Here, I think you're introducing an idea, or notion, that isn't necessary to the conversation; namely, the notion of luck. — fiveredapples
Hmm, I do not think you're right, but it doesn't really affect my point, which I'll elaborate on shortly.
First, you say that in the original clock example the agent does not really have a justification because broken clocks are not reliable time-tellers and the agent is looking at a broken clock.
Several things: first, intuitively the agent 'is' justified. They could not reasonably have been expected to know that the clock was not working. So, they were justified in believing it was working, and so subsequently justified in believing it was the time that it represented it to be.
Second, for the sake of argument let's test your analysis. If it is correct, then any belief about the time based on a broken clock's report should fail to qualify as knowledge. But I can imagine a case in which a person bases a belief about the time on a broken clock and their belief 'does' qualify as knowledge.
For example, imagine the clock has broken, but it has literally just broken - that is, it has broken at the point at which its hands reach 3 o clock. It was working fine up to that point. The agent then looks at the clock. Now, the agent is looking at a broken clock and, on the basis of its report, he forms the belief that it is 3 o clock (which it is). This time it seems clear enough that the agent does have knowledge, yes?
Yet their belief is based on the report of a broken clock.
But anyway, I am not married to the 'luck' analysis either, for my point is not that this or that analysis is always and everywhere correct, but rather that whatever diagnosis we give of why the agent lacks knowledge in the relevant case, we will be able to construct another case in which that 'key' ingredient is present and the agent lacks knowledge (or absent, and the agent possesses knowledge)
That's not to say that the diagnosis of the original case was wrong. It is just to note that there is no stability to what is, and is not doing the work of making it the case that the agent has knowledge (apart from possessing a true belief). So it is to say that the diagnosis does not locate an ingredient of knowledge, even when it is correct - that is, even when it correctly diagnoses why the agent lacked knowledge on this particular occasion.
So, take 'justification'. I understand that term to mean 'has normative reason to believe'. Now it seems to me that in many cases an agent knows something due to the fact they have a justification for their true belief.
But there also seem cases in which an agent has a justification for their true belief and lacks knowledge.
And there also seem to be cases in which an agent knows something yet lacks a justification (in an example I gave earlier, I might acquire a true belief, but the belief is so trivial there is no normative reason for me to believe it - yet intuitively I may still have knowledge in such a case).
So the lesson I take from the many hundreds, if not thousands of failed attempts that have been made to specify what ingredients knowledge is made from, is that there is no stable set of ingredients beyond 'true belief' (though obviously 'true beliefs' often fail to qualify as knowledge too).
In turn that tells us something important about knowledge. It isn't made of those ingredients. Rather, it is something that those ingredients typically bring about.
For an analogy: take the property of being 'delicious-to-Bartricks'. Now, there are plenty of things I find delicious and they often have things in common - such as containing chocolate, or lots of sugar, or whatever. But it would be a mistake to think that because I often find something delicious due to it containing chocolate, that therefore anything that contains chocolate I will find delicious. No, in fact sometimes I might dislike something due to it containing chocolate (a potato stuffed with chocolate - no, that's not delicious at all).
What conclusion would it be reasonable to draw from that? Well, that 'delicious-to-Bartricks' is not something made of ingredients, but is rather an attitude that certain combinations of ingredients, in certain circumstances, produce in Bartricks. The whole project of trying to figure out what ingredients delicious-to-Bartricks is made of is misguided.
I am drawing the same lesson in respect of knowledge. Knowledge has no stable ingredients beyond true belief (just as 'delicious-to-Bartricks' has no stable ingredients beyond edibility). Thus it is reasonable to conclude that 'knowledge' is an attitude that a person is adopting towards true beliefs. Not, I emphasis, an attitude one of us is adopting towards true beliefs, but an attitude Reason is adopting towards them.
What attitude? Well, the knowledge attitude. The attitude we are referring to when we 'feel' that we know something. Only it has to be felt by Reason, not us.