When Domains Collide: An Activity Theory Exploration of Cross-Disciplinary Collaboration

Zixuan Feng, Thomas Zimmermann, Lorenzo Pisani, Christopher Gooley, Jeremiah Wander, Anita Sarma · 2025

Read When Domains Collide: An Activity Theory Exploration of Cross-Disciplinary Collaboration by Zixuan Feng, Thomas Zimmermann, Lorenzo Pisani, Christopher Gooley, Jeremiah Wander, Anita Sarma (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-20063.txt.

Introduction

Feng and co-authors study friction when domain experts and professional software developers build production software together. Mixed teams bring complementary skill, but clashing expectations and priorities cause strain. Their 2025 study interviews twelve experts and twelve developers, then validates the findings with a survey of nearly three hundred people in both roles.

Developers held eight expectations of their partners, and experts held six of theirs. Mapping those hopes onto a classic activity framework exposed twenty-one distinct frictions and showed where each arises. The authors name the pattern cross-disciplinary software development and give it a grounded model. They close with advice for researchers, team leads, and tool builders.

Catalog record

CREATORZixuan Feng, Thomas Zimmermann, Lorenzo Pisani, Christopher Gooley, Jeremiah Wander, Anita Sarma
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE59 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-20063

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

When Domains Collide: An Activity Theory Exploration of Cross-Disciplinary Collaboration Zixuan Feng, Thomas Zimmermann, Lorenzo Pisani, Christopher Gooley, Jeremiah Wander, Anita Sarma Abstract Background: Software development teams are increasingly diverse, embedded, and cross-disciplinary. Domain experts (DEs) from different disciplines collaborate with professional software developers (SDEs), bringing complementary expertise in creating and maintaining complex production software. However, contested expectations, divergent problem-solving perspectives, and conflicting priorities lead to friction. Aims: This study aims to investigate the dynamics of emerging collaboration of cross-disciplinary software development (CDSD) by exploring the expectations held by DEs and SDEs and understanding how these frictions manifest in practice. Method: We utilize Activity Theory (AT), a well-established socio-technical framework, as an analytical lens in a grounded, empirical investigation, conducted through a mixed-method study involving 24 interviews (12 DEs and 12 SDEs) and a large-scale validation survey with 293 participants (161 DEs and 132 SDEs). Results: We conceptualize and empirically ground the CDSD dynamics. We identified eight expectations held by SDEs and six by DEs. By mapping these expectations to AT components, we revealed 21 frictions in CDSD and illustrated where and how they arise. Conclusions: This study offers a theoretical lens for understanding the dynamics and frictions in CDSD and provides actionable insights for future research, practitioners, and infrastructure design. I Introduction Consider a team of three software engineers and three medical physicists co-developing an MRI image processing software for early cancer detection with fluid roles; physicists configure the CI/CD pipeline, while software engineers implement the image processing models, with all team members sharing code review and deployment responsibilities. Such domain expert embedded, “Cross-Disciplinary Software Development (CDSD)” teams, where domain experts and software engineers co-own the product and share responsibilities, promise richer expertise and faster innovation. Yet subtle differences in priorities and tooling literacy can produce misaligned expectations, which then manifest as friction in collaboration. While CDSD teams are becoming increasingly common—ranging from biology [ 1 , 2 ] to health [ 3 , 4 ] to the automotive industry [ 5 ] , among others, our understanding of how they operate is still unclear. Prior works on cross‐disciplinary work have focused on Domain Experts (DE) serving as clients [ 6 , 7 ] , in peripheral roles (e.g., data scientists as isolated consultants) [ 8 , 9 ] , or participating in hand‐off collaboration models [ 10 ] (e.g., siloed DEs produce Machine Learning models that are integrated by Software Developers (SDE)), leaving the dynamics of fully embedded, co‐owned cross-disciplinary collaboration largely unexplored. To unpack the complex, multi-actor, role-fluid interactions and frictions in CDSD, we use the lens of Activity Theory (AT) [ 11 , 12 ] , a well-established socio-technical framework that conceptualizes work as an interplay among subjects (SDEs, DEs), tools (infrastructure, pipelines), rules (engineering practices), community, division of