This can be especially important if using the size of a particular font to approximate a certain width. It stinks to see carefully crafted navigation break to the next line.

Sweeping statements about the size and growth of the Internet are tough to swallow. So I'll break down the McKinsey report and pull out some wonderfully illustrative graphs to help it go down.

Naive Bayes classifiers often break down when the size of the training examples per class are not balanced or when the data is not independent enough.