The Role of the Retrospective Meetings in Detecting, Refactoring and Monitoring Community Smells

TL;DR

This paper examines how retrospective meetings in agile software development surface and address social debt expressed as Community Smells. Using semi-structured interviews with 15 practitioners and a qualitative analysis that blends a priori smell coding with Grounded Theory–inspired open coding, the study focuses on four smells: Lone Wolf, Organizational Silo, Radio Silence, and Black Cloud. Findings show retrospectives effectively detect smells and typically elicit refactoring actions, but follow-through and monitoring are uneven, while emphasis on positive aspects can help prevent smells. The work provides practical guidance for practitioners to strengthen action-item monitoring and outlines directions for researchers to evaluate refactoring strategies and their effectiveness across contexts.

Abstract

Retrospective meetings play a vital role in agile development by facilitating team reflection on past work to enhance effectiveness. These meetings address various social aspects, including team dynamics, individual performance, processes, and technologies, ultimately leading to actions for improvement. Despite their importance, limited research has explored how these meetings handle forms of social debt, particularly Community Smells -- recurring dysfunctional patterns in team dynamics, such as poor communication or isolated work practices. This study seeks to understand how retrospective meetings address a few core Community Smells, examining whether these meetings help identify smells, make it possible to formulate refactoring strategies, support the monitoring of refactoring actions, and contribute to preventing the most prominent Community Smells. We conducted semi-structured interviews with 15 practitioners from diverse organizations who regularly participate in retrospective meetings. The interviewees shared their experiences with retrospectives, the challenges discussed, and subsequent improvement actions. The study focused on the four most cited Community Smells in the literature -- Lone Wolf, Organizational Silo, Radio Silence, and Black Cloud. Data was analyzed iteratively using a priori coding to examine Community Smells and inductive open coding inspired by Grounded Theory. The findings indicate that retrospective meetings indeed enable the identification of core Community Smells. However, while strategies for refactoring are often formulated, their implementation and monitoring remain inconsistent. Additionally, an emphasis on positive aspects during these meetings may help in preventing Community Smells. This study offers valuable insights to practitioners and researchers, highlighting the importance of addressing social debt in software development within agile practices.

Figures (6)

• Figure 1: The coding process applied to the analysis to answer the research questions
• Figure 2: Detection of Community Smells in Retrospective Meetings
• Figure 3: Refactoring of Community Smells in Retrospective Meetings
• Figure 4: Monitoring of Community Smells in Retrospective Meetings
• Figure 5: Prevention of Community Smells in Retrospective Meetings