I start out with a basic subjectivist or idealist thesis: the individual mind's only interaction with any 'objective reality' is through the mediation of the senses and the perceptual faculties. In a sense, the mind uses the senses to build a virtual or Cartesian model of reality . These perceptual faculties are not perfect; they suffer from errors, falsifications, imprecision and incompleteness; the virtual model does not and cannot map perfectly onto the original reality.
Despite the fact that I cannot assert the existence of an objective reality outside the mediation of my senses and perceptual experience, I accept the existence of one on the basis of pragmatism and consistency. Despite the fact that I only percieve the existence of a coffee table in my living room through my senses, I accept that it is objectively real - it's in the same place every day, and it appears to hold my tea cup off the floor. When guests come over, they also behave as though my coffee table exists, and that it behaves in a similar way that I believe it to - it holds teacups and will cause you to stub your toe if you disbelieve in its existence.
However, I have no non-sensory way to prove that the coffee table exists. I have no way of proving that my friends who believe in the coffee table exist. At the same time, my subjective universe continues to operate in a consistent manner if I continue to believe in the coffee table.
In a similar fashion, I believe that there is bread in my kitchen. I cannot prove the objective existence of this bread. But when I put the subjective bread object in the subjective toaster object, and then eat the subjective toast object, my perceived feelings of hunger go away. The bread may be an illusion, my hunger may be an illusion, but the interaction of both serves a pragmatic purpose. I'm comfortable in saying the bread exists because it succeeds in being useful.
Having asserted the existence of an objective reality due to reasons of consistency and usefulness, it becomes difficult to speak reliably about it. Since all of my perceptions are mediated by my senses, and my senses may be incorrect or deceived, there is an element of uncertainty about all of my statements of reality. While I can assert with some confidence that my coffee table exists in some sense, my assessment of its various predicates and qualities are entirely subjective. I may assert that it is three feet long: I may be mistaken, and even if I am not, there is an inherent limit on the precision of that statement. Evin if I use some advanced technology (possibly involving lasers) to determine the length of the table, there is a chance that I will misuse the instrument, or misread it. In any case, I can only become more precise rather than perfectly precise.
Meanwhile, someone else who comes into my house to measure my coffee table might arrive at a different measurement of its length. He might say that it is 3.2 feet in length, while I maintain that it is 3.0 feet. Who is correct? It is impossible to say "the coffee table is 3.0 feet in length." At best, I may assert "I measure the coffee table as being 3.0 feet in length." My subjective reality and that of the coffee-table measuring intruder do not map consistently upon each other, and neither has a more legitimate claim to being mapped correctly onto objective reality.
The analogy that is often used here is one from philology, or textual analysis. Say a scholar has discovered pieces of a lost work of Sophocles, and wishes to attempt to reconstruct the original text. He begins with the assumption that there existed at one time a perfect text - the one that the scribe produced at the direction of and with the proof-reading of Sophocles himself [in this analogy, objective reality]. Unfortunately, the modern scholar does not have access to that manuscript. Instead, the scholar has only four different corrupted partial versions of that text [subjective models of reality]. He may attempt to reconstruct, as best he can, the original; however, he has - at best - simply created a new corrupted version of the text.
If the new version of the text turns out to be consistent, readable, and aesthetically enjoyable, though, the scholar will have succeeded in a sense. The new text, if not mapping perfectly on the original, succeeds in its purpose; it is useful. One can read it it, perhaps even perform it, and achieve some sort of aesthetic or intellectual satisfaction. Which is good enough, since we can't ask any subjective model of reality to be perfect, just useful.
Googlebombing for a cause: www.minnesotangos.org