This paper presents a novel methodology for defining and analyzing the dynamics of the collaboration networks of scientists working on general relativity from the mid-1920s–1970. During these four and a half decades the status of the theory underwent a radical transformation: from a marginal theory before the mid-1950s to a pillar of modern physics. To investigate this passage—known as the renaissance of general relativity—we used a definition of collaboration networks broader than the co-authorship relations retrievable from online datasets. We constructed a multilayer network, in which each layer represents a different kind of collaboration. After having analyzed the evolution over time of specific parameters of the co-authorship network, we investigated the effects of adding one type of collaboration edge at a time, in a cumulative fashion, on the values of these parameters and on the topology of the collaboration network through time, including rapid shifts in the dynamic evolution of the largest component. This analysis provides robust quantitative evidence that a shift in the structure of the relativity collaboration network occurred between the late 1950s and the early 1960s, when a giant component started forming. We interpret this shift as the central social dynamic of the renaissance process and then identify its central actors. Our analysis disproves common explanations of the renaissance process. It shows that this phenomenon was not a consequence of astrophysical discoveries in the 1960s, nor was it a simple by-product of socio-economic transformations in the physics landscape after World War II.