Going back and forth for three weeks about what epistemology is
better, I think it's time to start afresh.

First, let's define what someone considers to be important.

1) That which is within our ability for objective demonstration.

2) That which is outside our ability for objective demonstration.

These are two separate and disjoint questions. The way to obtain
objective information about stance 1 would be via repeated
observations. Everyone on earth ascribes to this at some point in
their lives. They need to know what food is poisonous, how high they
can jump before they die when they fall to the ground, how much gas to
put in their cars, what situations they should avoid when they carry
radioactive materials, etc, etc, etc. This is what I will consider the
"practical" stance.

The way to obtain objective information about stance 2 is to interpret
subjective inference, and assume axiomatically that these subjective
inferences accurately reflect God's will of what we should do. Not
everyone on earth ascribes to this philosophy, and not everyone on
earth ascribes to this philosophy in the same way. This is what I will
call the "esoteric" stance.

The atheist/agnostic/rationalist entirely ascribes to the "practical
stance". They will simply reason that what doesn't impact them via (1)
is irrelevant to their existence. That is, what is unfalsifiable is
not worth worrying about. They worry about God's existence and God's
desires for them as much as they worry about the Flying Spaghetti
Monster's existence and the Flying Spaghetti Monster's desires for
them (i.e not at all). This is a perfectly reasonable stance to take.
One's experiences defines who one actually is, and if one's
experiences do not point to the supernatural, one can hardly blame
them for not acting as if the supernatural exists.

The theist ascribes to the "esoteric" stance (unless they're driving
their car or jumping out a window, in which case they'll ascribe to
the "pragmatic" stance as well). They will axiomatically assume that
they know the will of God in some way, shape or form. They think there
is a definitive difference between the Flying Spaghetti Monster and
God. Their experiences convince them of this. As an a priori stance,
this is perfectly reasonable (after all, it's just an interpretation
of subjective events). However one must of course realize that what is
subjectively interpreted by oneself alone does not define an objective
truth when the objective truth in question is unfalsifiable, and hence
unknowable. Of course, when the statement is falsifiable the ambiguity
is resolved by simply testing the hypothesis (so statements like, "The
earth was created in 6 days" are falsifiable and perfectly open to
those appealing to stance 1).

The MAJOR discrepancy comes when there is a statement that is
falsifiable, and those in stance 2 simply ignore stance 1 in
determining the truth in question. This is inherently naive. If there
is a direct conflict between a dogmatic assertion of a falsifiable
prediction, and what the observation of that falsifiable event
actually IS, there is no reasonable way to justify the accepting of
that which is dogmatically asserted above what is observed. There is
simply no reasonable way to maintain this stance.

For instance, dogmatic statements from religion that are falsifiable
(and falsified) would be:
-The earth is 6000 years old and was created in 6 days.
-Human beings descended from two people within the last few thousand
years, who descended from mudpies plus divine intervention.
-There was a global flood.
-The earth stopped spinning on it's axis.
-The earth is flat.
-The earth is at the center of the universe.

These are all demonstrated to be false via observation. Asserting them
would amount to believing in a falsehood (i.e. the definition of being
"delusional"... I realize that is a loaded word, but that's the word
in the English language we have to describe believing in a
falsehood).

Dogmatic assertions from religion that are falsifiable but
insufficient evidence is currently present to decide their veracity
would be:
-There was a person who rose from the dead 2000 years ago.
-That same person made people who were blind see again by rubbing mud
in their eyes.
-That same person turned water into wine and multiplied food.

These aren't technically demonstrated to be false, per se, however via
our observations they are extraordinarily unlikely and it is perfectly
reasonable to not accept the claims of their veracity because they are
so extraordinarily unlikely. Should someone accept them on the off-
chance that they are true, they aren't technically falsified, so there
is no a priori case against believing in them. Of course, this is
simply a matter of suspending disbelief in their plausibility (theists
call this the "leap of faith", but I like "suspending disbelief" much
better, personally). I would consider the "default" stance here one
where the disbelief is not suspended, however (i.e. the leap of faith
is not demonstrated to be necessary at all, but if someone wants to
make that leap, more power to you... just keep the falsifiable
predictions in the realm of observational acquisitions of knowledge
and we'll be basically all set).

Dogmatic assertions from religion that are unfalsifiable would be:
-God exists.
-God created the universe using known physical processes (i.e. "God
works in natural ways").
-God talks to us each individually via some unknown process in our
brains to implant information there somehow.
-There is a heaven.
-God is moral.
-Obeying God's will is a way to enter heaven.
-We have a soul.
-We continue to be conscious in some way after we die.

These aren't demonstrated to be false at all, but neither are they
demonstrated to be true. They are, of course, all positive assertions,
so in order to get someone else to believe them the person in question
must supply some justification, or admit that such justifications are
impossible. Of course, in that case, such a person would have no
choice but to admit the reasonableness of the stance that doesn't
accept their positions). The suspension of disbelief is also present
here (aka the "leap of faith"), since there is no evidence for or
against such things at all. Of course, this applies to any
unfalsifiable statement like "There is a pot of gold at the end of a
rainbow" or "There is an invisible pink unicorn on my shoulder that
tells me what is moral and what is not", so selectively ascribing to
one of these statements has no more objective justification than
ascribing to any of the others. One could consider the fact that
theists are simply atheists who disbelieve one fewer unfalsifiable
assertion about the supernatural.

In summary, the different questions that are answerable with our
different epistemologies are simply a matter of what questions you're
asking. The atheist/agnostic/rationalist/nontheist/whatever will apply
the same principles to the question of God's existence than what they
would when they drive a car. The theist, on the other hand, will apply
different principles to the question of God's existence than what they
would when they drive a car, and selectively suspend disbelief (i.e.
make the "leap of faith") toward things they have a subjective "hunch"
might be true, and don't suspend disbelief toward things they DON'T
have a subjective "hunch" might be true.

When there is a conflict in a falsifiable position, then the only
reasonable stance is to accept that which is observed (stance 1).
Everything else is denial of reality based on no justification. I will
also maintain that positive assertions like "God exists" or "God
doesn't exist" require evidence for someone else to accept them. The
denial of a positive assertion like "God exists" or "God doesn't
exist" is not necessary... the burden of proof lies with the one
making the positive assertion. I will also maintain that someone not
acknowledging their uncertainty about unfalsifiable assertions is
simply mistaken. One can axiomatically assume they're right all they
want, however the ultimate answer is that the person is not certain
that unfalsifiable axioms are true, because they are unfalsifiable.
They are, of course, free to make axioms to formulate a theory,
however acting as if those axioms are certain is simply not supported
by the truth value of what an axiom is (i.e. "undefined" rather than
"true" or "false").