Value of knowledge and the problem of epistemic luck

Abstract

Imagine that you’ve just spent the last several months reading Don Quixote—and that you’re all but fifty pages away from finishing. Unfortunately for you, the book was due back before you could finish, and so begrudgingly, you turn it back in, having not known what happens in the end. Riddled with curiosity, you make your best guess about Quixote’s eventual fate and suppose it is the most likely scenario. Entirely unbeknownst to you, it turns out that you were right; Quixote’s ultimate destiny was just what you had supposed it would be! What luck! Quite naturally, we would say that (despite how impressed we are that you rightly anticipated Cervantes), when all is said and done, knowing what happens to Don Quixote in the end is surely better than merely believing truly what happens in the end—the predicament you find yourself in having not actually finished the book. After all, it was just by dumb luck that you guessed the ending right—a point you could deny only on the pain of some embarrassing claim of clairvoyance. We might put the idea more generally by saying that from a purely cognitive standpoint, it is better to know the truth than to stumble upon it by luck. This general idea betrays two distinct insights about knowledge. The first is the insight that while true belief is valuable, knowledge is distinctively so—knowing the truth is valuable in a way that merely having a true (but not-known) belief is not. The second insight here is that you lack knowledge if it’s just by dumb luck that the belief you have is true. Call these the value insight and the anti-luck insight: Value insight: Knowledge is distinctively valuable. Luck insight: Knowledge excludes luck. In contemporary epistemology, and especially over the past five years, separate projects have arisen in correspondence with these distinct intuitions: value-driven epistemology is concerned with issues surrounding the first insight, and projects under the description of anti-luck epistemology have arisen out of the second. Now we might reasonably suppose that whatever it is that makes knowledge relevantly un-lucky would be something we could cite in accounting for what makes knowledge distinctively valuable. This natural idea reflects the thought that the insights about value and luck should not be entirely disconnected. There is a problematic tension though between this reasonable expectation and the resources epistemologists have provided for us to accommodate it. What explains the tension is the fact that value-driven and anti-luck projects in epistemology have by and large been developed apart from each other, each focused on one of our two guiding insights at the expense of the other. Consequently, value-driven epistemology’s focus on the normative but not modal properties of knowledge leaves these two aspects of knowledge disconnected much in the way that anti-luck epistemology’s focus on the modal but not the normative properties of knowledge leaves these same aspects disconnected. The lacuna here between what value driven epistemologists tell us about what makes knowledge valuable and what the anti-luck epistemologists tell us about what makes knowledge exclude luck is troubling. We may resist that there should be such a disconnect if we avoid the common flaw shared by each of these projects pursued in isolation from the other. The flaw here is essentially one of naivety: that of supposing that we can give an account of knowledge exclusively in terms of conditions that would accommodate one of the two insights, while still managing to account for the other insight—which itself we did not appeal to directly in our theory of what knowledge is—e.g. in the analysis provided of it. This is the flaw behind the value-driven approaches that think about knowledge in terms of valuable properties they can’t explain to ensure modal robustness and the anti-luck projects that think of knowledge in terms of modal properties into which we can’t well smuggle the normativity needed to explain its value. To avoid this flaw, then, we should let both of these insights dictate the conditions our analysis offers as essential to knowing. The project with which I’ll be engaging here develops substantially on this widely overlooked and promising idea.