Dr. Jon Herlocker

The last straw fell mid-morning, roughly five years ago. I was a mid-tenure-track professor, working on a medium-importance but short deadline task to create a class module that I had promised my operating systems students for the class the next morning. As I was reading through my previous year’s course notes in preparation, a young seeker-of-knowledge chose to take advantage of my “always open-door” policy in order to enrich himself. I say that tongue-in-cheek – on this occasion it turned out this student was seeking to contest a grade I had given him. Nonetheless I engaged him enthusiastically and after some spirited debate, he left somewhat less dejected than he had arrived. When I returned to my work station, I sat for a moment thinking about what I needed to be working on, and settled on editing a grant proposal. The rest of the day quickly passed. When I arrived at class the next day, I was subjected to extreme embarrassment as the students demanded the forgotten class module I had promised them in preparation for the upcoming exam.

 Now, the information overload experts among you can immediately identify many tools and processes that I could have employed to help prevent the occurrence described above. For example, perhaps if I followed the Getting Things Done methodology, I would always have a list of tasks in front of me, reminding me that the course module was due the following day. Or perhaps if I had applied sophisticated interruption prevention techniques, like closing my door, I would have prevented the damaging interruption. But as a young computer science professor, I did the geeky thing – I thought about how the right technology could have prevented my woe by helping me to recover more effectively from interruptions.

What if there was software that knew, at every moment in time, the task on which you were working? Then perhaps whenever I activated my computer, it could tell me something like “you were last working on the course module for your operating systems class.” Thus was born the idea for TaskTracer (http://eecs.oregonstate.edu/TaskTracer) at Oregon State University and later its commercialized counterpart, Smart Desktop (http://www.smartdesktop.com/). Such software would run on my laptop or desktop computer, and would record my activity history. A majority of my life is channeled through my computer – be it reading email, browsing the web, reading and writing documents, writing software, or interacting with other software applications. Such a system could not only help me keep on task, it could also help me to restart tasks that had been interrupted days ago, by showing me records of my activities when I last worked on those tasks. For example, after working for some time on a different task, I might tell the software that I wanted to resume the “course module” task and it might in return tell me “when you last worked on the course module task, you were editing this presentation (cs411.ppt), would you like to open it?”

I view interruption recovery as a four stage process. First is the cognitive step of remembering that you should be working on a task. Second is the remembering what state in the process you were on that task. Third is remembering or generating a plan to reach the resources - documents emails, or web pages - that are necessary for the next step (likely the resources you were using when you were last interrupted on that task). Fourth is the physical manipulation of the keyboard or mouse to get you to those resources. All of the effort involved in this process is overhead – it doesn’t advance the task that you are trying to complete. Everything that we can do to reduce this overhead will reduce the cost of the interruption, and increase the amount of productive hours during the day. The TaskTracer research project demonstrated that software could reduce the overhead cost of each of the four stages of interruption recovery. In addition to reminding me which task I was working on, TaskTracer could present me with a list of the files, web pages, and emails that I had most recently touched when I was last working in the context of the task that I was resuming. This would trigger my memory about where I was in the task, and would virtually eliminate the overhead of recalling, locating, and opening the resources themselves.

We know personally that interruptions can be costly to our personal productivity. It is certainly important to find ways to remove or schedule some interruptions and I look forward to advances in those areas by my colleagues with the Information Overload Research Group. However, interruptions (or interrupts as we call them in computer science) are also one thing that enables us to be more productive through multi-tasking. Interruptions allow us to push tasks that are blocked (waiting for somebody else) out of our memory, freeing it up to work on other tasks, without having to continually check if those tasks are ready to continue. But in the end, there will always be interruptions, so there is considerable potential to apply technology to limit the effects of those interruptions on our productivity.

[Self-Disclaimer: the author is currently an employee of Pi Corporation, which is commercializing the TaskTracer/Smart Desktop described in this blog and a co-founder of the TaskTracer group with Thomas G. Dietterich (http://eecs.oregonstate.edu/~tgd). For peer-reviewed articles describing TaskTracer and the research associated with it, please visit http://eecs.oregonstate.edu/TaskTracer . Or make up your mind independently by trying it yourself – http://www.smartdesktop.com/  . ]

About the Author
Jon Herlocker is currently Chief Technical Officer of the Smart Desktop division of Pi Corporation. Until just very recently, he was Associate Professor of Computer Science at Oregon State University researching information overload issues. For more information see http://www.smartdesktop.com/