Content area
Full text
Author for correspondence: Nathan C. Brown, E-mail: [email protected]
Introduction: diversity in interactive computational design
In its most basic definition, design is the act of creating a plan for the construction of an object, system, or action (Design, 2017). The early stages of any design process often include rapidly generating a variety of potential solutions to a problem through brainstorming. Historically, this aspect of design has been driven exclusively by humans, frequently accompanied by sketching or other forms of representation to help communicate and clarify ideas (Goldschmidt, 1994). Especially in the field of architecture, a premium has often been placed on creativity and innovation during this process, valuing the ability to come up with many different ideas for possible buildings and structures, as well as ideas that have not been proposed before. Design in architecture and related fields has always required balancing aesthetic, technical, functional, economical, and other concerns, with different priorities taking precedence depending on the building and its context. Increasingly, computers are able to help analyze how a design is performing, navigate the relationships between different design goals, or even overcome human cognitive obstacles (McCaffrey & Spector, 2018). Despite these advances in how design priorities are pursued, fundamental aspects of design remain synthetic, creative, and human.
As such, many computational design methods require interaction with humans primarily in order to be effective, but also to appeal to human designers as a potential creative aid in the first place. In facilitating effective interaction, computers can be much faster than humans at certain tasks such as rapidly generating design variations within a provided scheme and evaluating their performance (Cvetkovic & Parmee, 2002), but they must do so in a way that enhances the creative process rather than restricting it. In many cases, this means that computers are used to create sets of potential design options, and these sets are presented to a human for comparison. Unless explicitly implementing optimization in the rare case that a single numerical objective overshadows all other considerations, the possibilities generated by computers must be varied and different, rather than all essentially representing the same or similar options. Otherwise, they cannot provide human designers with the flexibility to exert preferences or simultaneously consider design goals that are more difficult to quantify.
The...