Update: This post relates to my conversation with Kenneth at Time Immortal. He's responded to this essay as well.
The primary job of an epistemic system obviously is to find the truth.
"Obvious" and similar phrases—especially in philosophy—are big red flags that all too often mean, "Here's where I start bullshitting you." By qualifying a point as "obvious", the writer usually avoids substantiating the point. But if the point were truly obvious, why mention it at all? "Obvious" doesn't always mark bullshit, but it does so often enough that you should automatically view any point so qualified with extreme suspicion.
An enormous part of philosophical investigations of evidentiary epistemology, the connection between our perceptions and "truth" about reality, does not at all discuss whether our perceptions are related to truth about the world, but rather on establishing the lack of deductive rigor in the process itself, and how we should think about this lack. All of these principles, Hume's critique of induction, Putnam's Twin Earth paradoxes, Gettier problems, Probabilism, etc. all talk about the rigor of evidentiary epistemology, not about whether it even begins to talk about the real truth.
The classical Greek Skeptics denied all knowledge, and no philosopher has ever successfully challenged their view. Descartes made a valiant effort, but never got past Cogito, which even the most dedicated solipsist accepts.
Even if we have a completely rigorous epistemic method, such as mathematical deduction, how do we tell if our valid conclusions are actually true? If I put two rocks in a bowl, and then two more, I "know" I'll count four rocks in the bowl. Do I know this because 2+2=4 is a valid theorem of arithmetic? Or do I know that integer arithmetic is a good model of my experience about rocks by some other means? If I put two drops of water in a bowl, and then two more, I "know" I won't count four drops of water.
I don't want to put down philosophy that investigates the rigor of epistemic systems. But I want to make crystal clear that this sort of philosophy asks the question: Assuming our epistemology is doing the job of establishing truth, how well does it do that job? But we're simply assuming that our epistemology does the job we're asking of it.
When we ask whether (rather than how rigorously) our epistemology is finding the truth, we run into a vicious circularity. To determine whether our epistemology is finding the truth, we must know the truth independently of that system. But by definition we need some sort of epistemic system to know the truth.
Theism doesn't really help. Simply by defining "truth" in some God-dependent manner doesn't free us from assuming that our epistemic systems deliver the truth, it just dresses up the assumptions in rococo finery hoping to dazzle us with shininess. At worst, theism just places the assumption in the realm of religious dogma, guarded from doubt and philosophical investigation by threat of Hell.
As Plantinga's evolutionary critique of naturalism shows, if we define "naturalism" in the broadest sense as an epistemology dependent on human reason and perception, and if we define "truth" to be something supernatural in this sense, then we need a supernatural entity to relate supernatural truth to natural knowledge. It's a valid argument but obviously begs the question.
It's clear that we cannot ask whether an epistemology delivers the truth. Rather, we are forced to define "truth" in terms of what an epistemology delivers (a subtle and complicated enough task when our epistemology is not absolutely rigorous).
Much of what passes for "postmodernist" philosophy recognizes this fundamental inversion of the "modernist" relationship between epistemology and truth. If "truth" is whatever an epistemology delivers, then anything can serve as an epistemology; there is no basis in truth for privileging one epistemology over another. But no basis in truth does not—as many postmodernist philosophers assert—mean no basis at all. Even if it makes no sense to ask whether an epistemology delivers the truth, there are other jobs we can ask our epistemology to do, other ways of separating epistemology from bullshit.
 Plantinga fundamentally argues against metaphysical naturalism. He still assumes that epistemological naturalism does deliver truth.
 I'm convinced that 99% (if not 99% or possibly even 100%) of skill in philosophy consists not of constructing sound arguments but of hiding the fundamental fallacy of begging the question by various means, notably equivocation, enthymemes, hand-waving, and bold assertion.