What „Species” is Platform Work?

2. Conceptual Basis

The use of the internet has expanded significantly in recent decades, and can be divided into three phases in terms of time and function: in the early days (first phase), the internet was used to access and search for information; in the second phase, it was also used to sell products; and in the current third phase, labour, works, professions and time can also be sold [17], giving rise to new forms of work.[18] The sale of labour takes place through digital work platforms, which have become economically significant today. We must also recognise that economists consider digital platforms to be one of the most innovative parts of the economy, due to their diversity and continuous development. [19]

Ales emphasised the dematerialisation of companies in relation to the digitalisation of business life. [20]

Based on the European Union's previously referenced 2024 directive, a digital work platform is defined as a service provided by electronic means that includes, as an essential component, the organisation of work processes through a digital work platform, which is carried out by individuals in the context of transactions for consideration.

The ILO emphasises that these platforms are based on information technology and use the internet to quickly match work. De Stefano and Aloisi highlight three features: on-demand labour mobilisation, task-based remuneration and technology-driven matching.

A structural feature of platform work is the triadic legal relationship, i.e. the three-way construction of the platform, the worker and the client, which differs from the traditional two-party employment relationship. In some cases, this can blur lines of responsibility and make it difficult to apply traditional labour law categories [21].

So-called "automated decision-making systems", i.e. algorithmic management, form the technological basis of platform work. These systems make all important decisions, from recruitment to account suspension, while increasing information asymmetry and reducing human agency.

The central dimension of the typological division of platform work is spatial location. Location-based platform work, or "on-location platform work", requires physical presence. This includes passenger transport, delivery and domestic services, where workers compete in local markets. Typical examples are passenger transport (Uber, Bolt), food delivery (Deliveroo, Wolt) and domestic services (TaskRabbit).

Crowdwork refers to work carried out within the framework of crowdsourcing, and the term is used in this context. [22]

But what is 'crowdsourcing'? The term was first used in 2006 by American journalist Jeff Howe of Wired magazine: "Simply defined, crowdsourcing is the practice of outsourcing a function previously performed by employees to a predefined (typically large) group of people in the form of an open call. The task may be solved through collaborative work, but it is also common for it to be done individually. The essential conditions are an open call and a large network of potential participants." [23]

Outsourcing is the transfer of a well-defined task or job to a company that specialises in that particular area. Some authors use the term "online outsourcing" instead of "crowdsourcing", which more accurately reflects the nature of the institution. According to a World Bank study, in online outsourcing, a third-party contractor offers their services – their work – on an internet-based labour market. These digital intermediary channels enable companies to outsource activities that were previously performed by their own employees to a remote workforce.[24] Just as the definition of 'outsourcing' is extremely diverse, with Wikipedia listing more than 20 different concepts, 'crowdwork' also takes many forms.

Online outsourcing can be internal or external. In the former case, the company advertises the job opportunity to its own employees on its own electronic portal. In the latter case, the company places the tasks on an electronic platform accessible to anyone – a third party.

There are basically two types of external outsourcing: bilateral and trilateral. In the case of bilateral outsourcing, the company itself creates the website on which communication and the exchange of services take place. Other companies and customers cannot join this network. In trilateral outsourcing, a third party, a company, operates a web interface through which companies can search for employees. [25]

Crowdworking is therefore an activity that manifests itself in the field of crowdsourcing, the essence of which is that employers and employees meet on an online platform, where they often share and divide larger tasks among a "virtual cloud" of employees [26]. A characteristic feature is that technology companies are able to attract large numbers of job seekers to their online platforms, and as this creates an oversupply in the digital labour market, the impact on wages is often negative [27].

The nature of the work can be extremely diverse. It is very common to have to perform micro tasks such as tagging photos, evaluating emotions, and filling out questionnaires. In other cases, more serious and significant work is organised, such as designing a logo, developing a website, or participating in the initial phase of a marketing campaign [28].

Online platform work, or "crowdwork," is not dependent on geographical location, meaning that tasks can be performed from anywhere. There are two main subcategories of crowdwork: microtasks, which are short, repetitive, low-skilled activities (Amazon, Mechanical, Turk), and online freelancing, which requires specialised expertise, such as software development, graphic design or translation (Upwork, Freelancer).

