zlacker

>>mattjh+(OP)
Something I learned on this site: We're bad at estimating the average duration, but reasonable for the mean duration.

If you get 10 tasks of seemingly equal duration, 9 will go well and 1 will hit a reef of unexpected troubles and take forever.

So the practice of doubling is not that stupid. It leaves time in the first 9 to deal with the unexpected disaster.

>>hyperm+on1
except "work expands so as to fill the time available for its completion", and therefore, after each of the 9 tasks is completed, there is no time left for the 10th task. so this only works if the disaster happens very early, or if you do almost all of the tasks in parallel. neither is the expected case.