labor, and shared objects [ 12 ] to produce the outcome of project success. AT is especially suited for our investigation as it explicitly models mediating artifacts, rules, and division of labor, enabling it to reveal how mismatches in expectations and priorities cause frictions. We use the lens of AT to investigate: RQ1 : How do SDEs and DEs coordinate through tools, rules, and shared artifacts when co-developing software in CDSD teams? RQ2 : What frictions arise from mismatched expectations in CDSD collaborations? We answer these questions through a mixed-methods study [ 13 ] , 24 interviews spanning 12 SDEs and 12 DEs from diverse organizations, projects, and regions within Microsoft, followed by a large-scale survey validation of 294 Open Source Software developers with experience in CDSD. By grounding our mixed-methods study in AT, we identified 14 expectations each group holds for the other, and a set of 21 collaboration frictions arising from mismatched expectations. The contribution of our work is threefold: (1) We apply AT to conceptualize and empirically ground the fluid team dynamics that emerge in cross-disciplinary software development, offering a theoretical lens to understand how shared ownership and blurred roles unfold in practice. (2) We map expectations of DEs and SDEs to AT components to surface how and where frictions arise, revealing that collaboration breakdowns stem from systemic contradictions within the activity system, such as misaligned expectations in rules, tools, and the division of labor. Our findings equip SE researchers and practitioners with a diagnostic framework to anticipate, surface, and resolve contestation before it escalates. (3) We leverage AT to identify collaboration friction hotspots in CDSD, particularly around rules, tools, and community to offer actionable insights to: (i) enable practitioners focus efforts on high-impact areas (e.g., setting up shared validation practices that balance SDE rigor with DE agility), (ii) guide infrastructure designers to develop adaptive tools (e.g., lightweight rule enforcement mechanisms), and (iii) inspire software engineering researchers to explore new coordination mechanisms (e.g., reimagining Agile to accommodate contributors with limited software engineering backgrounds). Our findings suggest that collaboration in CDSD fails quietly, at the boundary of assumptions, not ability; “There’s a lot of philosophical differences in software development practices.” [DE8] (interview participant). Our work makes those boundaries visible and offers a structured AT lens to navigate them. To facilitate reproducibility, we include the interview guide, codebook, and survey questions in the supplemental material [ 14 ] . We do not include interview transcripts or survey responses to maintain participant confidentiality and comply with our IRB protocol. II Background Collaboration in software development teams is a long-studied phenomenon [ 15 , 16 ] , but as technology and practices evolve, so does how collaboration is conducted, and the frictions involved in it. Here we focus on the collaboration aspect of how SDE works with DEs, which can be across the team border as clients or, as in more recent situations, embedded within the team itself with a shared outcome and client focus. We categorized the different types of collaborations into four models. Professional Collaboration typically involves software professionals [ 15 , 17 ] , often working for clients in industries such as finance, transportation, and government [ 18 , 19 ] . Responsibility for creating the product and its ownership remains entirely with the software team. Collaboration to build the software is internal to the development team, with checkpoints with clients as required in the development process [ 20 , 21 ] . Adjacent Collaboration introduces DEs (e.g., data scientists) as part of the software development teams, but relegates them to peripheral roles [ 8 , 9 ] . In the case of data scientists, they are primarily involved in data-related helper tasks, such as processing, analyzing, and modeling data to inform the system design [ 9 , 22 ] , and their involvement remains isolated from the main software development workflows [ 23 ] . Interfaced Collaboration is a more cross-disciplinary collaboration type [ 10 ] , where DE teams, such as machine learning engineers, work with SDE teams at specific stages [ 24 ] . This form of collaboration involved more interactions between the DEs and SDEs, but their development roles were still siloed: DEs focused on building models, while SDEs were responsible for production and deployment [ 10 ] . Such phase-specific engagements creates fragmented ownership, misaligned expectations, and integration bottlenecks [ 10 ] . Dependency management across the teams is uneven

Read the complete text