Christian Kohls, Till Schümmer: Do you believe in patterns?

Patterns capture proven design solutions to recurrent problems. It is a method to generalize and extract the invariant features of good forms and best practices which have been tried and tested in real settings. However, there are still no agreed standards to judge upon the scientific validity of documented patterns. This makes it difficult to establish pattern mining as a research tool and it opens the door for misuses, e.g. to propose patterns without evidence.

What can we do to justify the knowledge that is captured in patterns? First, we have to understand the type of knowledge found in patterns and its implications. Are patterns science or art? Second, we have to analyze the current methods and paradigms in pattern mining and discuss their appropriateness for scientific research. There are qualitative methods such as interviews (DeLano, 1998), mind mapping (Cunningham, 199x), expert workshops (Mor & Winters, 2007; 2008), or the Writer’s Workshop (Gabriel, 2002). But are there also quantitative methods, e.g. questionnaires for evaluation or cluster analysis? Is the "Rule of Three" (Biggerstaff & Richter, 1987) sufficient for significant findings? Third, we have to reflect about the benefits and liabilities of patterns as a research tool and its capability to generate knowledge. Where is its place in epistemology? Is its format of reporting adequate?

The paper illustrates the differences and relations between patterns in the real world, the patterns in our heads, and the documented patterns as epistemological mini-theories (Kohls, 2008), and compares the design/generation spaces of each. It distinguishes between validity of structure, usefulness and usability and between proposed, observed and applied patterns. We will argue that just like any research paper needs to describe the methodology used, pattern papers should name the empirical source, the state of confidence and the research method used to mine the pattern.