Designing in the Loop, Co-evolutionary Design
- From biological analogy to collaborative intelligence
In recent architectural discourse, the analogy between biological systems and computational design processes has served as a provocative conceptual framework. While such analogies are often useful for framing ideas, they can easily lapse into rhetorical flourishes lacking epistemological rigour. A more productive pathway emerges when these analogies are operationalised as methodologies rather than metaphors. This article focuses on the concept of co-evolutionary design, where designers and computational systems engage in a mutual, adaptive design process inspired by, but not limited to, biological co-evolution.
Beyond the Metaphor: Co-evolution as a Design Methodology
Biologically, co-evolution refers to the process by which two or more species reciprocally affect each other's evolution (Thompson, 1994). In design, this concept is abstracted to describe mutual adaptation between human designers and computational agents. Rather than using evolutionary models solely as generative tools, co-evolutionary design involves an iterative feedback loop, where both human and machine adapt in response to each other's outputs (Frazer, 1995; Goulthorpe, 2000).
This methodology challenges the traditional notion of the designer as the sole author, positioning the machine as a co-creative partner. The dynamic interplay between designer intention and algorithmic behaviour forms a recursive loop of hypothesis, generation, evaluation, and refinement (Oxman, 2006).
Co-evolutionary approaches often make use of evolutionary algorithms (EAs) - computational models inspired by natural selection, mutation, and fitness evaluation (Deb et al., 2002). While EAs typically simulate selection based on predefined fitness criteria, co-evolutionary systems differ by integrating designer input as part of the selection mechanism.
Tools like the Snowflake plugin exemplify this model. Developed as a multi-objective optimisation platform grounded in the SPEA2 algorithm, Snowflake enables designers to iteratively guide the evolutionary process based on subjective aesthetic or performative judgments (Motlib & Eiz, 2023). The result is a system where computational logic and human creativity co-determine the trajectory of form-finding.
The Epistemology of Co-evolution
The philosophical implications of co-evolutionary design align with emerging models of distributed cognition and extended intelligence (Clark & Chalmers, 1998; Parisi, 2015). In this view, intelligence is not confined to the mind of the designer, but is distributed across systems, tools, and feedback mechanisms. This epistemic shift resonates with Bateson’s (1972) idea of ‘ecology of mind’, where learning and creativity are seen as systemic processes rather than isolated acts of cognition. By embedding the designer within a responsive, evolving system, co-evolutionary design enables a situated mode of intelligence - one that is adaptive, emergent, and deeply contextual.
Additionally, co-evolutionary design offers a compelling framework for addressing the complexity of contemporary architectural challenges, from climate-responsive buildings to computational fabrication. It reframes design not as a static act of problem-solving but as a living process of negotiation between competing forces, constraints, and intelligences.
By foregrounding interaction over imposition, co-evolution as a methodology encourages designers to engage the built environment as a system of interdependencies - where adaptation is not a byproduct, but the central mechanism of design intelligence.
References
Bateson, G. (1972). Steps to an Ecology of Mind. University of Chicago Press.
Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7-19.
Deb, K., Pratap, A., Agarwal, S., & Meyarivan, T. (2002). A fast and elitist multiobjective genetic algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6(2), 182-197.
Frazer, J. (1995). An Evolutionary Architecture. Architectural Association.
Goulthorpe, M. (2000). The possibility of (an)architecture. Architectural Design, 70(1), 26–31.
Motlib, Z., & Eiz, H. (2023). Snowflake: An Interactive Multi-objective Optimization Plugin for Parametric Design. [Unpublished plugin documentation].
Oxman, R. (2006). Theory and design in the first digital age. Design Studies, 27(3), 229–265.
Parisi, L. (2015). Instrumentality or the intelligent instrument? In Computational Culture, Issue 5.
Thompson, J. N. (1994). The Coevolutionary Process. University of Chicago Press.
