Amazing power of pairwise independence

In the currency of random bits, it is far cheaper to generate pairwise independent random variables than completely independent ones. This simple construction, viewed and used in a variety of ways, has found a blizzard of applications in theoretical computer science. I will survey the ideas leading to some of the following applications: (1) Derandomizing probabilistic algorithms and constructions (2) Utilizing weak random sources (3) Deterministic amplification (4) Efficient hashing schemes (5) The complexity of unique solutions (6) BPP and approximate counting in the polytime hierarchy (7) Public vs. private coins in interactive proofs (8) Pseudorandom generators for logspace (9) One-way functions, pseudorandomness, and average case complexity. Below we give a partial list of references. For each one we marked the application(s) to which it is relevant.