This paper describes FACIL, a classifier based on decision rules and border examples that avoids unnecessary revisions when virtual drifts are present in data. Rules in FACIL are both pure - consistent - and impure - inconsistent -. Pure rules classify new test examples by covering and impure rules classify them by distance as the nearest neighbor algorithm. In addition, the system provides an implicit forgetting heuristic so that positive and negative examples are removed from a rule when they are not near one another.