CONCEPT
Anomalous Problems
Problems a tradition's own commitments predict should not exist — more diagnostic of degeneration than ordinary unsolved problems, because their existence indicates that something in the framework's core is wrong.
Laudan's framework distinguishes sharply between unsolved problems and anomalous problems. An unsolved problem is one a tradition has not yet addressed, which is ordinary and appears in every tradition. An anomalous problem is different: it is a problem the tradition's own theoretical commitments predict should not exist. Anomalies are evidence that the framework generates predictions the world contradicts — that something in the tradition's core requires modification. A tradition with many unsolved problems may still be progressive if it is developing resources to address them. A tradition with accumulating anomalies is degenerative regardless of its other successes, because the anomalies indicate a structural failure that will eventually produce empirical collapse.
In The You On AI Field Guide
The distinction between unsolved and anomalous problems is subtle but decisive. An unsolved problem is a gap in the tradition's achievements. An anomaly is a contradiction between what the tradition predicts and what the world shows. Unsolved problems are normal. Anomalies are diagnostic.
Applied to the AI