Literature Review

In addition to secondary research from books and my own primary research, I reviewed academic journals to better understand the state of the art and and to see if I could find any foundations for my thesis from other sources.

In their study on fragmented work, Mark, Gonzales, and Harris cite their earlier study that claims people are interrupted about every three minutes. When a worker is interrupted, 77.2% resume the work they had been doing before the interruption, but after about 25 and a half minutes of doing another (not necessarily distracting) activity. Mark et al classify interruptions as either external, such as a coworker asking a question, and internal, where a worker decides to switch to a different task of their own volition. Typical workers see about an even split between the two kinds of interruption.

In “A Diary Study of Task Switching and Interruptions,” Czerwinski, Horvitz, and Wilhite recorded a diary study of information workers over a week, and they found similar results: 40% of interruptions are self-initiated.

The internal/external difference was an observation that appeared in my interviews, as well. Interview subjects frequently cited other people as being among the most distracting things in their environment. The statistic of a 50% split is also interesting because it helps me determine the possible scope of my solution. If I address only internal interruptions, I would only address 50% of the problem.

In “Self-interruption on the computer: A typology of discretionary task interleaving,” Jin and Dabbish describe seven types of self-interruptions: Adjustment, break, inquiry, recollection, routine, trigger, and wait. Of the seven, distraction would fall under the break category, which is defined as, “Temporarily switching to a more desirable task because of frustration or fatigue with the primary task.” The three causes of distraction (avoidance, boredom, and completion) would fall under this type.

In “Effects of instant messaging interruptions on computing tasks,” Cutrell et al studied the usage of instant messaging and how it impacted computing. In their study, they noticed the concept of “chunking behaviors:”

...it took longer to switch to [a new instant] message during our execution phase is reminiscent of the idea of chunking behaviors. This refers to the tendency to delay switching to another task until completion of a subtask (e.g., typing the search words). We informally observed users not attending to an interruption until they had completed typing their search keywords. The significant effect of relevance on message reading and task resumption times suggests that when interruptions are unrelated to ongoing tasks, they will take longer to process and it will be more difficult to get back on task following the interruption.

Conceptually, this is similar to the idea of “milestones” I had used, where the system would encourage users to take breaks, switch tasks, or post a status message after completing a significant piece of work.

In the paper “The effects of task interruption and information presentation on individual decision making,” Speier et al address the question of what happens to users given the frequent interruptions and distractions. Based on their research study, they found that as the number of interruptions increases, the quality of the decisions that are made is reduced.

In summary, I found that the academic literature in the topic was largely in agreement with what I had found through my own research. I continued on to create concepts and test a few prototypes.