The purpose of this paper is to present contaminated interaction as a design construct. Interactions with an object can be altered, positively, neutrally or negatively, due to some prior use. In such cases, the interaction departs from the designed condition and is said to be contaminated. This is particularly significant as objects, physical or non-physical, have multiple uses or are shared amongst users. We propose an ontological model of contaminated interaction based on a review of literature and an analysis of user experiences. The model outlines the process of contaminated interaction including the drivers and outcomes. In a negative context, contamination can lead to consumers misusing, negatively experiencing, or avoiding the object altogether. Positive contamination sees the opposite effect in which usability can increase, users report more positive experiences and users seek out or cherish the object. Together, this model presents an approach to understanding and addressing contamination in the design process to enable the creation and maintenance of meaningful experiences.

This work continues development and evaluation of a practical approach to a Design Research Methodology developed previously [1] and extended to practical application in [2] and concentrates on empirical techniques for research evaluation. We test a method of integrating an ethnographic approach with analytic observational techniques. The ethnographic study proved a rich source of novel insights into designers' knowledge searches during normal design activities over a 9-week period. These insights were used in conjunction with a diary study and interview data to interpret the self-report and observational data. This lead to the formation of models of the search process and the development of a number of categorisation schemes for search. A methodological analysis of a number of these schemes (1) a search tree based representation, (2) a feature commonality model, was made leading to the adoption of a scheme strongly based on the ethnographical insights rather than an abstract model. Some preliminary results from the use of this model are described and methods for verification and validation of the approach discussed.