Hands-on representations in a two-dimensional space for early stages of design

Abstract

In design, problem analysis is as important as solution synthesis. Strategic knowledge is required not only for constructing a solution but also for framing a problem. While externalized representations play critical roles in design tasks, different types of representations are necessary for different stages of a design task. In early stages of a design task, design support tools need to provide hands-on representations with which a designer can easily perform trial-and-error and examine the whole as well as parts of the whole, allowing the designer to represent any levels of preciseness, as he/she likes. Sketching and drawing with paper and pencil provide an ideal representation for this process. But what about supporting design domains, such as writing or programming, where no sketching exists? In this paper, we argue that two-dimensional positioning of objects in a design support tool serves for the same purpose as sketching does for architectural design. Two-dimensional positioning allows a designer to produce hands-on representations that “talk back” to him/her without forcing the designer to formalize or verbalize what to be externalized. Two systems, ART for writing and RemBoard for component-based programming, illustrate the framework.

Introduction

In ill-structured design, a problem and a solution co-evolve [1], [2]. In architectural design, writing, or programming, for instance, what components need to be constructed (problem analysis) and how they need to be integrated (solution synthesis) depend on each other — parts define the whole but the roles of parts are defined by the whole; a design process can be viewed as forming a hermeneutic circle [3].

Throughout a design process, a designer is engaged in a cycle of producing a representation (such as sketches, mockups, notes and solution forms), and reflecting on them [4]. The externalized representations serve as a “situation” that talks back to the designer. During the process, the designer has a conversation with a material asking questions such as:

• what parts are missing;

• how much the designer is “sure” about a newly created part;

• what the role of this newly created part is in terms of the whole design;

• what the role of this newly created part is in terms of other parts; or

• which direction the whole design is moving toward and whether the direction is in accordance with the intention behind the design.

A type of strategic knowledge that our research focuses on is related to low-level design decisions required to address these questions.

Such strategic knowledge is necessary in early stages of a design task when understanding what the problem is plays a large role. The strategic knowledge is required to explore a possible problem space by uncovering implicit requirements, by making trade-off among conflicting goals, and by setting up constraints.

Appropriately applying the strategic knowledge to a design situation requires the “right” kind of external representation. Most existing design support tools focus on providing representations in a solution domain. CAD systems, for instance, allow designers to produce detailed pretty-printed representations, allowing them to do precise simulation and detailed analysis of a produced artifact. On the other hand, these representations support designers very little in answering questions listed above.

In early stages of a design task, designers produce rough sketches using paper and pencil. Programmers record hand-written notes. Not all of those externalized representations would constitute the final solution. Most sketches are in fact, drawn as a part of a trial-and-error process or produced as a result of doodling, and abandoned later. Those representations serve for the application of the strategic knowledge in framing a problem by triggering creative ideas.

This paper presents our approach of supporting early stages of a design task by providing representations that are suitable for the application of the strategic knowledge. In early stages of a design task, design support tools need to provide hands-on representations with which a designer can easily perform trial-and-error and examine the whole as well as parts of the whole, allowing the designer to represent any levels of preciseness, as he/she likes. Acknowledging the widespread use of sketches and drawings, these can be considered as ideal representations for early stages of architectural design. We argue that two-dimensional (2D) positioning of objects serves as the same purpose as sketching does for domains where no sketches exist, such as writing or programming. We focused on the positioning of objects in the space and what types of positioning, including placing, moving, resizing and merging, emerge during the design process. Such objects can be any type of representations including parts of a final product, comments, or design rationale.

In what follows, we first argue for externalized representations for early stages of a design task. We discuss sketches and drawing as one of ideal representations and identify requirements for such representations. Section 3 presents 2D positioning of objects as an alternative representation for domains where no sketches exist. Section 4 presents two prototyped systems, ART for writing and RemBoard for component-based programming which illustrate the framework. The following sections discuss the two systems and conclude the paper.