I find the problem of the criterion very interesting. My take on the problem is that we have a concept of "knowledge" which most people seem to understand. The question is how do we come to understand what we mean when we use the term "knowledge". Can we understand the meaning without resorting to instances?
When we develop language I think it is fair to say that basic concepts are "taught" by pointing to examples and saying "that is an example of x". The subject then begins to recognise what is common between all of the different instances and thereby how to use the term in meaningful language.
If we accept that there is a definition of knowledge which comprises several other concepts such as belief, truth and justification, then is it the case that our understanding of these concepts must precede our understanding of knowledge?
Could there in fact be a concept of knowledge which exists independently of any cases of knowledge or independently of language?
It does seem odd that we have a gut feeling that the term "knowledge" is something more than just a label.
I realise I am asking a lot more questions than I am answering!