The efficient representation and manipulation of data is one of the fundamental tasks in the construction of large software systems. One of the most successful approach to date has been Hindley-Milner polymorphism which provides predefined mechanisms for manipulating data structures providing they are parametric in the data.

I will talk about recent work on containers which has lead to some intriguing new insights into polymorphism. In particular I will classify all the polymorphic programs between containers and show they have a remarkably simple form. I will then extend this to talk about derivatives of data structures as an example of generic programming.