Encounter Or Exposure? Simulating The Social Consequences Of Open-Plan Interior Layouts With Visibility Graphs, Agents And Machine Learning

Interior layouts shape who meets whom at work, yet the social consequences of everyday design decisions are rarely tested under controlled conditions. This study used a fully computational experiment to examine how five interior decisions in open-plan offices, namely core position, tea point strategy, workstation cluster size, desk screen height and the enclosure of meeting rooms, affect unplanned encounters and visual distraction. A full factorial set of 108 parametric layouts with 72 workstations each was analysed with visibility graphs and simulated with an agent-based model of a four-hour working block, and 60 randomised layouts served as an unseen test set. Most simulated encounters occurred beside desks and were tightly coupled with distraction, whereas encounters away from desks were unrelated to it. High screens reduced distraction by 32% with almost no loss of away-from-desk encounters. A single central tea point produced almost twice as many away-from-desk encounters as two distributed counters while causing the least distraction, larger workstation clusters reduced the share of cross-team contact, and enclosing meeting rooms had negligible effects. Machine learning surrogates trained on conventional visibility descriptors predicted unseen layouts poorly, but adding four simple route and catchment indices raised the explained variance to between 94% and 98% with a regularised linear model. The findings suggest that interior architects should move encounters from desks to shared nodes, and that theory-informed descriptors allow fast and transparent screening of layouts at the sketch stage.