Do or Delegate? Learning to Create and Manage Complexity in Collaborative Knowledge Creation

Delegation involves a delegator’s discretionary decision to defer authority and responsibility to a delegate (Yukl, 2013). Classical organization theory posits that delegation follows from a “fundamental organizational... [ view full abstract ]

Delegation involves a delegator’s discretionary decision to defer authority and responsibility to a delegate (Yukl, 2013). Classical organization theory posits that delegation follows from a “fundamental organizational problem” of the division of labor, and correspondingly, the process through which organizational actors divide overarching organizational goals into subgoals (e.g., Puranam et al., 2014). Delegating authority and responsibility over subgoals to subordinates enables delegators (e.g., managers) to economize on scarce time and attention, utilize dispersed expertise more efficiently (e.g., Dobrajska et al., 2015), and motivate employees (Gambardella, Khashabi, & Panico, 2014).
While offering insights into the rationale, process, and outcomes of delegation (e.g., Dobrajska et al., 2015; Leana, 1986; Yukl & Fu, 1999), prior research has been based on two questionable assumptions. The first assumption is that delegators can process information on potential delegates prior to choosing whether to delegate and to whom to delegate. For example, studies have shown that delegators rely on a formal reporting structure and role descriptions, congruence between the delegator and delegate’s goals, and knowledge of the delegate’s job competence, trustworthiness, and interest in assuming increased responsibility (e.g., Leana, 1986, 1987). The second assumption is that delegators operate within a hierarchy where they can use mechanisms such as formal control, incentives, knowledge transfer, and integration tools to minimize the uncertainty underlying the decision to delegate (e.g., Dobrajska et al., 2015; Gambardella, Panico, & Valentini, 2015).
These assumptions are contested by recent work on new forms of organizing, which challenge the traditional mechanisms enabled by hierarchy and information on the identity and capabilities of the delegates. We aim to increase understanding of delegation in new forms of organizing to shed light on the microfoundations (Baer, Dirks, & Nickerson, 2013) of organizational delegation—i.e., organizational actors’ decisions and behaviors relating to maintaining or delegating responsibility for executing tasks. We do so by focusing on an organizational setting where (1) information on potential delegates’ identity and capabilities is severely limited, and (2) delegators lack the formal authority, based on hierarchical structures, to impose tasks on delegates and control their execution.
Given the pervasiveness of new forms of organizing where delegators lack information and authority, we propose that current knowledge on delegation must be complemented with a new theoretical frame wherein the assumptions of hierarchy and information on delegates are relaxed. We conceptualize this as “peer-to-peer” (P2P) delegation and argue that in the absence of traditional design attributes, organizational members base their decision to either self-implement or resort to P2P delegation on the basis of (1) their prior experience with coordinating idea implementation (i.e., successful or failed self-implementation versus delegation) and (2) the complexity of the task at hand. Drawing on organizational learning theory concerned with the distinction between learning from success versus failure and experiential versus vicarious learning modes (e.g., Deichmann & Ende, 2014; KC, Staats, & Gino, 2013; Madsen & Desai, 2010), we disaggregate delegation patterns and ask the following research question: How does a decision maker’s experience from coordinating prior implementations and task complexity impact delegation decisions and their subsequent performance outcomes?
We will explore this question in the context of OpenStack—an online OSS community with 34,024 independent and firm-sponsored contributors from 557 member firms spanning 177 countries. OpenStack has a technology-enabled organizational design in which community members take part in collaborative knowledge creation, problem solving, and decision-making. To balance learning from peers and strengthening reputation by submitting high-quality contributions, community members need to decide when to self-implement their novel software development ideas and when to delegate responsibility for implementing these ideas. We will employ the OSS setting to study how this decision is impacted by the mode (i.e., self-implemented vs. delegated) and outcome (i.e., failed vs. successfully implemented) of prior experiences.
OSS development communities are an opportune research setting for studying delegation and individual learning for at least two reasons. First, delegation is particularly challenging in OSS communities because of the likelihood of delegating tasks to individuals without prerequisite expertise. In online communities, delegators generate ideas for software development yet delegate the implementation. Delegators have little discretion over who implements the task once they defer their responsibility for implementation. Rather, delegates self-select into the implementation of these “open tasks.” Second, because learning-by-doing is particularly potent in OSS participation (e.g., von Krogh et al,, 2003), OSS communities are suitable environments for developers to enhance their learning (Hars & Ou, 2002; Hertel, Niedner, & Herrmann, 2003; Lakhani & Wolf, 2005).
Initial data collection of about 9,000 software development ideas has been completed. At the time of writing (10.05.16), all data on the delegation structure and implementation status of each software development idea, and all data on the prior experience of each developer, have been collected, and about 60% of the data on developers’ prior network ties, and 70% of data on task complexity have been mapped.
We use logistic regression to estimate the likelihood that a given task was delegated and that it was successfully implemented (Deichmann & Ende, 2014; KC et al., 2013; Madsen & Desai, 2010). We will apply a mediation analysis using a structural equation modeling framework (Muthén, 2011) to explore the impact of task characteristics and developers’ prior experience on the likelihood of successful idea implementation through delegation. We expect to complete a comprehensive analysis on the collected data set and recognize emerging theories by conference presentation (1 August 2016).