According to the ILO, job security, social protection and collective bargaining rights are critical issues in stationary platform work. Platform workers often work irregular hours for low pay, while bearing the costs and risks themselves.

The situation is more mixed in the field of online platform work. Those performing microtasks face extreme vulnerability, often earning less than the minimum wage, according to Berg et al. In contrast, highly skilled online freelancers enjoy greater autonomy and income [29].

The control intensity grouping grouping is also considered. We can distinguish between platforms with low control (intermediary function), those with medium control (pricing, quality standards), and those with high control, which comprehensively regulate work, scheduling, and performance evaluation. The latter category is most relevant to the question of classification as an employment relationship. Furthermore, it can be concluded that different types of platform work generate different risk profiles, and the most heated debates typically revolve around high-control, location-bound platform work.

Recognising this complexity, the EU legislator applies multi-layered regulation. The presumption of an employment relationship is only activated in cases where there are facts indicating management and control, while the rules on algorithmic management apply to all platform workers, regardless of their employment status. This multi-level approach reflects the fact that, for certain types of work, classification as an employment relationship is justified, while at the same time all platform workers are entitled to basic guarantees in the areas of data protection, transparency and supervision.

Based on the above distinction, it is clear that platform work is not a homogeneous phenomenon, as the triadic legal relationship, spatial location and intensity of control result in radically different risk profiles. While classification as an employment relationship may typically be justified in the case of on-location platforms exercising a high degree of control, this is not at all the case in certain segments of crowdwork that offer a high degree of autonomy.

The multi-level EU regulatory solution can therefore be seen as a promising starting point, as it theoretically allows the most vulnerable platform workers to be protected by labour law guarantees, while at the same time seeking to enforce basic guarantees against algorithmic power for all platform workers. However, it is problematic that while the enforcement of fundamental guarantees is unquestionable in all cases, the practical enforceability of labour law guarantees is far from straightforward. In our view, this logic justifies further refinement of the regulations, primarily in the case of high-risk, on-location platform work involving strong control, since it is desirable for legal guarantees to be as simple and, if possible, automatic as possible. In order for this expectation to be realised as widely as possible , three potential solutions can be envisaged: the first would result in the creation of a third category, the second in the establishment of a fiction, and the third in the imposition of mandatory contractual content under clausulae cogentes rules. [30]

3. The Limitations of the Classic Binary Approach

The classic system for determining employment status is based on a dichotomy between two categories: employment and self-employment. This categorisation is of particular importance because employee status provides access to a wide range of labour and social security protections, while the self-employed are denied these protections [31,32]. According to the ILO, the classification of the legal relationship should be determined primarily by the facts relating to the actual work performed and the remuneration received.

Traditional classification tests use a multi-factor approach. The prevailing view is that three basic elements are necessary for an employment relationship to exist: work performed for the benefit of another person, direction and supervision by the person for whom the work is performed, and the existence of remuneration. The typological method also assesses a number of additional factors, such as organisational integration, economic dependence, personal obligation to perform the work, ownership of work equipment, nature of remuneration, risk exposure and length of the legal relationship. These criteria are not rigid rules, but guiding factors that need to be assessed together, according to De Stefano and Aloisi.

The structural characteristics of platform work call into question the validity and applicability of traditional classification tests in a number of respects. As discussed in the previous chapter, the triadic legal relationship, in which the platform appears as a triangle between the worker and the customer, differs from the classic two-party employment relationship. This can result in a more unhindered market, where the platform plays an intermediary role in matching supply and demand, but it can also happen that the dominant role of the platform can disrupt and upset the market. This multipolar setup in itself obscures the relationships between responsibility and interest, not to mention the fact that the low entry threshold allows platform workers to move freely between different rival platforms, which prevents employer rights and obligations from being clearly linked to a single entity.

The issue of control and instruction is perhaps the most problematic area. Traditional tests examine the direct control of a human superior and the employee's organisational embeddedness. In the case of platform work, however, control is typically indirect, using algorithmic tools. Automated decision-making systems make all important decisions, from recruitment to account suspension. Platforms employ strict monitoring based on customer ratings, GPS coordinates, barcode technology, deadlines, continuous measurements, screen recordings, response times and unknown algorithms. Algorithm-mediated control thus represents both strong control and apparent distance, which calls into question the relevance of traditional subordination tests.

