The Definition of ‘Digital Labour Platform’ in the Proposed Platform Work Directive

On 9 December 2021, the European Commission announced its proposal for a Directive on improving working conditions in platform work—the ‘Platform Work Directive.’ The Directive’s main goals are to reduce false self-employment among persons performing platform work, to regulate algorithmic management on digital labour platforms, and to provide legal certainty for platforms. The Directive is now in interinstitutional negotiations (‘trilogues’). Others have observed that the Council’s position has weakened the Directive significantly (see the upcoming blogposts by Kocher and Barrio).

This essay, however, focuses on an element of the proposed Directive that has gone relatively unremarked in the scholarly and policy debates so far: the definition of ‘digital labour platform.’ It appears from the preparatory materials—and from the press release announcing the proposal—that, in drafting it, the Commission mainly had in mind the most visible and widely-discussed type of platform work: on-location work, such as transportation and delivery. And it seems clear that the Directive’s definition includes platforms mediating this type of work. It is however not clear whether other important types of platform work are also included.

The diversity of platform work

We can identify at least five main types of platform work: on-location platform work, online ‘microtasking’; online freelancing; content marketplaces; and contest-based platform work. These different types of platform vary in the type of work they mediate and also in the methods they use to assign—or provide access to—and remunerate work. In ‘microtasking’ platforms such as Amazon Mechanical Turk, Clickworker, and Microworkers, large projects are typically broken down into smaller tasks, which are made available to anyone meeting pre-established criteria. Payment is made per task. In online freelancing platforms such as Upwork, in contrast, clients typically interview applicants who bid to work on larger projects. Freelancers often clarify clients’ needs and establish ‘milestone’-based payment plans.

Content marketplaces such as YouTube and iStock work differently. Here, ‘content creators’ post content directly to the platform. They typically choose what content to create not in response to explicit requests attached to offers of payment, but rather by extrapolating past content consumption trends. A variety of payment models may be involved, including advertising-based models in which the relationships with advertisers are managed entirely by the platform.

Finally, the ‘contest-based’ model is used by platforms for design and innovation, such as the graphic design platform 99designs. In this model, a client posts a ‘brief’ setting out criteria for a particular design output, such as a logo or website design. Many designers are invited to submit designs, and the client can provide feedback. At the end of the process, one or a few winners are selected. Only these winners are paid.

And within on-location work, platforms can be classified further by the type of work they mediate. In addition to transportation and delivery, platforms play a growing role in mediating janitorial work in offices and households, domestic work including home repair and cooking, and care work, such as childcare and eldercare.

It appears that all five types of labour platform are growing in size and economic importance. Microtask platforms, for example, have become a key part of the supply chain for ‘artificial intelligence’. And during the Covid-19 lockdowns, viewers at home turned to YouTube—possibly the world’s largest digital labour platform—for educational content, as schools ‘went online’; for public health and mental health information; and for ‘soothing’ content.

However, the Directive’s definition may not capture all five types. In particular, it may exclude content marketplaces.

The Platform Work Directive’s definition(s) of ‘digital labour platform’

The original Commission proposal set out this definition for ‘digital labour platform’ in Art. 2(1)(1):

‘digital labour platform’ means any natural or legal person providing a commercial service which meets all of the following requirements:

(a) it is provided, at least in part, at a distance through electronic means, such as a website or a mobile application;

(b) it is provided at the request of a recipient of the service;

(c) it involves, as a necessary and essential component, the organisation of work performed by individuals, irrespective of whether that work is performed online or in a certain location

The Council position (p. 38) adds a fourth criterion:

(d) it involves the use of automated monitoring or decision-making systems

The criterion in Art. 2(1)(1)(a) is met by all five types of digital labour platform. The new criterion (d) also seems likely to be met in most cases. However, the implications of criteria (b) and (c) are less clear.

‘provided at the request of a recipient of the service’

On its face, a content marketplace such as YouTube or iStock does not appear to meet criterion (b). This criterion seems to imply a particular workflow: first, a ‘recipient of the service’—i.e., a prospective customer—makes a ‘request’; next, a person offering platform work responds to the request; the work is carried out; finally, the person carrying out the work is paid.

This workflow does roughly describe how work is done on other types of digital labour platform. On content marketplaces, however, creators typically create content not in response to explicit requests but by extrapolating past content consumption trends. Where content is created in response to explicit requests, those requests are generally informal. This distinguishes them from the ‘tasks,’ ‘projects,’ or ‘briefs’ posted by customers on other types of digital labour platforms. The latter are typically constructed in platforms’ legal terms as ‘invitations to make an offer’ and may include specific payment amounts. That is, they are formal, and they establish clear terms: the person performing work will receive a specific payment in exchange for the work. The payment will be provided by the customer—that is, the ‘recipient’—although, of course, it may be processed by the platform.

It is therefore difficult to argue that content posted to content marketplaces generally meets criterion (b). Some content—and some other transactions—however, might. Within-platform ‘channel memberships,’ as well as messages attached to payments (e.g., YouTube’s ‘Super Chat’ and ‘Super Thanks’ features), do seem to be services provided ‘at the request’ of the recipient.

‘involves […] the organisation of work’

What about criterion (c)—the requirement that, to be considered a digital labour platform, a service must involve ‘the organisation of work performed by individuals’?

