Thus starts an analogy from Michael Wolfe to explain why software development task estimations are regularly off by a factor of 2-3.
