#separator:tab
#html:false
#notetype:Basic
#deck:Math::Learning to learn::The buffer hour
#tags:math learning planning
#columns:Front	Back	Tags
#tags column:3
What is the planning fallacy?	The habit of underestimating how long a task will take, even with experience of past overruns.	definition
Why do errors in task-time estimates not cancel out?	A task can hardly finish much early but can run far over, so the errors point one way.	skew
Can an estimate be honest and still too short to plan with?	Yes: the most likely time can be well below the average time when the misses are all late.	trap
In the model, a one-hour task averages how long?	1.35 hours: 1 h at 6/10, 1.5 h at 3/10, 3 h at 1/10.	skew
Five one-hour tasks planned in five hours: chance the day fits, in the model?	(6/10)^5, about 7.8%, one day in thirteen: every task must hit its estimate.	buffer
Five one-hour tasks with one buffer hour: chance the day fits?	46.7%, up from 7.8%.	buffer
Same five tasks with symmetric errors: chance a five-hour plan fits?	64%: early tasks pay for late ones, so skew is what breaks the plan.	skew
How do estimates actually improve?	Record estimated and actual time, take the average ratio, and multiply new estimates by it.	calibration
What is the buffer hour for, once estimates are scaled?	Days worse than average: the scaled plan matches the average day, not every day.	calibration
Set Up for Success: what if a to-do item is not done by its date?	Pick an extension; if it is not done by the extension, ask whether it needs doing at all.	definition