With some potential exceptions, content marketplaces do not generally seem to be involved in ‘organising’ work in the colloquial sense of telling creators what kind of content to create, when, where, and with what tools and methods; and paying them in direct exchange for the creation of that content. That is, they do not direct work.

However, the proposed Directive has a specific interpretation of ‘organising work.’ Recital 18 of the Commission proposal clarifies (emphases added):

Organising work performed by individuals should imply at a minimum a significant role in matching the demand for the service with the supply of labour by an individual who has a contractual relationship with the digital labour platform and who is available to perform a specific task, and can include other activities such as processing payments.

Four things can be noted about this explanation. First, it indicates that ‘organising work’ is primarily about ‘matching demand […] with the supply of labour.’ Second, it notes that organising work ‘should imply at a minimum a significant role’ in matching demand and supply. Because playing a significant role in matching demand and supply is ‘a minimum,’ this would appear to indicate that a platform can play a significant role in matching demand and supply—but nonetheless not be considered as ‘organising work.’ Other criteria that might be decisive in determining whether a platform is to be considered as organising work are not articulated. The possibility for platforms to undertake ‘other activities such as processing payments’ is noted, but it is not clear if such activities are to be understood as indicating that a platform is ‘organising work.’ Third, the matching of demand and supply by the platform should consider individuals ‘available to perform a specific task.’ That is, it should consider the supply of labour in time. Fourth and finally, it establishes that ‘service providers’ should have a contractual relationship with the platform for it to be considered as ‘organising work.’

Like criterion (b), Recital 18 seems to assume the model of location-based platform work: a customer makes a request; an available service provider responds to the request; the work is carried out; the service provider is paid. As already noted, content marketplaces follow a different model: content creators make content in response to anticipated future demand, not specific formal requests. Additionally, unlike in other types of labour platform, the work product is non-rival: many viewers can watch a video, or many licensees license a photo, without impeding others’ ability to do the same. Algorithmic systems primarily manage creators indirectly and even implicitly—through rules about what kinds of content will receive advertising and be promoted, and real-time data about which content is ‘performing well’—rather than explicitly directing them. And in the case of YouTube, the primary paying customer is not the viewer but rather the advertiser. This raises the question of what, exactly, the ‘service’ is: is it the provision of video content to viewers, or the provision of an audience to advertisers? And who, then, is the ‘recipient’?

Put shortly, trying to use Recital 18 to determine if content marketplaces ‘organise work’ is difficult. The overall impression is not that the legislator made a considered decision to exclude content marketplaces but rather that they were simply not considered.

Considering criteria (b) and (c) together seems to lead to a similar conclusion. It can be argued that some content marketplaces, or at least some transactions on them, fall within the scope of the Directive. But the hoped-for legal certainty has not been achieved.

Refining the concept of ‘digital labour platform’

This raises a question: should the definition of ‘digital labour platform’ in the proposed Platform Work Directive be expanded to included content marketplaces? There are at least three reasons to argue it should. First, by any common sense understanding of the term, content marketplaces are digital labour platforms. Second, the literature documents that creators on content marketplaces experience the same challenges as other platform workers: opaque, error-prone algorithmic management; lack of representation; and, in some cases, potential false self-employment. (For a review, see ‘Content marketplaces as digital labour platforms,’ pp. 4-7.) Third, while other EU laws do provide protections content creators could use to address some of challenges they experience arising from algorithmic management—in particular the so-called Platform-to-Business Regulation (2019/1150) and the Digital Services Act (especially Art. 17(1))—these protections are not as comprehensive as those established by the Platform Work Directive. The right to human review of algorithmic decisions (Art. 8 of the Commission proposal) and the provisions on communication channels (Art. 15) are unique to the Directive.

Of course, it may be too late to refine the definition in the Platform Work Directive. There is however another reason to develop a definition of ‘digital labour platform’ that includes content marketplaces. There have been discussions about a possible international labour standard on ‘platform work’ through the International Labour Organization. This instrument would need a definition of ‘digital labour platform.’ This definition should include content marketplaces.

In developing this definition, it may be useful to distinguish the conceptual task of developing a clear definition of ‘digital labour platform’ from the policy tasks of clarifying the indicators of an employment relationship and reducing false self-employment. The existence of ‘platform work’ should not be taken to imply in every case the existence of an employment relationship. Some persons performing platform work do appear genuinely self-employed. Others seem ‘not quite genuinely self-employed,’ but ‘closer’ to being genuinely self-employed than to being employees. It may therefore be helpful to separate the question of ‘What is a digital labour platform?’ from ‘How can we identify false self-employment, especially in digital labour platforms?’

Acknowledgments

Thanks in particular to Eva Kocher and Alberto Barrio Fernandez for recent discussions on the Platform Work Directive. This essay is based on research funded in part by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 947806).

Some of the text in this essay is adapted from M. Six Silberman, Sangh Rakshita, Halefom Abraha, and Jeremias Adams-Prassl, ‘Content marketplaces as digital labour platforms: towards accountable algorithmic management and decent work for content creators,’ prepared for the 8th Conference on Regulating for Decent Work, International Labour Office, Geneva, 10–12 July 2023.