The Gettier cases

What it tests Is knowledge the same as justified true belief, or does knowing need something more?

The two cases

Gettier’s paper, three pages long, offers two cases. In the first, Smith and Jones have applied for the same job. Smith has strong evidence that Jones will get it (the company president told him so) and that Jones has ten coins in his pocket (he counted them). He infers that the man who will get the job has ten coins in his pocket. But it is Smith who gets the job, and Smith, as it happens, also has ten coins in his pocket. His belief was justified, and it was true, but what made it true was not what he had evidence for.

In the second, Smith has good evidence that Jones owns a Ford, and from it he infers, quite correctly, a string of disjunctions: either Jones owns a Ford or Brown is in Boston; either Jones owns a Ford or Brown is in Barcelona; either Jones owns a Ford or Brown is in Brest-Litovsk. Smith has no idea where Brown is. Jones does not own a Ford, but Brown happens to be in Barcelona, so the second disjunction is true. Smith believes it, with justification, and it is true. He does not know it.

Why it matters

Before 1963 many philosophers took knowledge to be justified true belief: you know that p when p is true, you believe it, and you are justified in believing it. The definition goes back, in one form, to the end of Plato’s Theaetetus, where Socrates considers whether knowledge is true opinion with an account, and finds it wanting. Gettier’s cases show the gap precisely. Justification and truth can come apart in the middle: the belief is justified by one thing and made true by another, and the coincidence leaves the believer right by luck.

The lesson most philosophers took is that knowledge excludes a certain kind of luck. The difficulty has been to say which kind, since all of our beliefs depend in some way on things going right.

The replies

The first repairs added a fourth condition. One required that the belief not be inferred from anything false (Smith’s inference ran through the false belief that Jones would get the job); new cases were soon built in which no false step is involved. Causal and reliabilist theories required the belief to be produced by the fact it is about, or by a process that reliably produces truths. Others required the belief to be safe (it could not easily have been false) or sensitive (had it been false, the believer would not have held it). Virtue theories hold that knowledge is a true belief whose truth is creditable to the believer’s competence rather than to luck.

Each proposal handles some cases and is troubled by others, and the long run of counterexamples persuaded some philosophers, Timothy Williamson most influentially, that knowledge cannot be analysed into belief plus extra conditions at all.

Earlier cases

Gettier was not the first to describe the phenomenon. The Buddhist philosopher Dharmottara, in the eighth century, imagined a traveller who sees a mirage and takes it for water, then finds water at the spot by chance. Bertrand Russell, in 1948, described a man who looks at a clock that has stopped and happens to read the right time. What Gettier added was the target: he aimed the cases at the definition itself.

Gettier and machines

The cases have a modern descendant. A system can produce a correct answer for reasons that have nothing to do with what makes it correct, as the stopped clock does twice a day. My own work on whether language models can transmit knowledge turns on a version of this question: whether the truth of what a source says figures in why it says it (the Tracking Minimum).