At the same time: the flexible opportunities provided by platforms and the freedom of platform workers, such as the freedom to choose when to log in, point towards entrepreneurial independence. Nevertheless, it is also important to note that this flexibility often comes at a price, as the prevailing view is that de facto economic dependence and the indirect coercion exercised by platform algorithms create a situation in which workers remain dependent despite this formal flexibility, for example due to the possibility of sanctions for inactivity. According to De Stefano and Aloisi, in many cases, true autonomy and control remain unattainable for most platform workers.

This raises the question of whether the situation would be different for entrepreneurs in the classical sense. As we have already stated, platforms play an intermediary role, with the aim of connecting supply and demand. Algorithmic operation favours fast and reliable individuals in order to satisfy demand as best as possible, while more inactive, less efficient individuals are pushed into the background. This essentially simulates the functioning of the market, supply and demand, i.e. it artificially recreates the exact same conditions under which businesses compete. It is natural for an inefficient, inactive entrepreneur to be at a disadvantage compared to their faster, more skilful, more stable and more reliable competitors. This raises the following questions:.

Can the simulation of market mechanisms be considered management, or is the platform merely a tool in the hands of entrepreneurs competing in the arena of capitalism? Can entrepreneurs in the classical sense sit back and relax? Do they have greater autonomy and control? Who does the platform worker actually depend on, the platform or what the market dictates?

The specific nature of sanctions raises further difficulties with regard to supervision and control. In traditional employment relationships, the employer's disciplinary measures are surrounded by transparent procedural guarantees. In contrast, in the context of platform work, the form of sanction is deactivation or withdrawal of access, which functions as a quasi-employer disciplinary measure, yet does not meet the formal requirements of classic employer disciplinary rights. The preamble to the directive emphasises that control and supervision may take different forms in certain cases and that platforms may also exercise control through various sanctions or other forms of pressure.

The picture is further complicated by the fact that the issue of personal performance and substitution is also ambivalent. Certain platforms allow substitution, which argues in favour of entrepreneurial status. In other cases, however, personal performance is mandatory, which again points to the nature of the employment relationship. A similar duality can be observed in the area of economic risk, as platform workers bear the costs of purchasing equipment themselves, while the determination of remuneration and control over access to job opportunities is entirely in the hands of the platform. The latter functions as a quasi-price cartel and creates uniform rules and opportunities for workers. Further ambivalence can be observed in the area of rapid interchangeability between different platforms, which is largely contrary to the principles of labour law, such as the obligation to cooperate and provide information, and the protection of the legitimate economic interests of employers [33]. According to an ILO survey, the majority of platform workers are economically dependent on the platform. For example, in the field of crowdwork, platform work is the main source of income for 32% of workers, but if we also take into account those who do not have paid work, this proportion rises to 48%. All this indicates that, compared to the classic employer-employee relationship, the relationship between the platform and the worker is looser, their interests are less closely aligned, their fates are less intertwined, and the loyalty, mentality and basic principles, typically such as the principles of cooperation, information, good faith and fairness, and fair consideration, as well as the protection of the employer's legitimate economic interests.

Some of these principles have a different meaning in this relationship (duty to cooperate), exist only on one side (principles of good faith and fairness), or are completely absent (principle of fair consideration). .

De Stefano and Aloisi conclude that there is often a discrepancy between the contractual status of platform workers and the working conditions dictated by the platform, as self-employed status and the working conditions dictated by platforms are incompatible. Therefore, in some cases, the contractual classification constitutes obvious legal pretence, according to De Stefano and Aloisi. .

Gyulavári sees a clear trend towards the spread of self-employment at the expense of employment in an employment relationship, which has also brought about a change in the principle of subordination and the spread of outsourcing and subcontracting chains [34,35].

According to György Kiss, the dual model of employment and self-employment is outdated because the labour law structure that has developed in Europe has been unable to keep pace with market changes, to which contracts are generally able to respond [36,37].

