The word knowledge is used with two different meanings, one weak and one strong. In a weak sense knowledge means basically the same thing as “justified belief” (or better yet: beliefs we should have given our epistemic position), which means that the things we know may very well turn out to be false. In a strong sense knowledge refers only to warranted true beliefs, where warrant is simply a label for whatever is needed to be added to a true belief to make it knowledge (note: warrant is not the same as justification, it can be shown that warrant is necessarily stronger than justification). The strong sense of knowledge is felt by many to be the more intuitive understanding. But already there seems to be something absurd about the strong definition. Specifically from our point of view knowledge, so defined, seems simply to be a label to hang on whatever we believe to be true. Since we are rational people believing something to be true is basically equivalent to believing that we know it. And moreover it becomes hard to see how the concept of strong knowledge is a valuable tool in addition to our concept of truth; since strong knowledge requires the belief to be true a procedure to separate true beliefs from false beliefs must be a necessary part of any procedure to determine when we have knowledge, since only true beliefs can be knowledge. But with such a procedure we no longer have any need for knowledge, since we already have truth. And if we can’t devise a procedure to sort the true beliefs from false beliefs then we can’t devise a procedure to determine when we have knowledge. So either way strong knowledge seems superfluous.
But the absurdity of strong knowledge doesn’t stop there. Recently Michael Blome-Tillmann has written a paper entitled “The folly of trying to define knowledge” where he expands on Trenton Merricks’ work to show a surprising fact about warrant. Merricks’ argument shows that warrant, whatever it is, entails truth. Consider: if a belief can be warranted and false then it can be warranted and accidentally true (to be accidentally true is when some fact unrelated to the warrant makes the belief true). But a belief cannot be warranted and accidentally true, because that would make that belief knowledge, by definition, and there are several thought experiments which imply that strong knowledge requires that the truth of the belief be connected to its warrant. And thus a belief cannot be warranted and false. And so by this argument if a belief is warranted then it is true. And Blome-Tillmann has expanded this argument to show that this also holds for belief, because if it is possible for someone to be warranted to believe some fact and not actually believe it then it is possible for them to be warranted to believe it and believe it accidentally, which makes it an accidentally true belief. And so by the same reasoning warrant must entail belief. Blome-Tillmann thus concludes that if warrant implies a true belief then warrant implies knowledge. And obviously knowledge implies warrant. So knowledge is not distinct from warrant, rather warrant is simply another name for knowledge.
Now obviously if warrant implies truth and belief it must be fairly strong itself. One candidate for warrant is justification from true premises. Another is belief for reasons that are actually reasons for the belief being true. Warrant then is probably as strong as knowledge in its strongest sense. Which means that the earlier problem I posed for strong knowledge, namely its being a basically redundant label for truth, can’t be solved by an appeal to warrant. By an appeal to warrant I refer to the idea that maybe a study of knowledge (specifically a study of the warrant component of knowledge) can improve our epistemic situation in a way that a study of truth cannot. If warrant didn’t imply truth we might reason that while a procedure for determining when we had warrant wasn’t as good as one for determining when we had truth, but perhaps it could give us some clue as to which of our beliefs are more likely to be true. But, given that warrant entails truth, a procedure for determining truth is necessarily simpler than one for determining when we have warrant (or at least not more complex).
Another possibility for saving strong knowledge might be to throw in justification (or something similar) as an additional component of knowledge. But it is hard to defend this move, since given the strength of warrant justification is redundant. Since having warrant necessarily entails truth then having a warrant is justification, and so adding justification to this is pointless. A better move might be to break warrant itself into components, and then look for a way to determine when we have some of those components, making the possibility that we have warrant more certain. But there are problems with this move as well. Since warrant entails truth together the components of warrant must entail truth. Perhaps it is just my lack of imagination, but I can’t see any way to construct warrant that doesn’t slip in truth in some form. For example, if warrant is justification from true premises then obviously knowing when we have true premises requires us to be able to determine what is true. In general to guarantee truth about claims in a domain we need to start with some true claims in that domain (for example, in mathematics we start with axioms). Of course we can still study whatever else makes up warrant. But then aren’t we just studying weak knowledge? Why pretend that we are trying to understand strong knowledge when it is only weak knowledge that we are really after a better understanding of?