← Blog

Background for an Executive Productivity Aide

bundo productivity time GTD RIP executive function research

tl;dr: This is the first in a series of items I will be publishing on bundo. I talk about productivity systems and applications, empirical studies on time blindness among human experts, and then synthesize the idea of a new productivity application called bundo.

Landscape of Productivity Systems

Getting Things Done (GTD)

GTD (by David Allen) focuses on capturing, clarifying, and organizing tasks to clear mental clutter, while the Eisenhower Matrix (urgency vs importance) provides a framework to prioritize them. Taken together, the GTD process is a structured system which is a fully-mapped and widely discussed way of externalizing the sort of centralizing executive function tasks needed for creative work. The following describe the GTD system in general:

Understanding Tasks

  1. Begins with a 'brain dump' to write down and chronicle all tasks, ideas, and commitments in one place
  2. These are then tagged, or categorized, placing similar tasks together; ie, 'work', 'personal', 'family', etc.
  3. The user evaluates the impact of the task or commitment: 1 low impact, 2 medium impact, 3 high impact
  4. The user evaluates the urgency of the task: A low urgency, B medium urgency, C high urgency
  5. Actions to take on tasks can then be assessed, as follows.

Eisenhower Matrix Task Assessment

  • low impact + low urgency -> defer/archive/delete
  • low impact + medium urgency -> delegate
  • low impact + high urgency -> quick action
  • medium impact + low urgency -> schedule the task
  • medium impact + medium urgency -> prioritize
  • medium impact + high urgency -> focus and resolve
  • high impact + low urgency -> plan
  • high impact + medium urgency -> action
  • high impact + high urgency -> do now

Decision Tree on Work in progress

From Stuff (tasks, commitments, ideas) 'in': -> if not actionable, then either eliminate or incubate -> if actionable, continue

What is next action? -> Projects -> planning and review actions for plans for projects (sub tree traversal) -> Tasks -> do anything which will take less than 2 minutes now! -> delegatable tasks - checked on a daily or weekly loop -> deferred tasks: either scheduled via calendar or queued

Task Execution

  1. using SMART goals, break down any given task into defined steps with logical sequence and knowable metrics measured and timebound due dates
  2. focused work in 30-50 minute blocks, without distractions within blocks.
  3. evaluation immediately after the block, and at the end of the day allowing adaption

Randomized Iterative Productivity

  • Developed by Paul Town, and available as a paid application, this method uses tasks which are either minute-metric tasks, or unit-metric tasks (minutes of programming, sets of pushups, cups of water, etc.)
  • The main idea is that it's a hassle to write out what you're going to do each day, especially if you're doing similar things each day.
  • The psychological feeling of accomplishment from checking off daily lists is quickly diminished if it is presented on a daily basis rather than tracked and displayed holistically.
  • Therefore, the RIP system randomly generates, based on the metric of the activity, and the number of times per week the user would like to do that task, with a magnitude range of metrics for completion; then, the application generates a semi-random range and frequency of activities daily for the user to perform.

Puke Planner

  • Now canceled application teased by Conrad Barski and never openly discussed beyond teaser marketing
  • Minimizes data entry and human analysis paralysis by only allowing three input metrics per task: discomfort capacity of task (symbolically, the puke), the time the task will take, and [to be used sparingly] how close a task's deadline is.
  • Every day, the user gives the planner the amount of total time they can work, and a 'puke budget' of how hard they feel up to working that day.
  • Overall design idea: humans are actually bad at picking the right things to work on.
  • Purported "finely tuned ten-factor task selection algorithm" was teased, but never documented by Dr Barski (unfortunately) which assigns tasks based on the budget any given day.
  • Does not seem to have any understanding of periodicity of tasks.

Other Systems

  • Many todo systems provided as applications often just an extension of the Pomodoro Technique without leveraging the power of computerized systems. A to-do list + a timer is potentially helpful but not often ideal.
  • Other systems attempt gamified reinforcement (with 'streaks', point systems, or leaderboards,) but this can set someone up for failure, or create extrinsic pressure that initially motivates but rapidly produces anxiety.

Real-time Scheduling and Human Estimation

All of the above systems describe systems in which

In Why software projects take longer than you think, Eric Bernhardsson writes on the difficulty of getting human experts to correctly estimate the time a task might take. Human experts intuitively understand the median, rather than mean or skew, of the time it takes a task to complete, leading to time estimates which are an honest and unbiased estimate, yet do not cover contingencies. This means that by and large, a human expert will report an estimation which does account for the real world probability density of the task they are describing: we can consider that the overall distribution of actual time has a probability distribution with a long tail, a median at some n = 1.0, and a mean of 1.65, giving an inferred inverse gamma distribution with alpha = 4.97, and beta = 4.13, such that the 'blowup' time factor of the task at the 99th percentile is 32x longer than the estimated time.

Executive Dysfunction and Planning

Neurodivergence or mental health issues can lead to a deficiency in executive function, decreasing the ability for the individual to plan, centralize, and integrate information. For people who may be suffering from ADHD, ASD, GAD, or other issues, the above systems can fail in the following ways:

  • GTD fails at the review and prioritization stage. The "brain dump" is perfect for externalizing, but the complex matrix and weekly reviews become a "wall of awful" that inspires avoidance.
  • RIP fails at project-based work. It's brilliant for "build consistency in habits" (e.g., exercise, practice) but can't handle a multi-step project with a hard deadline.
  • Puke Planner (conceptually) fails at periodicity and complexity. It brilliantly addresses activation energy and spoon theory but has no mechanism for recurring tasks or project dependencies.

A New System

Bundo is a proposed application and real-time scheduling system which proposes to synthesize the above systems and extend them, with the goal to improve the productive lifestyles of individual creative people who may be neurodivergent.

Using the take-aways from previous systems, we can provide the user with:

  • A way to minimize data entry portions and thinking about the tasks themselves, externalizing many executive function tasks
  • A single system which handles appointment-like tasks, habitual tasks, and complex deadlined/project tasks in a single processing queue.
  • A real-time scheduling system which utilizes both empirical research in time management of creative people and understandings in neurodivergent time perception and energy.
  • An app which does not provide a surface for attempts at direct gamification of the system, but which instead encourages self-evaluation.
  • An algorithm which combines algorithmic and probabilistic elements to prevent boredom, based on user preferences.