Our analysis shows that most traditional classification criteria present an ambivalent picture in the case of platform work. Subordination exists, but to a much lesser extent, completely lacking in personal character, only in algorithmic form, simulating the functioning of the market; economic dependence, although typically to a lesser extent, still exists, but is coupled with a greater degree of autonomy than in an employment relationship; the work is predominantly personal in nature, but in many cases the platform is unable and unwilling to control this; there is a lack of traditional organisational embeddedness; remuneration is regular but does not take the form of a classic wage; the platform worker provides the work tools; there is a lack of loyalty and the relationship between the parties is completely different from that of an employment relationship. Platform work therefore does not clearly meet the criteria for either employment or self-employment, which means that the application of the classic binary classification necessarily leads to misleading results.

This classification uncertainty is not merely a theoretical problem. The preamble to the Directive also points out that, due to information asymmetry, it is often difficult for platform workers to obtain adequate access to the tools and information necessary to prove the true nature of their contractual relationship and their rights under it to the competent authorities. Thus, information asymmetry, the unilateral practice of contract modification by platforms, and the opacity of algorithmic management make it difficult for platform workers to correctly assess their legal status and enforce appropriate legal protection.

Any classification disputes often result in court proceedings that drag on for years, during which time the workers remain in a precarious legal situation. This is further complicated by the fact that a significant proportion of platform workers are foreign nationals, so a lack of knowledge of the language and applicable law is an additional barrier to the effective exercise of their rights. Not to mention those who work on platforms as a second or third job, for whom the positive outcome of a lawsuit is a negligible prospect in view of the effort and capital investment required to assert their interests and the possibility of losing the case. Consequently, it can be concluded that, given the practical difficulties of asserting their interests, the current regulations do not provide adequate protection for platform workers.

In summary, the triadic legal relationship and algorithmic management described in Chapter 2 already indicate that platform work is a phenomenon that goes beyond traditional legal categories, and that the application of multi-factor tests can lead to contradictory results in some cases, as certain criteria point to an employment relationship, while others point towards self-employment. It is particularly problematic that traditional tests do not take into account the dynamic nature of the platform economy, frequent contract modifications and the constraints generated by reputation systems. In addition, platform workers often find that, although they formally operate as autonomous entrepreneurs, in practice their working conditions and income are largely determined by unilateral decisions made by the platforms. For all these reasons, the application of the classic binary classification not only causes a high degree of uncertainty, but also inevitably leads to misleading results. Assuming, but not allowing, that our platform can be clearly classified as one of the classic employment relationships, the complexity and intricacy of the method poses a further problem. The practical difficulties and uncertainties described above pose significant obstacles to the effective enforcement of interests, thereby hindering the successful enforcement of the labour and social security guarantees to which platform workers are otherwise entitled under the regulations. For this reason, instead of rigidly applying binary categorisation, it may be justified to use an alternative approach that offers a more flexible and, above all, automatically applicable system of protection tailored to the specific characteristics of platform work.

The Supreme Court's Ruling on Platform Work – Hungarian Case Law

In December 2023, Hungary's highest court, the Supreme Court, ruled for the first time on the labour law status of platform workers. In relation to a food delivery driver employed by Foodora, it found that the person "does not perform his food delivery activities within the framework of an employment relationship, and the evidence did not allow for a contrary finding..." [38] The panel of judges examined the legal dispute on the basis of the substantive characteristics of the employment relationship, based on the indicators previously issued by two ministries (the Ministry of Employment and the Ministry of Finance) and already repealed by Joint Directive 7001/2005. (MK 170.) FMM-PM.

With regard to the job description, the court emphasised that it had to take a position on whether the parties had defined the work to be performed as a job description. "The law does not define the concept of a job description, but according to the prevailing legal interpretation, it should be understood as the totality and nature of the work that the employee is required to perform. In addition to the general wording, according to legal literature, a job description refers to the specific definition of the tasks that employees are required to perform, expressing what tasks the employer assigns to a worker under the employment contract in accordance with the organisational division of labour."

