How are buffers sized?
The most common basic approach to sizing buffers for a particular chain of tasks is to start with the 2-point range of estimates, and use half of the difference between the sum of the longer “safe” estimates and of the shorter “aggressive but achievable” estimates. Since buffers are based in the ability of “luckier” tasks making up for the “unlucky” tasks, when chains of tasks get short, with perhaps only four or less tasks, and alternative calculation based on the “square root of the sum of the squares of the differences” of these two task estimates is sometimes used. Buffer size is not limited to this accounting for task variation alone; they can also be subject to modification/augmentation to deal with other sources of variation, such as iteration uncertainty.
