A Generative Pre-Design Framework for Early-Stage Architectural Problem-Framing Reconceptualizing Optimization in Architectural Practice

This paper is aimed to address a critical misalignment between research and practice in applying computational design optimization to early-stage architectural design. A critical challenge exists in using the optimization method designed for structured problem-solving to address the ill-defined and wicked nature of early architectural design phases. In response, this paper introduces a generative pre-design framework that reconceptualizes computational optimization by positioning it as a source of analytical design prototypes that are generated from abstract massing logic. The framework leverages these prototypes as a medium for analysis and reflection, helping designers deconstruct performance trade-offs and synthesize knowledge, operating through a feedforward mechanism where insights actively shape problem definition rather than merely refining solutions. Developed through critical synthesis of real-world applications using EvoMass, the framework demonstrates how optimization can support reflective design exploration during initial problem-framing and conceptualization before conventional design development begins. These examples show how optimization can generate valuable insights for comprehensive, performance-informed problem-framing, finally supporting a paradigm shift in using computational optimization in early-stage architectural design.