Why software projects take longer than you think – a statistical model

Erik Bernhardsson analysis SiP Effort Estimation dataset to illustration why software estimations turn out to be inaccurate and summaries his conclusion:

The mean turns out to be substantially worse than the median, due to the distribution being skewed (log-normally).

Tasks with the most uncertainty (rather the biggest size) can often dominate the mean time it takes to complete all tasks.