In the case in question, under the framework contract of mandate, the courier's sole task was to deliver food and beverages from a specific address to a specific address. In the Court's view, neither the content of the General Terms and Conditions (GTC) nor the individual orders allowed the conclusion that the platform defendant had actually defined the tasks as a job. A particularly strong feature of litigation is that the party who has an interest in having their version of events accepted as true by the court must prove it. In the present case, due to procedural obligations, the court ruled that the platform employee was unable to prove that the platform had any influence on the performance of the task, or that it had determined the manner of performance or the essential circumstances thereof. In our opinion, the manner and circumstances of performance are defined in practice by the job description. We note that the job description is not a labour law institution included in the Labour Code. However, the intended absence of this does not allow us to conclude that the platform has no influence on the circumstances of performance, as it is precisely its extensive powers of control and reward that show that it wishes to achieve (better) performance from the employee in this way. International literature is also increasingly unanimous on the issue that one of the most worrying dimensions of artificial intelligence (AI) in today's workplaces is that these systems can exert almost unlimited influence and serve to control workers as representatives of managerial power [39].

According to the facts established by the court, the acceptance of the GTC established a two-part remuneration system between the parties (fixed hourly wage plus a fee for deliveries). From this, the court concluded that the parties did not agree on a base wage, even in the form of performance-based pay, and that the amount of remuneration was not fixed or capped. This form or methodology of remuneration is not uncommon in other atypical contracts, and can also be established in combination; in this regard, the free agreement between the parties is not contrary to the rules of the Labour Code.

The court examined the relevance of the employer's main obligation, the fulfilment of the employment obligation in the case. In this regard, it noted that the platform employer did not assume any employment obligation in the general terms and conditions, even during the employee's active periods: the platform – similarly to the employee – was entitled to unilaterally terminate even an active period that had already begun, for example, if it was unable to assign a specific task. Under Hungarian law, the employer is obliged to employ the employee in accordance with the employment contract and the rules governing the employment relationship and, unless otherwise agreed by the parties, to provide the conditions necessary for the performance of the work [40].

It is characteristic of platform work that the person undertaking the work can report for work at the times specified by them and for a period of time that suits their needs. The court found that no unequal work schedule had been specified, as the parties had not agreed on the number of working hours in the contract. The worker could decide to work 2, 4 or 8 hours a day, varying from time to time. In the court's view, this does not constitute a flexible work schedule either. The court also noted that the person undertaking the work "was entitled to terminate the agreed active period without having to report this or obtain permission." However, this has financial consequences, and it is not in the interest of the person undertaking the work not to work the entire agreed period, as the algorithm immediately evaluates this and "penalises" it. This meant that they could only apply for their desired time slot later, after the others. Thus, the worker may have little chance of getting the more desirable time slots, which may result in less income.

According to the court, determining the number and length of active periods based on an algorithm does not constitute a broad right to issue instructions. In this respect, the broad right to issue instructions may be extended by the above, as this cannot be done on the basis of a consensus between the parties, but only by the employer's unilateral decision and exercise of its rights. According to the court, this circumstance cannot be assessed to the detriment of the platform employer, as it is in its business interest and affects its efficiency to give preference to those who are reliable, diligent and regularly apply for longer active periods.

In general, it can be said that all employers in the competitive sector are profit-oriented, so increasing the organisation's performance is their clear goal. The cornerstones of this are the employees, so if an employee performs exceptionally well, it also increases profits, and this characteristic is not unique to platforms. It can also be noted that organisations have countless tools for motivating employees (bonuses, employee of the month, rewards for those not on sick leave), but this does not mean that employees who are encouraged by their employer to work more efficiently are self-employed.

Managerial powers form a dense network of power, which can be formally distinguished in the form of management, control and discipline, but in practice is so closely intertwined that it functions as a unified exercise of power. Together, these powers form mutually reinforcing functions that define the concept of employer Aloisi - Valerio De Stefano.

The relationship of subordination between the parties is of particular importance in employment relationships. According to the Supreme Court, subordination cannot be established in this case either, as the person undertaking the work did not perform their work within the platform's organisation and their activities were not linked to its corporate structure. The court concluded from this that the employer did not have unilateral management and instruction powers. However, according to the court of second instance, the platform unilaterally determined the remuneration, the transport tasks, the place and time of their performance, and the active periods, which, in our opinion, implies that the platform had extensive rights to give instructions and exercise control. Although the platform did not provide the employee with a means of transport for work, this does not directly imply self-employment. In Hungary, it is not common for employers to determine what type of vehicle employees should use, and employers do not provide cars to health visitors, for example, even though they also visit clients for most of their working hours. Health visitors are not self-employed, which is also a fact.

