Okay, that might be me not really getting your point. It seems like your fundamental argument is this:

Fundamentally, people disagree because most of their beliefs are supported by evidence that is at best fragmentary and at worst imaginary.

I disagree with this statement. The argument I'm making is this:

Simply put, we have evolved to focus on and pay attention to information that affirms our preconceptions, while we are evolved to tune out and ignore information that contradicts our preconceptions.

Elaborating on your argument, you dive into complexity and model-making, even throwing Heisenberg's Uncertainty Principle in there. Here's a good quote:

The difference between the freezing water proposition and the theft proposition is not one of kind, but one of complexity. Both really are just physics problems – dependent on the behavior of physical things that are subject of physical constraints.

My argument is that you are incorrect: It is not a matter of complexity that causes people to disagree. It is a matter of prejudice, and a matter of selection bias. As simply put as possible, the suitability of Sarah Palin for higher office is a pretty simple discussion. She was mayor of a tiny town in Alaska, she spent two years on Murkowski's coat tails, and she was inches away from a massive scandal when she got scooped up to tart up McCain's ticket. But if you're predisposed to see the "presidential" qualities of Sarah Palin, you're going to ignore and downplay all that in favor of intangibles like "charm" and "earthiness" because you've already decided she'd be a good vice president. Objectively it's not even a contest - consider the laughingstock that Dan Quayle was. Picking a fight with Murphy Brown! The horror! But look at Quayle's record before the VP spot. Do you think Sarah Palin would have survived if Republican selection bias didn't require Republicans to find her worthy?

The complexity of the problem is inconsequential. We're in a "shoot first, ask questions later" mindset about pretty much everything, and any argument we engage in, we've picked sides before we've opened our mouths. Your argument is this:

We know little about politics, economics, sociology, etc – in the same sense that we know the rudiments of the physical sciences… Every historical event, in time, ceases to be a collision of physical causes and becomes an author’s narrative.

Except no. Take James Carville and Mary Matalin. She ran George Bush's campaign in 1992, worked for Reagan, etc. He ran Clinton's campaign in 1992, worked for Howard Dean, etc. They've been married since 1993 and are uncontested experts in politics. Yet they disagree completely as far as political ideology.

By your argument, their far-more-perfect knowledge than the average schmo should lend itself towards a far-more-perfect agreement over the major points but it hasn't. In this case, and in all cases, understanding the situation better doesn't get you closer to the "truth" in complex problems, it makes your arguments more compelling and gives you more certainty over your correctness.