Classical Invariantism and the Puzzle of Fallibilism

This paper revisits a puzzle that arises for theories of knowledge according to which one can know on the basis of merely inductive grounds. No matter how strong such theories require inductive grounds to be if a belief based on them is to qualify as knowledge, there are certain beliefs (namely, about the outcome of fair lotteries) that are based on even stronger inductive grounds, while, intuitively, they do not qualify as knowledge. This paper discusses what is often regarded as the most promising classical invariantist solution to the puzzle, namely, that beliefs about the outcomes of fair lotteries do not qualify as knowledge because they are too lucky to do so (or, relatedly, because they do not satisfy a safety condition on knowledge), while other beliefs based on potentially weaker inductive grounds are not too lucky (or, relatedly, because they are safe). A case is presented that shows that this solution to the puzzle is actually not viable. It is argued that there is no obvious alternative solution in sight and that therefore the puzzle still awaits a classical invariantist solution.