> Parameters for the model are estimated from literature examples of known scandals, and the factors influencing conspiracy success and failure are explored.
I've only read the abstract so far, so it might be they try to account for this - but I think they are at risk of being hit by some version of the "WW2 aircraft" problem if they only use historical data.
Basically, if you say, "all historical conspiracies eventually were exposed" than that's sort of a truism - if they weren't exposed, we wouldn't know about them.
In the same sense, if you make a statistical model about conspiracies and only base it on the conspiracies that became historically known, you're sort of biasing your data towards failed conspiracies. That might lead you to underestimate what kind of conspiracies are possible that just successfully managed to keep secrecy.
Of course the other extreme is just as bad. You can always say "well that's what they want you to think / well they are just that good at hiding..." at which point your actual conspiracy theory becomes unfalsifiable.
Concretely re your question: yes, such a conspiracy is conceivable, however, such large conspiracies are probabilistically doomed to get exposed: https://journals.plos.org/plosone/article?id=10.1371/journal...