Failure of a conspiracy can be predicted by a model that assumes a certain level of uncertain loyalty spread out across the ranks of the conspirators. Conspiracies that would have to have a large number of conspirators, either initially or over time, would be more likely to fail over time than conspiracies involve a small number of individuals that do not need to recruit fresh conspirators.
Researchers created a model of conspiracy collapse and ran some of the more common conspiracies through it. The results? Large, global conspiracies involving thousands of players are almost impossible to maintain.
Conspiracies above from upper left: Moon Landing, Global Warming, Vaccine Manufacturers, Suppressed Cancer Cure.