This week I read Turnover-induced knowledge loss in practice by Martin P. Robillard, a professor of computer science at McGill University. This study explores when knowledge loss happens as well as the implications of knowledge loss.
My summary of the paper
Previous research has observed that when contributors to a software project leave, the quality of the code and productivity of the team are impacted. This understanding has given rise to the notion of considering a project’s “bus factor” to determine how resilient it is to developer turnover. However, bus factor assumes a sudden and complete departure, and infers severe consequences. This study sought to provide a clearer understanding of the problem.
This study was carried out by conducting qualitative interviews with software developers and engineering managers working at different companies. After the data was analyzed, there was a follow-up survey to validate the results.
This paper has two main contributions: a framework for characterizing the types of departures that cause knowledge loss, and a description of a set of the implications of knowledge loss.
Characterizing the types of departures that cause knowledge loss
To better understand the implications of knowledge loss, a framework was devised to characterize the types of departures. The framework presents different departure events and their level of impact on teams along three dimensions: permanent vs. temporary, sudden vs. anticipated, and complete vs. partial.
Complete vs. partial departures. A complete departure means a knowledge worker is no longer available for knowledge sharing (e.g., they are not reachable, or they have forgotten the knowledge). A partial departure means a knowledge owner remains available for knowledge sharing on a discretionary basis (e.g., the person is still reachable after leaving the company or transitioning internally).
Sudden vs. anticipated departures. A departure is perceived as sudden if normal expectations about knowledge transfer cannot be met. Even when an employee gives notice, a departure can feel sudden when there is not enough time to transfer knowledge. Anticipated departures, such as leaves or internal transfers, are those that allow for knowledge sharing.
Permanent vs. temporary departures. Departures are permanent when the knowledge owner is not expected to return to the project. Departures are temporary when they are expected to return. Temporary departures are problematic when the information needs cannot be anticipated in advance of a leave, for example the team having to handle requests or incidents they haven’t experienced before.
This framework helps us see how the simple anticipation of a departure can play a major role in preventing knowledge loss. We can also observe the impact of internal transfers, which can amount to a complete departure from the point of view of knowledge loss, as the knowledge owners progressively lose their knowledge.
The implications of knowledge loss
This study also looked at the experience of knowledge loss from the perspective of those dealing with it. Four themes emerged: lacking guidance and information, relying on documentation, working with colleagues, and recreating the knowledge.
The theme lacking guidance and information generally captures the disorientation caused by the disengagement of the knowledge owner. The person or team lacking the knowledge may be uncertain and feel like they need to make guesses to move forward.
A team may also feel they need to rely on documentation that is hard to find or insufficient.
Developers may attempt to rely on socialization to recover some of the lost knowledge, or in other words ask around to other colleagues to see if anyone was involved in relevant projects.
Developers may also attempt to recreate the knowledge by doing their own research and learning.
Final thoughts
While the SE Research newsletter has summarized research on the related topics of developer satisfaction and happiness, this is the first issue that covers the topic of turnover. I see so much still yet to be explored as far as research that helps us understand what causes attrition, as well as the impacts of it on teams. However, this study provides a helpful starting point for organizations to discuss potential mitigation strategies to one of the more well-known consequences of attrition — knowledge loss.
That’s it for this week! Share your thoughts and feedback anytime by replying to this email.
If this email was forwarded to you, subscribe here so you don’t miss any future posts:
-Abi