Feb 27, 2013
My idea is software which automatically implements productivity strategies, measures the effectiveness of those strategies, and analyses which strategies work best for you. Hopefully, using the software would result in a sustained increase in your productivity over time.
By "productivity strategies" I mean things like: the recommendations in the the anti-procrastination algorithm, the pomodoro technique, exercising regularly, pre-commitment, experimenting with sleep patterns, gamifying your tasks and so forth.
In practical terms, what I'm envisioning is an extensible software framework. The core program would be a simple task list manager: add tasks to be done in the future, check off items as done when completed and send notifications to the user.
This core framework would then be extended by plugins, which represented different productivity strategies. For example, the pomodoro plugin might make your first task at 9am each morning to review your task list and choose the most important three tasks (MITs), your second task to set and begin a timer for 30 minutes and your third task to complete that top MIT you chose. After 30 minutes, it would add a new task of taking a five minute relaxation break and send you a notification to let you know. Five minutes later, it would notify you again to finish your relaxation break task, with a fresh task to re-start the timer and then back to your MITs for a further 30 minutes.
The software could independently activate and deactivate the plugins in order to collect sufficient data to suggest which strategies were most effective for you. Over time, more plugins would be written as people made further suggestions. Existing plugins could be potentially improved and automatically reviewed using A/B testing.
When deciding whether a strategy is "effective", I mean that a large number of tasks are completed, that the remaining number of tasks on the list is small and that the age of those tasks is not too great. However, the criteria could be extended to ask for an indication of mood from the user, to allow for low stress optimisation, for example. Perhaps stochastic self sampling would work well here.
If users were willing to opt into providing anonymous data, the software could automate a community review of the strategies: which strategies seem to be most commonly effective? Affinity analysis could even be used to recommend plugins that were helpful to other people who responded to similar strategies as you.
What are your comments, and specifically criticisms, of this idea? Would you try using software like this if it existed? Would you like to assist in writing software like this?