The Best-Guess Phenomenon and Degree of Belief
You're a lost tourist. You hit up a local for directions. With apparent confidence and fluency, the local sends you in utterly the wrong direction. I experience this so regularly in travelling that I now no longer ask a single local for directions. About half the time, the second gives a completely different version from the first. Occasionally, someone will express a degree of uncertainty. Almost always, if she's at all uncertain, she's completely wrong; and many who seem perfectly confident are equally wrong.
I've been on the other end of this, too, at least twice, realizing that my directions were seriously mistaken only after having given them. In one case, I sent a Mexican family so far astray that I fear it would take them at least an hour recover, were they unwise enough to trust me!
Similarly, but more serious: The doctor tells you you have disease X, with evident assurance, no visible uncertainty. Another tells you you have disease Y. Or even: The doctor starts out seemingly uncertain, undecided, then settles on something as -- you might think, her best guess -- then, as she spells out the guess to you becomes seemingly confident about it, confident enough, apparently, to stake your health on it. But she's wrong, and in fact changes her opinion easily when the next doctor you see calls up the first and describes his different, better diagnosis.
Let's call this the "best guess phenomenon". In certain situations, when Person A is presumably an expert and Person B has no resources to challenge Person A's opinion, Person A will give her best guess, conveying it with authority and confidence regardless of how well-founded the opinion is. No malice is intended, nor any disguise. It's not that Person A knows she's uncertain and aims to conceal that fact. Rather, the situation invites Person A to take on the mantle of expertise, with very little sensitivity to the proper degree of confidence.
One model I think won't suffice for such cases: Conventional philosophical/economic treatments in terms of "degree of belief" on a scale from 0 to 1. Best-guess phenomena are not, I think, best described as cases in which Person A has an irrationally high degree of confidence. For example, if asked to make a serious wager -- for example, if the local wanted to get there herself, or if the doctor's own health were at stake -- she'd balk, admit uncertainty, consult elsewhere. Rather, it's more like degree of confidence doesn't arise as an issue: Person A is neither certain nor uncertain, really. She's just talking, playing authority as part of a social role, without much thought about how much certainty is justified.