The planning fallacy: why projects usually take longer than the estimate

Students writing their thesis were asked for a finish date in case everything went badly, and fewer than half made even that one. Why do estimates of software, and of any project, come up short, even from people who know they ran late before? The planning fallacy, the inside view, unknown unknowns, delays that add up while early finishes vanish, and Hofstadter's law. Then what beats a gut feeling: the outside view, reference class forecasting, and what to promise instead of one date.