The Gettier problem
Gettier case · justified true belief
A three-page paper from 1963 showing that a belief can be true, and justified, and still obviously not knowledge — because the justification and the truth can come apart by luck. It ended a definition that had stood since Plato.
In practice
A forecast that was right for a reason that turned out to be false is the same structure. The number was correct, the reasoning was sound on the evidence available, and treating the outcome as a vindication of the method is the error the case isolates.
The common mistake
Reading it as skepticism. Gettier does not argue that knowledge is unattainable; he argues that a particular analysis of it admits cases everyone recognizes as failures, which is a claim about the definition rather than about us.
You glance at the clock in the hall. It reads 3:15. You believe it is 3:15, and you are right — it is.
The clock stopped exactly twelve hours ago.
Your belief is true. It is justified, since reading a normally reliable clock is as good a reason as anyone has for a belief about the time, and you had no indication of anything wrong. And it is, plainly, not knowledge. You were right by accident.
That is a Gettier case. Edmund Gettier's Is Justified True Belief Knowledge? (1963) runs to three pages, contains two examples and no theory, and dismantled an account of knowledge that had been assumed since the Theaetetus.
What the target was
The standard analysis held that a person knows a proposition when three conditions are met, jointly sufficient as well as necessary: the proposition is true, the person believes it, and the belief is justified.
Each condition earns its place. Drop truth and you can know falsehoods. Drop belief and you can know things you do not think. Drop justification and a lucky guess counts.
Gettier's move was to accept all three and construct a case where they are satisfied and knowledge is absent. Take a justified false belief, derive from it something that happens to be true for an unrelated reason, and you have one. The stopped clock is the cleanest version; Gettier's own cases involve a man inferring that whoever gets a job has ten coins in his pocket, and getting it right about the wrong man.
What broke, precisely
The three conditions are each about a different thing, and nothing in the analysis requires them to be about the same thing together.
Justification concerns your reasons. Truth concerns the world. In a good case these are connected — your reason is good because it tracks how things are. In a Gettier case that connection is severed while both conditions independently survive: the clock is a good reason, the time is 3:15, and the two facts have nothing to do with each other.
So the diagnosis is not that one condition was too weak. It is that the analysis was a conjunction when it needed to describe a relation.
Twenty years of patches
The obvious response was a fourth condition, and the search for one became a minor industry.
No false lemmas — the belief must not rest on anything false. This handles Gettier's own cases neatly and fails on the clock, where nothing false is believed at any point.
Defeasibility — no further truth, if learned, should undercut the justification. The clock is stopped is such a truth, so this works here; the difficulty is stating it without also ruling out ordinary knowledge, since for almost any belief there is some misleading truth that would shake it.
Reliabilism — the belief must be produced by a process that generally yields truths. A stopped clock is not such a process. This is the most durable of the repairs and moves the subject decisively away from reasons a knower could cite.
Each handles some cases, each admits new ones, and new counterexamples arrived faster than conditions. Linda Zagzebski showed in 1994 why: as long as justification is compatible with falsity at all, you can always manufacture a case where a false-but-justified belief is rescued by luck. The recipe is general, so no patch of that kind can work.
The lesson that outlasted the problem
The most interesting response is that the project itself was misconceived.
Timothy Williamson argued in Knowledge and Its Limits (2000) that knowledge is not analyzable into more basic components at all. It is a primitive mental state, and the right direction of explanation runs the other way — you explain justified belief in terms of knowledge, not knowledge in terms of belief.
Whether or not that is right, Gettier's real achievement is visible now in a way it was not then. What the cases isolate is luck, and what fifty years of failed repairs suggest is that excluding luck is not one more condition to be added to a list. Every attempt to say precisely what makes a true belief non-accidental has either been too strong, ruling out ordinary knowledge, or too weak, letting a new case through.
Three pages, no theory, and the definition has not been put back together.
Concept web
Open the full webQuestions
What is the Gettier problem?
The demonstration that a belief can be true, believed and justified without being knowledge. Gettier's 1963 paper gives cases where the justification and the truth hold for unrelated reasons, so the believer is right only by luck.
What is the stopped clock example?
You read a stopped clock that happens to show the correct time. Your belief about the time is true, and reading a normally reliable clock justifies it, but you are right by accident, so it is not knowledge.
Has the Gettier problem been solved?
No repair has held. Proposed fourth conditions such as no false lemmas, defeasibility and reliabilism each handle some cases and admit others. Zagzebski showed the counterexamples can always be regenerated, and some philosophers conclude knowledge is not analyzable at all.