Our work, in brief

Overview:

Our work began with a hard question: Why weren’t designers of safety-critical systems getting complete and correct requirements from the people who’d be using those systems, even though those users knew that they’d probably die if the system failed due to incomplete or incorrect requirements?

The easy-looking answers didn’t work; the problem wasn’t money, or motivation, or reluctance to talk about difficult topics. Finding a better answer took a long time, but we eventually found one. That answer made sense of other questions related to the same underlying problem; for instance, how best to find out what people really want, or think, or believe, in fields ranging from market research to social science, and from public opinion surveys to clinical psychology.

That answer also sheds light on issues such as why some visual representations make more sense than others, why people make the types of mistake that they do, and how best to transfer knowledge in teaching and training.

We applied these insights to some hard real-world problems, and found solutions that previous researchers had missed. One was a problem that had defied the world’s best codebreakers for over eighty years, the Voynich Manuscript.

Over the last couple of years, we’ve been building this body of work into a systematic framework that produces clear, practical insights and methods from the underlying theory and principles. This website contains the key parts of that framework.

Our work falls into four main areas.

Elicitation involves ways of gathering information from human beings. We’ve been working on this since the 1980s. A key finding from this work is that interviews, questionnaires and focus groups only get at the knowledge that people do talk about; you need different methods to get at the knowledge that people don’t talk about (for instance, because they think it’s not worth mentioning) and to get at the knowledge that people can’t put into words (for instance, highly practised physical skills) or won’t put into words (for instance, trade secrets, or taboo topics). Our work draws heavily on a framework developed by Gordon Rugg and Neil Maiden for mapping these knowledge types onto appropriate elicitation methods.

Representation involves ways of presenting information and knowledge to people; our work in this area has mainly involved finding representations that will give better insights, and/or will reduce the risk of error and misunderstanding. One example of this is our work on treating liking and disliking as two separate concepts, rather than as opposite ends of a single scale.

Error has far-reaching implications. Our work has involved not only simple everyday errors, but also errors in expert research into difficult problems, such as code breaking, and autism. This article in Wired magazine describes some of that research.

Education theory and practice can be improved by drawing on findings from elicitation, representation and error. For instance, elicitation methods help find out where a student is misunderstanding a concept; choosing the appropriate representation can help that student see the concept correctly, and understanding human error can help prevent that misunderstanding from happening in the first place.

More about our work: An extract from the Knowledge Modelling Handbook