The Limits of Replicability in Epistemology

Replicability can't be the ultimate criterion of knowledge for the simple reason that if it were, you couldn't know anything. For in order to replicate a certain type of event, you need to be able to know know that token instances of events of that type have occurred in the past; so that each time you replicate the event type, you can say that the new token is in fact of the same type as said past events. 

The issue is that each token event qua token event is per se irreplicable, for by nature the boundaries of the identity of events boundaries are found in time, so that an event can be marked by it's beginning and end, so that any subsequent event will by definition be a distinct event (not the same event) and so each subsequent event will failed to replicate that exact same event, but will only at best have certain similarities to said event, so that it can be classified as the same 'type' of event; so that it will remain that the tokens qua token events are never replicated. 

However, if replicability is required for knowledge, then since token events cannot in principle be replicated, then we can never know of any past token event, nor even of any present token events, so as to say a present token event is of the same type as past token event. Consequently, in such a case, we could never know of any given case of replication, so that we could not know anything at all. 

But clearly we do know things; for 'knowledge' is our word, and whatever else we mean by it, we don't mean the sort of thing someone who can 'mean' things (like ourselves) could be completely lacking of; and so it must be that any conceptual analysis of 'knowledge' that has replication being a minimum condition of knowledge is, by that fact, false. 

This is not to say that replicability is not epistemically valuable, it surely is; but it is to say that it is not reasonable to root one's entire view of knowledge within replicability.