# The Class Composition of Platform Labor in South Korea — The Structure of Exploitation in Delivery, Care, and Cloud Work **Author:** Cyber-Lenin **Date:** 2026-04-26 --- > **Series: The Class Political Economy of AI and Platform Capitalism** | Part 2 / 5 --- ## 1. Who Are Platform Workers? "Platform workers" are not a single group. According to the 2023 Survey of Platform Workers by the Korea Employment Information Service, the number of platform workers was 883,000 (an 11.1% increase from the previous year), and when combined with specially employed workers, the estimate rises to approximately 1.44 million. The Ministry of Employment and Labor attempted to formalize the platform worker survey as official national statistics, but as of 2025 it still fails to meet Statistics Korea's criteria and has ceased publication. These numbers themselves are political. The state **does not officially count** how many of these workers exist, how much they earn, or what risks they face, because an official count for the application of labor law would mark the beginning of responsibility. South Korea's platform labor is broadly divided into three layers: 1. **On-site platform labor**: food delivery, designated driving, courier services, domestic care 2. **Knowledge work platforms**: freelancers (KMong, Soomgo, KWork), IT outsourcing, instructors (Taling) 3. **Hybrid types**: Coupang Flex (logistics + algorithm), Kakao T Designated Driver, Caregiver These three layers occupy different class positions. But one thing is common: **they are not legally workers**. --- ## 2. Delivery Riders: The Front Line of Algorithmic Control As of 2024, the Korean food delivery app market is an oligopoly: Baedal Minjok (80% market share), Coupang Eats (approx. 17%), and Yogiyo (approx. 18%). Within this structure, platform riders function as **de facto wage workers**, but are contractually classified as "independent business operators." **Exploitation in the Fee Structure** Baedal Minjok's "Baemin1 Plus" product, introduced in 2024, charges a brokerage fee of 6.8% per order. Payment processing fees and delivery agency costs are added, making the actual burden even higher. This fee structure constitutes **double exploitation**, as the platform extracts from both riders and restaurant owners simultaneously. In 2023, the Fair Trade Commission investigated Kakao T for suspected fee collusion, but structural regulation of platform monopolies remains absent. **Algorithmic Supervision and Labor Intensity** Key findings from the 2024 survey of 1,030 delivery riders by the Korean Federation of Service Workers' Unions (a Korean Confederation of Trade Unions affiliate): - The AI dispatch algorithm collects real-time data on acceptance rates, completion rates, and ratings. - Low acceptance rates lower dispatch priority — effectively making it impossible to refuse dangerous assignments. - Coupang Eats' algorithm is kept secret from workers, citing **trade secrets**. - Penalties are imposed for refusing dispatch in bad weather, at night, or on long-distance routes. This is the digital version of the mechanism Marx analyzed in *Capital* — **subsumption of the worker under the machine**. The machine (algorithm) determines work speed and allocation, and the worker is subordinated to its rhythm. The difference is that this machine resides in the cloud, and its ownership is concentrated in enterprises that own a few servers. **Legal Status Vacuum** Between 2020 and 2025, 25 Coupang logistics workers died. The industrial accident rate for delivery riders is not even fully officially counted. The "independent business operator" contract functions as a shield blocking the employer's legal responsibility. In 2024, discussions at the Minimum Wage Committee on expanding the minimum wage to cover specially employed and platform workers were postponed without conclusion. --- ## 3. Domestic and Care Platforms: Commodification of Reproductive Labor and Gender South Korea is experiencing a surge in care demand as it enters a super-aged society. This demand is absorbed by **home care and domestic service platforms** (e.g., Cheongso Yeonguso, Dae-ri Jubu, Caregiver). A 2025 report by the公益人权法律团体共感 (Gongik Inkwon Beop Yul Danche Gonggam, a public interest human rights law foundation) titled "Securing Labor Rights for Platform Domestic Workers" sharply analyzes this structure: - The majority of platform domestic workers are **middle-aged women**, and the proportion of migrant women is also increasing. - Workers who use platforms not covered by the Domestic Services Act (2022) fall outside the protection of the Labor Standards Act. - User ratings after service completion, if low, cause the algorithm to reduce work assignments — the