Providing action plans helps people complete tasks

People complete tasks more quickly when they have concrete plans. However, task lists place the burden of developing plans on the user. This paper investigates whether crowds can provide people with action plans. A betweensubjects experiment found that people who received action plans from the crowd completed more tasks than a control group that did not. We next describe two techniques for scaling this approach. First, we created a social variant where participants both used the task lists and provided action plans for others. In a between-subjects experiment, participants using this social approach completed more tasks than the control condition. Second, we developed a natural language processing technique that uses WordNet and synonym detection to group similar tasks and reuse action plans across them. Running this technique on a corpus of 2872 tasks with two human auditors of the results found that 69.4% tasks had at least one match. This suggests automatically providing action plans is quite scalable. We have incorporated these principles into a system, the TaskGenies Community. Author