Sunday, February 16, 2014

Related Work: TaskGenies


TaskGenies: Automatically Providing Action Plans Helps People Complete Tasks
Nicolas Kokkalis, Thomas Köhn, Johannes Huebner, Moontae Lee, Florian Schulze and Scott Klemmer
TOCHI, 20 (5): 2013.

The proverb, “A journey of a thousand miles begins with a single step,” is attributed to the sixth century BCE philosopher Laozi. People have attempted to accomplish large tasks by decomposing them into manageable parts for millennia, and modern approaches to time management continue to take Laozi’s words to heart. For example, agile software development breaks large, failure-prone software projects into smaller tasks with corresponding time estimates.

The TaskGenies paper presents a crowd-based approach for breaking tasks down into action plans, or the concrete steps necessary to implement the tasks. You can try the approach out for your own tasks the TaskGenies website. Because context and consistency are important in creating plans, TaskGenies uses just a few crowd workers to create each plan. To support the creation of quality plans, workers are provided with related plans to use a models part way through the creation process.

The approach is the first that I am aware of to attempt to automatically create action plans for people. Of course, not all tasks are well suited to being broken down by the crowd. Crowd workers cannot create plans for very personal tasks that require context, or for poorly described, vague tasks. In such cases, the plan-maker probably needs the intimate knowledge typically held only by the task-doer. Additionally, the process of creating a plan may help a person reflect on the task they are about to undertake. Nonetheless, the research presented in this paper suggests the crowd can often provide useful plans, and that it may be the presence of a plan – versus the process of creating it – that really helps a person get things done.

To study whether people could successfully complete externally created action plans, the authors asked several hundred to share 20 self-selected personal tasks such as “mop the floor” or “update my resume.” Participants were then asked to use action plans for their tasks that were created one of two ways:
  1. The crowd created the action plan for the person, or
  2. That person was prompted to create their own action plan.
They found that people who were provided with action plans by the crowd were more likely to finish their tasks than those who were prompted to create their own plan. Crowd-created action plans appeared particularly useful for tasks that were “lingering” (specifically “have been lingering on [the task owner's] to-do list for a while”), “high level” (as judged the by the task owner, v. “small & well-defined”) and “not vague” (as judged by crowd workers). Of course, the tasks that a task-owner thought were “high level” were also more likely to be called “vague” by crowd workers, which suggests the crowd is best at breaking down tasks that are sufficiently well described but not too well defined.

While the findings presented in the paper are enticing, I would really like to see the same study design applied to instances where participants were not just prompted to create action plans for themselves, but also actually created them. My guess is that most participants, despite the prompting, did not actually create plans for themselves, making it hard to say anything about how they would have performed with self-created action plans. Additionally, capturing those plans would provide an interesting counterpoint to the crowd-created plans. Crowd-created plans may, for example, lack important details because their creators lack the necessary context, or they may provide fresh insights and diverse perspectives.

No comments:

Post a Comment