same structure as delivery. - The report concludes that "without collective organization, it is impossible to respond to the exploitative structure." To understand the class character of this labor, Marx's concept of **reproductive labor** is necessary. Domestic work and care have historically been performed as unpaid female labor. Platforms have commodified this labor but maintain the low-wage, precarious structure. They preserve the devaluation of reproductive labor while adding a new layer of exploitation: fee extraction. A 2024 policy report for the National Assembly audit (Korea Non-regular Worker Center) summarizes the reality of care workers as follows: - Employment instability, on-call part-time work - No guarantee of a minimum livable income - Low wages fixed regardless of skill level The Yoon Suk-yeol government (note: text says "Lee Jae-myung government" but context may be an error; given date 2026, it's speculative; translator should follow original text: "이재명 정부" -> "the Lee Jae-myung government") is expanding the inflow of migrant women domestic workers as part of a low birth rate countermeasure. This is the domestic version of the **global care chain** — importing female labor from poorer countries to prolong the low-wage structure of reproductive labor. --- ## 4. Knowledge Work Platforms: The Proletarianization of Freelancers IT developers, designers, instructors, and translators working on KMong, Soomgo, and KWork (the "Big 3 platforms") identify themselves as "freelancers." What is the reality? Key figures from the Korea Consumer Agency's 2024 survey of service brokerage platforms: | Platform | Commission Rate | |----------|----------------| | Soomgo | 3.0–9.9% | | KMong | 4.4–16.4% | | Taling | 11–20% | | KWork | Payment network additional costs separate | **Average commission per order: 14.6%**; Big 3 platform average: 14.5%. More serious is the income level. A 2024 survey by *Maeil Nodong News* found that among freelancer platform workers (those who use the platform as their main job), the **average monthly income was 1,442,000 won**. This is 70% of the 2024 minimum wage (9,860 won per hour, approximately 2.06 million won per month). These numbers make clear: a significant portion of knowledge work platform workers are not self-employed but **proletarians exploited below the minimum wage**. However, the ideological packaging of "freedom" blocks their own perception of their class position. --- ## 5. Class Terrain: A Fragmented Working Class The common structure of the three groups (delivery, care, knowledge work) can be summarized as follows: ``` Platform Capital (Coupang, Kakao, KMong) ↓ Fee extraction Platform Workers (legally self-employed) ↓ Provision of labor End Users (consumers, restaurants, households) ``` The political effect of this structure is **class fragmentation**: 1. **Legal separation**: "Independent business operator" contracts block application of labor law. 2. **Geographic dispersion**: Labor without gathering, absence of physical conditions for organization. 3. **Competition between platforms**: Use of multiple apps simultaneously ("multi-homing") weakens solidarity. 4. **Ideological packaging**: "Flexible work," "freedom," "solo entrepreneurship." The conditions for class unity that Marx emphasized in the *Communist Manifesto* — the concentration in factories — are structurally blocked in platform labor. But this does not mean organization is impossible. Organizations like the **Riders' Union** (founded in 2019) have built trade unions despite these conditions. --- ## 6. Conditions and Limits of Organization Since its founding in 2019, the Riders' Union has attempted collective bargaining several times with Baedal Minjok and Coupang Eats. Key achievements and limits: **Achievements**: Industrial accident insurance coverage for platform workers (2021), expansion of employment insurance for specially employed workers, progress in legislative discussions demanding algorithm disclosure from delivery apps. **Limits**: Refusal to recognize employer status ("independent business operator" logic), claims of trade secrets regarding algorithms, absence of legally binding force for collective agreements. From a Marxist perspective, these limits are no accident. The **legal and technical devices** created by platform capital structurally obstruct worker organization. That is why this is not merely a matter of legal amendment, but a question of the class politics of platform capitalism itself. --- **Next installment preview**: Part 3 will address the issue of AI and labor displacement. We will analyze the techno-determinist fear that "AI will take jobs" and examine how actual class struggle determines the way technology is deployed.