• Procedural rights (rights 2 through 17, 19, and 20 above): In general, these procedural rights do not appear to impose significant costs on platform operators and do not appear to elicit significant principled objections from “reasonable” platform operators. The challenge is to establish them as a priority for platform operators. This can be done via a combination of informal social dialogue (e.g., the cooperation between platforms and trade unionists in evolving and enforcing the “Crowdsourcing Code of Conduct”); establishment of standards via bodies such as DIN, CEN/CENELEC/ETSI, and ISO; application of existing laws (in the EU context, especially the General Data Protection Regulation and potentially the proposed Regulation on Promoting Fairness and Transparency for Business Users of Online Intermediation Services); negotiation of collective agreements; and establishment of new national and international legislation and labor conventions.
  
  
• Right to organize and negotiate collective agreements: There is currently some legal uncertainty around the right of self employed platform workers to negotiate collective agreements (e.g., regarding the rights listed above) with platform operators and/or clients. Policy action may be required to clarify the conditions under which such agreements would (and would not) violate competition law.

From a trade union perspective, of course, organizing is seen as an essential prerequisite for encouraging various actors, including clients, platform operators, and policy makers, to take the steps necessary to establish and safeguard these rights.

■