Daniel Tunkelang wrote about Herb Simon’s attention economy and ways to measure the way people allocate attention. His example of attention-switching and interruptions with e-mail made me think about individual differences. People differ in the willingness to engage in an activity, and self-interruption is a common practice. You can measure time on task, but for complex cognitive tasks it is not clear that time is a good predictor of performance. The problem of measurement is more complex than simply aggregating times or counting switches.

In HCI, we have a notion of the fallacy of the average user — the notion that if you design for characteristics averaged over a large number of people, there may not exist a single person for whom the design is ideal. This due to the fact that certain phenomena have bimodal distributions rather than those with a central tendency. For example, Hudson et al. found that individual preferences in interruptibility suggested a bimodal distribution.

While often the reality of engineering is that you have to make compromises, it is still useful during design to think more contextually. For example, rather than asking how to prioritize e-mail-based disruptions, we should study the dependency of prioritization strategies on contextual factors such as the nature of the task the person is engaged in and on the cognitive style of the person. No single strategy is likely to fit all people or all situations, and often a combination of two different designs that optimize different (opposite) aspects can work better than a single “average” design. The challenge, of course, is to understand when each strategy is appropriate and how to allow people to control the application of different strategies in an effective and transparent manner.

Share on: