Abstract for: Regeneration of social housing estates: making mental models explicit using a novel CLD explorer

Housing in the UK is a wicked problem, shaped by conflicting interests and implicit mental models of what is desirable and achievable. Amid rising disrepair, child mortality, and rough sleeping, estate regeneration - particularly redevelopment - is often promoted as a silver bullet, despite contributing to a shrinking social housing stock. While retrofit is praised, redevelopment logics persist. This study aims to expose these tensions to support dialogue across the sector. CLDs help capture the diverse mental models underpinning complex issues. Yet communicating results remains challenging: aggregated diagrams risk distorting or erasing stakeholder boundaries and tensions. This study introduces the CLD Explorer, a tool that aggregates diagrams around a chosen variable while preserving each interviewee’s structure. Built in Python, it supports similarity-based mergers, interactive visualisation, and exploration of causes and consequences across stakeholders. We generated 38 individual CLDs in interviews with stakeholders involved in social housing regeneration. Aggregation was tested with an early version of the tool presented at ISDC 2025. The tool was subsequently improved through iterative development and verification tests. Validation compared manual and tool‑based mergers for CLDs from different stakeholder groups, focusing on causes and consequences of demolition. Initial findings show clear differences between housing association and resident perspectives. Our preliminary findings reveal both overlaps and differences in how stakeholder groups understand the drivers and consequences of social housing redevelopment. Minor discrepancies between manual and tool‑based mergers pointed to human error, underscoring the tool’s usefulness for managing large CLDs. Next steps involve merging similar causes and consequences and examining how emerging patterns align or diverge across stakeholder groups. AI was used for the tool development.