In this note we present the worst-character rule, an efficient variation of
the bad-character heuristic for the exact string matching problem, firstly
introduced in the well-known Boyer-Moore algorithm. Our proposed rule selects a
position relative to the current shift which yields the largest average
advancement, according to the characters distribution in the text. Experimental
results show that the worst-character rule achieves very good results
especially in the case of long patterns or small alphabets in random texts and
in the case of texts in natural languages.