CT's are just folktales and explanations for things people don't understand or control. A conspiracy theory only becomes dangerous or harmful when it threatens to upset an established order, which makes it oddly self fulfilling, since its real purpose is to facilitate organizing people around counter-establishment narratives. When you look at this paper as a new way to use technology to automatically detect counter-establishment narratives, it seems like pretty standard playbook for a secret elite coordinating to secure and expand their powers, which is hilarious, to me anyway.
However, the conclusion includes criteria for detecting actual conspiracies as separate from theories, "We hypothesize that three features—a single domain of interaction, a robustness to deletions of nodes and relationships, and a proliferation of peripheral actants and relationships—are key characteristics of an actual conspiracy and may be helpful in distinguishing actual conspiracies from conspiracy theories. "
I think they buried the lede on that one, as a heuristic for evaluating CT's and accusations of them could be super valuable.