In most of social sciences prediction is based on extrapolation of central tendencies. In reality, almost everything that is important is the consequence of a rare event. The alphabet model describes how seemingly unimportant rare events may govern social dynamics.