ABSTRACT
This chapter examines those circumstances when automatic item generation (AIG) would be useful and when it would not be. Because AIG is so tightly bound to computerized testing, those circumstances when computerized testing is sensible are examined. In addition some problems with computerized testing are discussed, as are some plausible solutions. These solutions are tied to Zipf’s Law which is illustrated in several situations ranging from the use of verbal items on the computerized version of Graduate Record Exam to the number of deaths due to horsekicks in the Prussian army between 1875 and 1894. I conclude with reflections on the wisdom of the Talmud.
