# Predicting how long a task will take

As a software engineer I have always struggled with predicting how long a project would take by underestimating and I have noticed that all of my colleagues have the same problem. In this video at 3:32 Luke Jones labels this problem The familiar road phenomena where people underestimate how long a task will take when they are familiar with the task.

• What studies have been conducted on the familiar road phenomena?
• Why do humans do this and what is the evolutionary advantage?
-
More as a comment: Evidence Based Scheduling - Joel on Software and search for "unrealistic optimism" or optimism bias plus many delays are due to "unpredictable" problems, so you can't just include them in a schedule. –  Piotr Migdal Feb 5 '13 at 17:08
The book Software Estimation: Demystifying the Black Art has a good example that shows that most people's assumption of "90% confidence" is drastically below 90%.. it's in fact closer to 20% confidence. The more familiar you know something, the narrower your estimate will be, and the more likely you'll miss your estimate. –  Muz Feb 28 '13 at 10:08