The talk gives a retrospective birds-eye view on hinges and hinge trees and their applications. Originally, hinges ans hinge trees were introduced in the context of databases, more specifically join dependencies (JDs) and hypergraphs. Hinge trees are a tool to decompose a JD into an equivalent set of JDs with fewer components. In addition, (1) hinge trees can be computed in uniform polynomial time, (2) the largest number of edges in a hinge of a hinge tree is an invariant, called the degree of cyclicity (or hinge tree width), and (3) the degree of cyclicity can be computed in polynomial time. A classical characterization of acyclicity can be generalized in that degree of cyclicity at most k is equivalent to k-wise consistency implying global consistency. Here, rephrased in CSP terminology, k-wise consistency means that each constraint tuple can be extended to a partial solution on any k scopes including the scope of the given constraint tuple, and global consistency, also known as minimality, means that every constraint tuple can be extended to a global solution. Later, hinges resurfaced in the context of constraint satisfaction problems (CSPs). It was shown (under moderate conditions on the size of the domains) that, in the context of minimal constraints, a set of scopes is a hinge if and only if every partial solution to that set can always be extended to a global solution. Also, a hybrid algorithm was proposed, where tree clustering is applied to the nodes of a hinge tree. Finally, the place of hinge trees in the theory of decompositions was discussed in the talks of Scarcello and Cohen.