We survey the set-theoretic methods of module theory that make it possible to equip roots of the contravariant Ext functor with filtrations built from the small roots. The power of these methods is illustrated by several applications: a solution to the Kaplansky problem on Baer modules and some of the related problems for relative Baer modules, the structure of tilting modules and classes, the structure of Matlis localizations of commutative rings, and in particular cases, proofs of the finitistic dimension conjectures, and of the telescope conjecture for module categories.The bifunctor Ext 1 R plays a crucial role in understanding properties of the category Mod-R of all right R-modules over an associative unital ring R.Though Ext 1 R (A, B) = 0 whenever A is a projective module or B is an injective module, in general it is quite difficult to characterize the zeros of Ext, that is, the pairs (A, B) such that Ext 1 R (A, B) = 0. Given a class of modules C, we will call a module A a root of Ext for C provided that Ext 1 R (A, C) = 0 for all C ∈ C. For example, if R is a commutative domain and C is the class of all torsion modules, then the roots of Ext for C are called Baer modules. Kaplansky asked for the structure of these modules in 1962 [35], but only recently a complete answer was given in [2]: the Baer modules coincide