A Causal Safety Criterion for Knowledge

Safety purports to explain why cases of accidentally true belief are not knowledge, addressing Gettier and based on statistical evidence. However, problems arise for using safety as a condition knowledge: is necessary knowledge cannot always the evidence it meant address. In this paper, I argue new modal designed capture non-accidental relationship between facts required causal safety. that possible errors in can be captured by accounting deviations relationships there natural way characterize which relevant an epistemic situation. Using this, develop analogue safety, where one’s p causally safe if all worlds one believes p. Causal argue, better address avoid arguments raised against necessity