When experts from different fields collaborate—such as software engineers working with user experience designers, or data scientists partnering with ethicists—they often hit a wall. Each discipline brings its own specialized jargon, distinct values, and preferred metrics. Direct interpersonal communication frequently breaks down because what is obvious to one group is completely opaque to another.
To bridge these gaps, teams rely on boundary objects: artifacts that inhabit multiple intersecting social or professional worlds and satisfy the informational requirements of each. Rather than forcing everyone to adopt a single perspective, boundary objects act as anchors for communication. They allow diverse stakeholders to collaborate on shared tasks without needing to reach a total consensus on underlying values.
The Structural Duality of Boundary Objects
Why do boundary objects work? Their effectiveness stems from a unique structural duality: they are flexible enough to adapt to local needs, yet robust enough to maintain a common identity across different domains.
Consider the challenge of bridging user-centered design and software engineering. As explored by Mieke Haesen et al. in MuiCSer: A Process Framework for Multi-disciplinary User-Centred Software Engineering Processes, task models can serve as a communication bridge precisely because of this dual nature. They are intuitive enough for non-technical stakeholders to validate, yet structured enough for software engineers to implement.
Similarly, in machine learning development, Alex Bäuerle et al. demonstrate in Symphony: Composing Interactive Interfaces for Machine Learning how interactive data interfaces act as boundary objects. These interfaces allow data scientists to conduct deep exploratory analysis while simultaneously providing high-level stakeholders with a clear visual context to understand model behavior. The artifact changes its utility depending on who is looking at it, yet it keeps both parties anchored to the same underlying reality.
Translating the Abstract into the Tangible
Boundary objects do not just display information; they actively translate abstract, complex concepts into representations that can be manipulated and evaluated by all parties.
This translation mechanism is highly visible in socio-technical domains like AI ethics and fairness. In industry settings, data scientists often think in terms of mathematical fairness metrics, while product managers and designers focus on qualitative user impacts. Wesley Hanwen Deng et al. show in Investigating Practices and Opportunities for Cross-functional Collaboration around AI Fairness in Industry Practice that practitioners bridge this gap by co-creating shared glossaries and co-evaluation documentation. These artifacts translate abstract statistical metrics into concrete descriptions of user harms, aligning engineering goals with qualitative insights.
The same principle applies to physical engineering and interdisciplinary learning. When students from different disciplines collaborate to build a complex system, a physical tool can serve as a shared interface. In Weaving Collaboration: Promoting Effective Interdisciplinary Learning with a Robotic Loom, Santiago Ojeda-Ramírez et al. found that a robotic loom forced participants to translate abstract theoretical concepts from their respective fields into a common tangible output. The physical artifact becomes a site of mutual teaching and shared exploration, making abstract concepts actionable.
Collaboration Without Consensus
A common misconception is that successful collaboration requires everyone to agree on everything. Boundary objects prove otherwise. They enable what the authors of Emergent Knowledge Artifacts for Supporting Trialogical E-Learning call "trialogical learning"—where groups collaboratively build and evolve shared knowledge artifacts rather than relying solely on direct interpersonal alignment.
As argued in Towards a global participatory platform by Simon Buckingham Shum et al., tools like interactive visualizations and argument maps allow diverse stakeholders to cooperate on highly complex problems without requiring them to reach a total consensus on underlying values or goals.
Similarly, Sophia Worth et al. highlight in Brokering Ethics: A Retrospective Study on the Use of a Data Ethics Framework how data ethics frameworks and canvases function as non-specialist languages. They allow individuals across different organizational levels—from legal teams to engineers—to voice concerns and align on project risks without needing to master each other's specific methodologies or share the same professional worldview.
Strategic and Cultural Anchors
Boundary objects are often thought of as passive documents, but in practice, they operate as active strategic tools. When tech organizations distribute human-AI design guidelines, they are rarely used as strict rulebooks. Instead, as Nur Yildirim et al. discovered in Investigating How Practitioners Use Human-AI Guidelines: A Case Study on the People + AI Guidebook, practitioners use resources like Google's People + AI Guidebook as boundary objects to build team culture, align cross-functional stakeholders, and win arguments for user-centered features. By providing a common technical vocabulary, the guidebook helps stakeholders negotiate complex system requirements and technical trade-offs.
The same mechanism underpins a wide variety of collaborative tools across disciplines:
- Qualitative Hierarchical Models: In A qualitative, network-centric method for modeling socio-technical systems, with applications to evaluating interventions on social media platforms to increase social equality, the authors use an abstraction hierarchy to link low-level technical components directly to high-level societal goals, helping computer scientists and social scientists align their expertise.
- Novel Visualizations: In Reflections on the Interdisciplinary Collaborative Design of Mapping the Universe, the authors show how specialized data visualizations act as translators that enable astronomers and information scientists to converge on a common understanding.
- Functional Boundaries: In Human-Computer Interaction for AI Systems Design: Reflections on an Online Course on Human-AI Interaction for Professionals, the authors demonstrate that adopting a shared language of functions and system boundaries allows technical engineers and user experience designers to decompose complex goals into manageable sub-functions.
By shifting the burden of alignment from fragile interpersonal communication to robust, multi-interpretable artifacts, boundary objects allow interdisciplinary teams to navigate the messy reality of collaborative problem-solving.
Key Foundational Work
Foundations and profiles: splicing metaphors in genetic databases and biobanks (2005) — contributed 5 concept elements to this direction.
Emergent Knowledge Artifacts for Supporting Trialogical E-Learning (2006) — contributed 1 concept elements to this direction.