http://plato.stanford.edu/entries/brain-vat/
The event level, according to Korzybski, (and I'm not going to bother to look it up.) represents all our highest level abstractions as to what is going on as well as any potential sources of abstraction.
Metaphysically, it could be the evil genious as the skeptics might suggest; it could be something independent of all observerser as the realists suggest.
Epistemologically, we simply do not know in any strong sense; so it can be taken ambigously to represent any process from which we may form abstractions.
Recall that metaphysics and epistemology together are like the wave particle duality. We can put our finger on particles, but not on waves. Metaphysics requires an assumption as to what "is"; whereas epistemology requires an assumption as to how we know. Internal, direct knowledge is immediat, and we can put our mental finger on it, but once it is projected by the mechanism we call "reference", we no longer have direct access - we can't put our mental finger on it.
Korzybski's (not original) answer is that we know through the biological process of our nervous system in the context of our environments, so study that, and we will know how we know what we know.
What is the status of "what we know"? To the best of our current understanding it's at best a hypothetical model to account for our "mental" experiences - including the ones projected on our bodies and externally.
If we "throw out" the "event level", we don't have a place onto which to project our theories.
Moreover, we would make general semantics totally untenable to anyone with even a smidgen of "realism" in their view.
So, let's keep it.
Ben, Characteristics do not get "taken in". The nervous system "responds" with a characteristic that is different from what it may have been stimulated by. A "characteristic" at the event level "is not", "is not the same as", "does not match", etc., a characteristic at the object level. One is "theoretically" some property of some putative "thing" (presumed structure) while the other is a neurological discharge or a synaptic chemical reaction, etc.; at each level the "characteristic" is different from any other level. We can draw a thread through characteristics, like stringing beads that are all different, with one "bead" indicating what happens to be the characteristic at that level.
In a few cases, the characteristics at some levels may be of the same type - nerolological responses - at object levels - words at verbal levels, but they can differ even within these levels.
When we take the tags going back to the event level, it just, it seems to me, means that we are projecting our abstract structure onto what is going on, and consequently we get future abstractions that agree with these projections.
At that level the parabola holds both the territory and the abstract map of it.
What we have no direct access to is the territory.
What we describe, perceive, cognize, etc., about what is going on holds the highest level abstractions.
We talk about what is going on as if the things we say are out there. When we mention electrons, protrons, light, livers, eyes, nerological responses, etc., we are talking as if these things exist independently of us, but we are also projecting our psycho-social model, and since that model is shared by cultural learning and time-binding, we have the presumption that we each understand each other.
General semantics proves we don't. 
Thanks, Ben,
I've noted over the years that some novice general semanticsts mistake or assume that a characteristic somehow is a constant from level to level; that it does not change. I wanted to be clear that that is not the case. "Takes in" seems to suggest this constancy, as perhaps "registers" might, depending on context, but "ascribes" and "responds" do not appear to do so.
People routinely abstract different sources into their neurological representations and then compare them, thinking that they are comparing the sources. In general semantics terms, this is identification of the responses with that from which it is abstracted. We experience and compare our responses, and those are maps distinct from the presumed territories.
But it's so "real" and so "common" and so "natural" and so much a part of our "common sense" world view, that we may have difficulty assimilating the "concept by postulation" that we are not experiencing or comparing the actual territory.
Thomas wrote "I think the SD is designed to show the progression of science and not how individuals communicate and evaluate."
It shows a common abstraction from the process by which humans extract information from their environment, abstract and process it and represent it in words as well as how science abstracts information from the collective environment, abstracts and process it and represents it in words - a model of the environment.
"Science" - the collective body of knowledge grows and evolves by the process of abstracting, building models, and testing and revising them through use. Individual human beings do "the same thing"; a persons body of knowledge grows and evolves by the process of abstracting, building models, and testing and revising them though use.
The structural differential represents the common structucture abstracted from both processes.
Ben, when you compare anything to anything else, the process takes place in your brain after abstraction. You are not comparing the territories; you are comparing your abstract maps - internal to your brain - after abstraction has converted different sources into a common internal representation.