Three Pages That Broke It

In 1963, Edmund Gettier published a paper of about three pages in the journal Analysis. It was called "Is Justified True Belief Knowledge?" and the answer was no.

He did not argue against the definition in general terms. He did something much harder to answer: he constructed cases where all three conditions hold and where calling the result knowledge is obviously wrong. He gave two. Here is one in the form it usually gets taught.

You look at the clock in the hall. It says two. It is, in fact, two. You believe it is two, it is true that it is two, and you are entirely justified, because that clock has been reliable for years and trusting a working clock is exactly what a reasonable person does.

The clock stopped twenty-four hours ago.

Every condition is satisfied. Belief, truth, justification. And nobody wants to say you knew the time, because you would have believed it was two whatever time it actually was. You were right by accident, using a method that was not connected to the fact at all.

That is the shape of every Gettier case. The reasoning is good. The conclusion is true. And in the middle, where the connection between them should be, there is luck.

What is worth noticing is what happened next, because it is not what you might expect.

Nobody concluded that knowledge is an incoherent concept. Nobody decided the question had been silly all along. The field accepted the counterexamples immediately, which is itself unusual, and went looking for the missing fourth condition.

The proposals are interesting and none has won. Perhaps knowledge requires that no false step appears anywhere in your reasoning, which handles the clock but struggles elsewhere. Perhaps it requires a reliable process, which raises the question of how reliable and reliable in which circumstances. Perhaps it requires that your belief tracks the truth, meaning you would not have believed it if it had been false, which is elegant and has generated its own counterexamples.

Sixty years of this. The concept is still in use, still doing work, still without a definition that survives examination.

That fact deserves more attention than it usually gets, because it is not a failure. Most of the concepts you rely on every day are in exactly this position. Nobody can define a game, or a chair, or a heap, in a way that survives a determined objector, and this has not made those words unusable.

What Gettier showed is narrower and sharper than the general point. He showed that a specific, confident, precisely stated definition, held by serious people for two thousand years, could be taken apart by somebody who thought carefully about cases for a few pages.

That is worth keeping in mind whenever an account of something looks finished.