The platform used a GPS system, which, according to the court, does not in itself constitute employer supervision. However, the GPS system stores data which, when analysed, allows for strict supervision of workers on the one hand, and on the other hand, checks whether deliveries are made properly and which courier is closest to a given address and can be offered the job.

The judgment found that the worker performed his work using his own equipment (vehicle, mobile phone), which, in his view, supports the existence of a business relationship. Based on the above, this is debatable, especially since the platform provides work clothes and a heated locker, which the worker is required to use. [41] Hidden behind marketing objectives, this employer's ticket may indeed appear to be a feature of the employment relationship, but, for example, the work clothes of an employee working in a BOSCH factory are also labelled, which serves the same employer branding and marketing purposes, yet the employee cannot be classified as self-employed.

According to the Supreme Court, the place of work is not decisive either, because the employee's task was to deliver food from one address to another, and he could decide for himself where to stay between deliveries. This is also contradicted by the type of work where the employee does not usually perform his or her activities at the place of business, such as a washing machine repairman or a DPD courier

4. Conclusions

The mule analogy referred to in the previous chapter is not only a narrative device, but also an apt description of the legal nature of platform work. A mule is a hybrid animal, neither a horse nor a donkey, but a unique, distinctive and independent creature that possesses the characteristics of both. Similarly, platform work cannot be considered purely as an employment relationship or self-employment, but rather as a specific legal relationship that is characterised by elements of both categories, yet does not fit completely into either.

The binary classification system represents a dogmatic constraint that forces the law enforcer to make a choice. The hybrid nature of platform work shows precisely that the classic categorisation is unable to capture the true nature of the legal relationship, as this now outdated dichotomy ignores the newly emerging needs arising from changes in society, which legislators would be well advised to follow in light of their regulatory objectives.

De Stefano and Aloisi point out that many Member States already have intermediate categories, such as the German 'employee-like person', the Italian 'para-subordinate' or the Spanish 'economically dependent autonomous worker'. These categories are based on the recognition that some platform workers display only certain characteristics of employees, while others do not, and therefore only certain labour law rules should apply to them. The purpose of the intermediate category is therefore to provide partial protection to those who do not qualify as employees, according to the opinion expressed by De Stefano and Aloisi in their aforementioned work.

Binary classification prevents the development of function-specific protection solutions. In contrast, the hybrid analogy suggests that recognising the specific nature of platform work and classifying it as a separate, third category would allow for the development of a minimum guarantee system ( ) that is tailored to the actual working conditions and the degree of control exercised by the platform, as well as its specific characteristics, rather than depending on the rigid application of predefined legal categories.

However, the main shortcoming of the current regulation is not theoretical inaccuracies, but the fact that its practical implementation is extremely complex and protracted. The categorisation process is uncertain and lengthy, not to mention the fact that, in most cases, platform workers are kept in a vulnerable and uncertain position due to information asymmetry and economic inequalities. In theory, however, a decisive, automatic choice could provide a solution. In the second alternative, we could consider the mule to be a horse in order to get rid of the lengthy, uncertain procedure of classification, which lacks legal certainty and effective protection, and to ensure more effective practical protection.

Finally, as a third potential alternative, we could consider the mule to be a stronger donkey, which, due to its nature and characteristics, has corresponding additional rights. The solution to this would be to prescribe mandatory contractual content.

In summary, in light of the above findings, the classic binary classification is outdated, as in its current state it is unable to provide effective protection for platform workers in practice. As a result of the investigation, it is necessary to rethink the regulatory environment and the dogmatic constraints behind it. In this regard, three alternatives are mentioned in the study, which would be worth examining critically, but the scope of this study does not allow for this as a further direction of research. However, a clear picture can only be obtained by thoroughly examining all the alternatives and evaluating them together with the current regulations in the light of legal practice.