The Pagoda Design Space: Extending the Scope of Design - Principia Designae: Pre-Design, Design, and Post-Design - page 57

combining expertise and design creativity in one design space in which different
components are independent yet free to interact in a complex network of links, is
a more affording account of the notion of design that extends from pre-design to
post-design.
5.1 Design Space
The term design space was coined in artificial intelligence (AI). It has a wide range
of interpretations but typically it denotes the “space of possible designs for behav-
ing systems” (Sloman 1995 , p. 1), where the systems may be natural or artificial,
and they include “architectures, mechanisms, formalisms, inference systems, and
the like” (Sloman 1995 ) . The design space is explored in order to “characterize
diverse behavioral capabilities and the environments in which they are required, or
possible” (Sloman 1995 ). Design space exploration (DSE) has become a major field
of study in AI and engineering, especially software design. In engineering design,
or problem-solving, “the design space is a representation of all possible solutions”
(Westerlund 2005 , p. 1), and design space exploration suggests exploring alterna-
tive solutions (e.g., Saxena and Karsai 2010 ; Woodbury and Burrow 2006 ) .
Instances of design, or alternatives, inhabit the design space in the form of repre-
sentations. In engineering we encounter the model-driven design approach, wherein
these representations are often models, and accordingly we find definitions such as:
“Design space exploration (DSE) aims at searching through various models
representing different design candidates” (Heged¨s et al. 2011 , p. 1). The purpose
of DSE is to arrive at a high quality solution in a cost-effective manner while
ensuring that a sufficient number of alternatives are compared and assessed against
goals and constraints. According to Woodbury and Burrow ( 2006 ), DSE research
typically “addresses representation, search algorithms, task description, or interac-
tion design ... most research focuses on design states and on making action
explicit” (p. 64). They go on to state that “relatively little work focuses on the
design space itself ... Yet, it would appear that the design space itself is where the
largest gains are to be made” (Woodbury and Burrow 2006 ) . As this research topic
is based in AI, most of the work on the design space concerns itself with the
computability of design spaces and computational access to them (Woodbury and
Burrow 2006 ). However, the design space, albeit under different appellations, has
also been addressed by researchers from other fields, notably architecture. Some of
this work is quite old by now, and does not refer to computation. We shall briefly
review samples of this work, which is relevant to the current discussion.
When the notions of problem space and later solution space were introduced
(Newell and Simon 1972 ), models of problem solving in various domains were
crafted to fit the principles on which they were based. Beside a goal, the basic
problem space contained a set of states of knowledge, and operators by which states
are changed from one to another. In addition, the problem space encompassed
constraints on applying operators, and what was termed 'control knowledge' by