When the representation of a data structure changes we would expect that the behaviour of clients using the data structure remains “the same”, provided that the data structure is fully encapsulated.

How do we demonstrate such representation independence? This unpolished talk will describe work in progress.