To recap, here's the basic faith-including epistemological system:
Use non-faith epistemological methods where they work. When science, philosophy, and mathematics get results, those results are accepted. However, where these methods fail, we anticipate a revalation. This revalation comes from some higher power. The results of these revelations are believed because we have faith in the higher power; faith in its ability and in its benevelence towards us. These revelations are not later supported by the other epistemologies because such support would remove the system's reliance on faith. In order to retain faith as a part of the system, there must be some revelations that go unsupported by the other methodologies.
The first problem facing this system is one of inter-epistemology conflict resolution. In the event that multiple involved methods (science, philosophy, mathematics, revalation, intuition if you care to include it) provide conflicting answers, which answer will we embrace?
This first problem isn't unique to faith-including systems. It is a problem that must be addressed by any system that involves potentially-conflicting epistemologies. One potential answer is to develop a hierarchy, so that one epistemology trumps another which trumps another and so-on down the line. Of course, the particular ordering of this higherarchy will significantly alter the results of the system when analyzing such multi-answered questions.
The second answer is to deny the potential conflict. This type of response is expressed in things like NOMA, which assert that science and faith answer different questions, and thus there can be no conflict. Of corse, to really apply such a response to the method considered here, we would also have to assert that science and math answer different questions (try telling that to a physicist), as do faith and math, science and philosophy, and so on.
We can see clear cases where the second answer is insufficient, such as the Indiana Pi Bill, the intelligent design movement, and any Newtonian-based metaphysics. There are some cases where apparent conflict can be resolved without establishing a hierarchy (such as the apparent conflict between Euclidean geometry and the fact that a large enough triangle on the earth's surface does not have an angle sum of 180 degrees, which is resolved by concluding that the earth's surface is not a Euclidean plane), but there are many instances where such conflict needs to be resolved by determining a hierarchy.
So a hierarchy needs to be developed, and its structure will determine the conclusions our system reaches for the set of multi-answered questions. This hierarchy is ultimately what separates the religious fundamentalists from the religious moderates. For fundamentalists, faith tends to be at the top (and perhaps has so wide a scope as to make the others pointless). For moderates, faith tends to be at or very near the bottom, at least below science and mathematics and maybe only above intuition. (Of course, the hierarchy can change from question to question, but such a practice will reqire us to devise a system by which we can establish the hierarchy for any given question). I have already spent some time (and will eventually spend some more) pointing out how faith-based epistemologies lack many desireable epistemological qualities. This demonstrates that if faith is to be included anywhere at all, it ought to be low on the higherarchy, so that our system can maintain as many of these desireable qualities as possible.
This is the minimalist approach, a system in which faith is a "last resort," used only when the other methods (except perhaps intution) fail. It is the "God of the gaps" approach, where God is inserted to answer questions where science and math and philosophy fail. And only those questions. This is where faith has to be in order to minimize the ammount of convergence, predictive power, and self-analysis we lose by adding faith. In other words, in order to make a faith-including epistemology at least as strong as a faith-less epistemology (assuming such a task is possible), we must place faith at the bottom of the hierarchy.
What this results in is essentially the skeptic's epistemology with "when that fails, use faith" tacked on. At first blush, this may look like an improvement. After all, it appears to allow the epistemology to address questions that previously went unaddressed. It "fills the gaps," so to speak. So why do I reject such an add-on? There are several reasons, but one is surprisingly simple. The gaps keep closing.
There was once a time when lightning wasn't understood scientifically, and religion was used to provide an "explanation." Since then, science has answered the questions and driven religion away from this particular phenomenon. The gap in our understanding closed. As far as I know, this closure was relatively clean and fast. But if we look at evolution, we see a closure that is still being resisted, and it is being resisted precisely because it was once filled with a conflicting faith-based conclusion.
Religions almost always present their teachings as more than a guess. And these teachings are almost always held to be highly important. This makes it more difficult to relinquish that territory when the gap closes. It took over a century for the Catholic church to stop rejecting the heliocentric model (while still not embracing it, that took nearly two centuries). I am perfectly fine with taking a guess and reporting it as a guess. The problem with inserting faith into these gaps is that the faith isn't treated as a guess. And since faith does a poor job of self-analysis, this poses an obstacle when science (or math or philosophy or whatever) reaches a conflicting conclusion. It hinders progress.
Imagine how different things would be if, for centuries, people had asserted that they just didn't know how mankind got here. They might say that if they had to guess, they'd go with divine intervention, but that this is just a guess. This would remove an enormous stumbling block in the path of widespread acceptance of the theory of evolution, which would allow biologists to stop combatting ID and get on with biology. Similar practices would have removed enormous stumbling blocks in the path of widespread acceptance of a spherical earth, the heliocentric model, the negative conclusions of scientific research into the paranormal, and so on.
Even if we take great pains to minimize the impact of faith on our epistemology, we are still left with a sharp decrease in the system's ability to self-correct. Taking what the skeptic treats as a guess and presenting it as divinely inspired truth hinders the progress of convergent, self-analizing, prediction-giving epistemological methods. It leads to overly-strong positions on matters that would otherwise be recognized as unknowns, which in turn generates conflict while simultaneously hindering convergent attempts to resolve such conflict.
Adding faith, even at the bottom of an epistemological system, hinders that system's development. It produces overconfidence in conclusions that must eventually be overturned. It generates conflict in what would otherwise be seen as an "unknown" area. And while it purports to answer questions left un-addressed by the skeptic's epistemology, the non-convergent nature of faith makes these answers as disparate and not-conflict-resolving as the mere guesswork employed by the skeptic. Its contribution to the epistemological endeavor is at best non-positive and at worst remarkably negative. Thus, it is removed.