Digital labor platforms (DLPs) digitally connect human workers (i.e., service providers) with service consumers. Examples of DLPs include Uber, Deliveroo, Amazon Mechanical Turk (MTurk), Upwork etc. Mediated by DLPs, humans interact with algorithms to accomplish work such as driving taxis, delivering food, and executing Human Intelligence Tasks on platforms such as MTurk. This type of work is known as algorithmic work. DLPs use algorithms to digitally manage and control work allocation, worker behaviors and worker–customer interaction. However, DLPs differ from traditional organizations in terms of workers’ independent contractor status and a technologically mediated task environment. The practice of managerial control in DLPs is thus different from that in traditional organizations. How are workers on DLPs controlled? Current IS research has mostly answered this question using the concept of algorithmic control (AC), which is defined as using algorithms as a means to align worker behaviors with controllers’ objectives. However, in focusing on AC, the current literature reveals a number of problematic scenarios. First, human workers often do not fully understand the outputs generated by the algorithms and the mechanisms behind the outputs, leading to them experiencing confusion, role conflict and role ambiguity. Second, algorithms cannot make fair judgements on ambiguous and complex issues, such as reasons for late food deliveries and validity of customer negative ratings. Third, AC is associated with loss of autonomy, privacy and identification by the workers, which is detrimental to worker well-being. These examples show how AC does not provide adequate explanation and feedback to workers, cannot make contextually appropriate decisions in ambiguous situations, and does not take interpersonal and empathetic considerations into account when making judgments. The objective of this dissertation is to investigate how the control of algorithmic work on DLPs can be accomplished through both AC and non-AC means, from the perspectives of both controllers (e.g., DLPs) and controlees (i.e., workers). To achieve this goal, we investigate (1) alternative control means in addition to AC, from the controllers’ perspective, and (2) workers' judgments and reactions to AC, from the controlees’ perspective. In the first paper, we ask the following research questions: how are workers on DLPs controlled? and how and why do control mechanisms influence workers’ judgments and reactions? To tackle these questions, we conduct a systematic literature review to synthesize the discussions of prior research and answer the research questions. We shed light on how workers are affected by AC, by using the Micro Level Legitimacy Process Model as the theoretical framework. Specifically, we illustrate granular AC mechanisms, worker judgments and reactions, and relevant AC characteristics that potentially moderate the relationships between AC mechanisms and worker judgements. Our analysis unveils mixed findings about legitimacy in literature, and we propose suggestions for future research accordingly. In the second paper, we conduct qualitative research to explain how DLPs control workers, from the perspective of controllers. The research question we ask is: how do humans and algorithms work together in the control of algorithmic work? We conducted interviews with 25 food delivery riders and managers who oversaw the riders. We found that human managers complement the capabilities of algorithms to enact algorithmic control. We theorize our findings as “the augmentation of algorithmic control by human control”. Specifically, human managers should augment AC by undertaking control mechanisms that are personalized, ambiguous, and emotionally impactful. That is because humans have more advanced capabilities in some areas, such as sensitivity to changing facts, intuitive thinking, common-sense based judgment, advanced communication, and interpretation of emotional