Chapter 2

2.1 Motivation and History

When we use a computer for solving a statistical problem, most of the time we trust the computer that the solution will be (at least approximately) correct. Today, nearly no statistician thinks about such basic questions like “What can a computer compute?”, “How does a computer compute?”, or “How exact does a computer compute?”. We just “trust.” Knowledge about such basic facts threatens to disappear since we nearly always act as if the computer always produces the correct answer. To understand that this is not true is one of the aims of this chapter. In particular, we will discuss the above questions and what the answers mean for statisticians.