Philosophy & Ethics · Knowledge & truth
When is a belief justified?
A true guess and a well-supported belief are different achievements. Evidence, reliable methods and the problem of being right by luck all bear on the difference.

What you will learn
- Tell a true belief apart from a supported one
- Explain why confidence alone doesn't supply justification
- Tell evidence-focused accounts of justification apart from reliability-focused ones
- Explain why the evidence available and the actual basis of a belief can differ
Lesson 1 of 4 · 4 min
Why a lucky guess isn't knowledge
A card lies face down on a table. Jo says it's blue, and Jo is completely sure. Someone turns the card over. It's blue. Did Jo know?
Jo had never seen the card and had no evidence about it. Mina also believed it was blue, but Mina saw it before it was turned face down. Both were right, and only one of them had a reason. Philosophers who study knowledge care about that gap between being right and having a basis for it.
A belief can be scored in three separate ways. Truth is whether it matches how things are. Justification is the support that makes it reasonable to hold. Confidence is how sure the person feels. Jo's certainty said nothing about the hidden card, and you can be confident on weak support or cautious with strong support.
Plato made the point with a courtroom. A lawyer uses trickery to win over a jury, and the verdict he argues for happens to be correct. The jurors believe something true. They don't know it, because a belief won by trickery isn't well grounded.
So the traditional definition says knowledge is justified true belief. Even that has a hole, and a clock shows it. In 1948 Bertrand Russell imagined looking at a clock you believe is running. It has stopped, but you happen to look at the one moment it shows the right time. Your belief is true, and trusting a clock is reasonable. You were still right by luck.
So ask what each person relied on. Jo relied on nothing. Mina relied on her eyes. The clock-reader relied on something that usually works and, that day, didn't.
Justification doesn't mean you've argued your case. Take your belief that five plus five is ten. You've probably never tried to justify it, and you might not know where to start. It's still knowledge. What counts is the support underneath a belief, whether or not you've ever put it into words.
The split runs the other way too. Say Mina had only glimpsed the card in dim light and taken a green card for blue. She'd have been wrong, yet her belief would still have been reasonably formed. Good support doesn't make anyone infallible.
That gives you three questions to keep apart when you judge a belief, your own included. Is it true? What supports it? And how sure does the person feel? The answer to one tells you little about the other two.
Try to recallTwo people predict the same election result, one by flipping a coin and one from careful polling. Both turn out right. Why doesn't the result show that their beliefs were equally good?
The result shows only that both beliefs were true. It says nothing about the support behind them. The coin gave no reason to expect that outcome and the polling did, so only the second belief was justified.
Jo was right with no support. The clock-reader was right with support that had quietly failed. Mina was right because of her support, and hers is the only belief of the three we'd call knowledge. The next time someone says "I knew it," ask what they were going on.
Lessons in this topic
Why a lucky guess isn't knowledge
A belief can be true and confident and still have nothing behind it. See how being right differs from having a reason to be.
What justifies a belief, your evidence or your method?
Philosophers split over whether a belief is justified by the evidence you can see or by how reliable your method is. A second distinction separates having a good reason from using it.
The stopped clock that tells the right time
A stopped clock can give you a true, reasonable belief that still isn't knowledge. See why that broke a famous definition, and why it doesn't show that nobody knows anything.
Can one extra rule rescue the definition of knowledge?
Adding a "no false steps" rule to justified true belief handles Gettier's own cases. A robot dog and a county of fake barns show why that still isn't enough.
Three quick stories from this topic
The mirage and the water
A knowledge puzzle made famous in 1963 was posed with a desert mirage around 770 CE
The Indian philosopher Dharmottara imagined people seeking water on a hot day. They see a mirage, walk to the spot and find real water under a rock. Their belief was true and reasonable, yet they were just lucky, so it wasn't knowledge.
The 1963 version
Edmund Gettier's paper used cases of the same shape to argue that a justified true belief can fall short of knowledge. They're now called Gettier cases, despite their older history.
Why the luck matters
Their evidence was the mirage, and the mirage had nothing to do with the water under the rock. Their support and the truth weren't connected.
Gettier's short paper
A two-and-a-half-page paper convinced most philosophers they had misdefined knowledge
Edmund Gettier's 1963 case was short too. Smith has strong evidence that Jones owns a Ford, so he believes "either Jones owns a Ford or Brown is in Barcelona." Jones's car is rented, but Brown happens to be in Barcelona. Smith's belief is justified and true, and it isn't knowledge.
How the belief came out true
Jones doesn't own a Ford. He's driving a rented car. By sheer coincidence, Brown really is in Barcelona. Smith's belief is justified and true, yet, as Gettier observed, Smith doesn't know it. The paper was titled "Is Justified True Belief Knowledge?"
An old problem with a new name
Cases like this had been discussed throughout the history of Eastern and Western epistemology, including a stopped-clock version from Bertrand Russell in 1948. After Gettier's influential paper, counterexamples to the justified-true-belief analysis came to be called Gettier cases.
Russell's stopped clock
Reading the right time off a stopped clock doesn't count as knowing the time
Bertrand Russell's example has you glance at a stopped clock at the one moment it's correct. Your belief is true, and trusting a clock is reasonable. But a minute later the same glance would have fooled you, so Russell said you can't correctly be said to know.
The belief is accidentally true
The same method would mislead you a minute later. The support and the truth don't connect the way they should.
A definition faces a test
The example challenges a simple justified-true-belief account of knowledge. It doesn't prove knowledge is impossible.
Sources
- The Analysis of Knowledge, Stanford Encyclopedia of Philosophy
- Bertrand Russell, Stanford Encyclopedia of Philosophy
- Gettier Problems, Internet Encyclopedia of Philosophy
- Human Knowledge: Its Scope and Limits, George Allen & Unwin





