“…Prior works formalized security problems using game theory (e.g., FLIPIT [41], [24]), "weird machines" [10], attack trees [43], Markov models [40], and other methods. Prior notions of attacker quality include O-complexity [9], expected information loss [38], or success probability [30], which is similar to our concept of ∀ versus ∃-attackers. Attacker synthesis work exists in cyber-physical systems [33,5,20,26,30].…”