Cheryl Stevens has started in interesting discussion on her Plain language, please Facebook page, about responding to clients who complain bitterly about trivial typos. Her point is that typos are actually inevitable because perfection in any kind of quality control is elusive. She cites the guru of quality management, W. Edward Deming.
I responded with an anecdote from my time working at the Open University.
I recall that one of our mathematics textbooks was found by students to contain quite a number of errors which had escaped the proof-readers. The errors were corrected and it was reprinted. The students then found further errors but fewer of them. This went on for several editions, and the story goes that the authors then used the declining number of errors in each edition as an exercise for students to calculate the probable number of errors still remaining.
As if to prove the point, just after I wrote the paragraph above I noticed a typo, so went back in to edit. The next morning I noticed another error and did the same thing. Immediately I noticed yet another... oh dear.
Adrienne Montgomerie responded by drawing attention to Ray Panko's website reviewing research on human error. He includes a list of research on proofreading - none of the studies found 100% of errors caught.