Except real life is not a program, and the input data is flawed (human and machines' errors). The acceptance tests are just predictions, based on, again, fallible analyses of the flawed data from history. So many layers of errors that compound