What-If Demolition: Home-First Narrative Anchoring for Public-Facing Gentrification Analytics
Authors
Ibrahim Al-Hazwani (University of Zurich), Yanick Bachmann (ETH Zurich), Mennatallah El-Assady (ETH Zürich)
Presentation
- Session
- Data, Meet Human: Vis That Cares
- Time
- Wednesday, Nov 11, 11:03 – 11:12 (US/Eastern) · session 10:00 – 11:30
- Location
- Hall America north
Keywords
Narrative visualization, Urban analytics, Data humanism, What-if analysis
Abstract
Predictive urban models capture rich spatiotemporal dynamics of housing markets, but their outputs rarely reach the residents most affected by neighborhood change. We present What-If Demolition, an interactive gentrification explorer that couples an Attentive Neural Process for rent prediction with a narrative interface designed for non-expert civic audiences. We abstract a transferable pattern from the design process, which we call home-first narrative anchoring, that grounds exploration in the user's own address, progressively disclosing citywide patterns and "what-if" scenarios under uncertainty. We operationalize four principles of Data Humanism as design decisions and reflect on the deployed prototype through a study with eight residents of Zurich. Participants engaged with counterfactual scenarios anchored in their own homes, but also identified interpretive risks related to uncertainty and causality, which we discuss as lessons for public-facing predictive visualization. These findings suggest that home-first narrative anchoring can generalize beyond housing to other civic domains where spatiotemporal predictions must be made legible to the general users they affect.
For Practitioners
The paper would be most relevant to three groups of practitioners: 1) Visualization designers and HCI researchers: The home-first narrative anchoring pattern and four design decisions offer a transferable template for wrapping any civic predictive model in a resident-facing narrative interface. 2) Data journalists: The paper's treatment of counterfactual framing and uncertainty communication offers a principled vocabulary for responsible interactive storytelling with predictive models.