Tuesday, January 06, 2015

What makes a prototype novel?

At the NordiCHI  Conference in October, Mikael Wiberg and I presented a paper:

Wiberg, M. & Stolterman, E. (2014) What Makes a Prototype Novel? – A Knowledge Contribution Concern for Interaction Design Research. NordiCHI '14, October 26 - 30 2014, Helsinki, Finland.

The paper is about something anyone engaged in HCI research is familiar with. Here is how we present it in the introduction of the paper:

"Every time we review a paper describing a new interactive system, every time we go to a conference or when we are presented with a new interactive system from industry we repeatedly find ourselves asking “Have I not seen this system before?” or stating “This system does not remind me of anything I have ever seen...”. These questions are related to a fundamental research consideration concerning how it is possible to conceptualize and relate different designs to each other and to the existing body of knowledge. In short, we address the question “what makes a prototype novel?” and accordingly a manifestation of something new, a knowledge contribution to our field. "

This is both a simple and difficult question. How do we know that a new design actually represents a new knowledge contribution or not? Most of us in the field agree that prototypes are crucial as part of the knowledge production process, but in what way? In the paper we discuss some of the earlier attempts that have been made to solve this question. We also propose some potential ways to move forward in a more structured and formal way. We propose the notion of generic design thinking or concepts. 

Below is the Discussion section from the paper. Of course it is not easy to understand just the final section of a paper, but it may give a sense of what we are trying to do. If you are interested in these questions, let us know.


When generic design concepts are used in architecture, there are two ways of handling designs that are new: as novel or as unique. Importantly, for something to be considered new it is not sufficient merely to be novel in the sense of having “odd properties”. Instead, for a design to be unique or new, it must involve at least one of the following criteria: 

the application of an established generic model to a new problem or in a new domain 
a design that combines elements from multiple established generic models 
the addition of a new element to a known generic model manifested in a design 
a combination of a new generic model and a design that defines a new design space such that the design demonstrates the potential scope of the new space. 

In this context, novelty that stems from an evolution of a design’s underlying model reconfigures the landscape of design spaces; if done particularly well, it creates new space within this landscape that others can join and exploit. 

We see several important implications of this suggestion for the advancement of HCI design research. 

First, generic design thinking reject designs that are not properly situated within a web of existing and already known design ideas. The new cannot be advanced without understanding how it relates to existing design ideas. That is to say, a new ultimate particular (a concrete design) needs to be anchored in the general (that is, in some theoretically articulated idea). 

Secondly, generic design thinking implies a shift in focus away from specific properties of a given ultimate particular towards generic dimensions in new designs. This shift has implications for what we need to express with a particular design. It also raises questions about which factors should be incorporated into a design and which can be omitted when designing prototypes during the research process. This could potentially reduce the difficulty of developing research prototypes as fully implemented systems (and the need to include a lot of specific system features, etc). 

Thirdly, generic design thinking implies the need for more deliberate work in HCI on the formulation of classes of interactions. Today, direct manipulation, embodied modes of interaction, and agent-based interaction models could be seen as some relatively stable classes that are important for the formulation of generic design principles in HCI. But what other kinds and ways of grouping interactions and interaction technologies can we imagine? And how can we move forward and become more specific? And what are the existing good examples? 

Finally, generic design thinking provides a practical tool to improve our ability to compare and evaluate different designs. In this way, it could provide a foundation from which to address design quality and to make judgments about designs that are rooted in more than just the opinion of an individual designer. This aligns well with the proposed concept of interaction criticism [2]. 

In this paper we have proposed, described and exemplified generic design thinking in the format of a four-step method and approach to systematically move forward (design) while also more systematically understand and learn (analyze) from past designs. Although we have so far only described this as a first draft of a method we are convinced that this approach might redirect our field slightly from being heavily future-oriented to also acknowledge the utility of working backwards from a design to its conceptual roots – to trace design ideas through the analysis of designs. Importantly, generic design approaches require critical analysis of the history of design within HCI in order to anchor the new and novel in the history of ideas. 

In wrapping up our paper, we should return to its basic message. We do have recent research stressing the importance of concept-driven design research [25] and we do have a good understanding of how ‘strong concepts’ [13] can advance our field. At the same time we lack methods to systematically relate different concepts to each other and in relation to particular designs. Here is where generic design thinking can play an important role as to systematically advance our field while keeping our design-driven approach.  Given this take on the subject we should state that in order to answer the most central question for design-driven HCI “when is a new design a knowledge contribution?”, we must first, as a field of research, establish the method and approach for guiding the systematic work of conceptualizing and theorizing these designs. In this paper we have suggested ‘generic design thinking’ as an initial attempt to move in the direction of the development of one such method and approach."

1 comment: