Ever since Plato, there's been an epistemological tradition of defining knowledge as true belief plus a reason. (He called it a Logos, but that's Greek for you.) This is what reduces to our common sense notion (used in science) that knowledge is justified true belief (where the justification process is now a scientific one.)

This was all thrown into chaos, when Earnest Gettier who was an American philosopher in the early 1960's (1963 to be precise I think) pointed out that while a person may hold a belief, and it may be true, and justified, but it may not relate to the truth in the right way. In other words it is relatively accidental, or it's just lucky the person is right.

This is counter intuitive when first heard, and so he furnished a series of examples, the aforementioned Gettier Examples, to illustrate his point. Lets say I reasonably and justifiably saw the Pope receiving a bottle of whiskey and thus rightly concluded that the Pontiff was a drinking man who enjoyed the occasional bottle of Jack's Best. Now the truth is that his holiness does, but on this occasion he was in fact taking delivery of a medical specimen for his dear aunt in Poland. Now in this case my belief is true, and justified, nevertheless I don't know that he drinks whiskey, since the truth is just accidentally relative to my belief. He could be a teetotaller.

As you can imagine, this sparked a long and furious debate over the kinds of conditions that could be put in place to offer a verifiable model of knowledge, or whether this was a futile exercise, and such a model couldn't in fact be said to exist at all. This debate was used to add fire to the philosophies of relativism, nihilism, and oddly enough some aspects of phenomenology. The logical positivists had a hard time with it as well, because they couldn't create a formal system that got around the notions, but they were to have enough troubles dealing with Godel.