This is just the usual wisdom on why estimates are off - nothing new here, except for an emphasis on regularly re-prioritizing.<p>The real problem lies in not recognizing that some things either can't be estimated, or to get an estimate takes just as long as doing the thing.<p>Example 1: Everyone's favorite magically autonomous puppet, Pinnochio! Imagine Pinnochio at rest. A simple system. What happens when ol' Pinnoak blurts out "My nose is growing!" What then, Jiminy Cricket?<p>Example 2: Solving Rubik's Cube. Simple methods exist (<a href="http://www.dickgrune.com/Cult/RubCub/" rel="nofollow">http://www.dickgrune.com/Cult/RubCub/</a>). You only need 20 face-turns to solve it (<a href="http://en.wikipedia.org/wiki/Optimal_solutions_for_Rubik%27s_Cube" rel="nofollow">http://en.wikipedia.org/wiki/Optimal_solutions_for_Rubik%27s...</a>). Your Official CMMi Level 5 process requires you to have a written procedure so as to repeatedly be able to solve a problem within schedule and budget bounds. It's faster to take Dick Grune's document and solve the damn thing, rather than look the cube over, write down all 20 moves, and then do the minimum number of moves.