How to organize digital dirty work

A human relation approach

Jeonghun Kim, Korea University Business School
Heeeun Jang, Yonsei Business Research Institute

One of the newly emerging digital jobs, content moderation, is investigated in this research. Content moderation refers to the process by which social media platforms monitor and filter user-generated improper contents using human labor and artificial intelligence. Improper contents contain the pornographic, the obscene, the violent, the illegal, and the hateful. Because these contents could damage the advertising revenue of social media platforms, a growing number of employees are being hired to moderate them. As managing data is the fundamental building block of the digital economy, it is self-evident that content moderation becomes a salient and elementary job in the era of digitalization. Accordingly, in the present study, the ways to organize content moderators are reviewed, and especially a human relation approach as an organizing way is investigated through the case study.

According to prior literature, a tayloristic approach is known as a widespread way to organize content moderation tasks. Digitalized tasks are fragmented into the smallest components and globally distributed through crowdsourcing platforms such as Mechanical Turk and oDesk. Globally disperse and isolated moderators perform these microtasks and are rewarded by a piece-rate system. Under the condition that the number of tasks streamed into crowdsourcing platforms is volatile, fragmented workers constantly experience a sense of precariousness. Repetitive and tedious tasks engender a feeling of stress and exposure to inappropriate contents sometimes triggers psychological trauma. For these dark-sides, content moderation often is called “digital dirty work”. Based on this approach, the growth of digital economy results in low-grade jobs.

However, we investigate a company in the current research that utilized an alternative approach to organize content moderators. This company, which was established in 2013, represents itself as an elderly-friendly company. Four hundred fifty employees were employed in this company in 2018, of which average age is 60 years old, to mainly process contend moderation tasks such as map blurring, monitoring illegal and unethical postings, and managing flagged contents. Content moderation tasks have been subcontracted from various IT companies, including one of the biggest IT company in South Korea. The most intriguing point is that this company internalized elderly workforce rather than distributing content moderation tasks to crowdsourcing platforms. The emphasis was placed on core values such as equality, sharing, development to induce horizontal organizational culture. Most employees were employed as part-time employment which is suitable for the nature of elderly workforce, but guaranteed long-term contract. In addition, they were grouped by teams and worked together in the off-line workplaces. Various HR practices were adopted to increase employee motivation and skills, including socialization programs, intensive trainings, and fringe benefits.  

The internalization of workforce resulted in high performance in terms of productivity and quality. The interviews with several employees uncovered two mechanisms to explain this result: social identification and collective learning. First, employees’ social identification with occupation was an essential factor to lead to high performance. Since the elderly workforce is not rarely given the chance to participate in South Korea labor market, positive symbolic resources provided by company are effectively used by employees to reconstruct their positive -self. Many employees had a sense of mission about their job and called themselves as “digital county” rather than felt ashamed to do “digital dirty work”. The occupational identification induced a high level of employee involvement in the work process, and thus high-quality decision making. Second, collective learning, which is facilitated by HR practices, were effective to cope with improper contents that are always changing. Along with intensive training to develop employees’ human capital, social capital among employees fostered by team activities, socialization programs, and egalitarian culture led to collective learning and increased problem-solving skills. In sum, the internalization, based on the human relation approach, was effective to handle content moderation tasks that require ethical and contextual decision making and ongoing learning.

The results of this research showed that the human relation approach could be an effective way to organize digitalized microtasks. The tayloristic approach base on the premise that digitalization inevitably deteriorates the quality of existing jobs and generates low-grade jobs. However, fragmenting crowding workers has limitations in that the potential productivity embedded within social relations is overlooked. For this reason, HR practices designed to foster social relations within workplaces are still important to effectively process digitalized tasks. Especially, motivation coming from social meaning and collective learning through cooperation between co-workers could be utilized as core resources to increase organizations’ competitive advantage, which is not easy to be organized among crowding workers. In conclusion, this research sheds light on the importance of managing human resources in the era of digitalization.