Proving lower bounds on complexity and inventing algorithms that have theoretically better complexity but are most times useless in practice (for example the O(

Proving lower bounds on complexity and inventing algorithms that have theoretically better complexity but are most times useless in practice (for example the O(n^2.376) algorithm for matrix multiplication).