I'll answer your hypothetical, but then I would appreciate if you answered one of mine.
> I am the All-Knower-and-All-Seer.
My prior that this is true is extraordinarily low. I put this at a 0.000000001% chance of likely to be true. Suppose you provide me a theory that Iraq possess Weapons of Mass Destruction, and must be invaded in order to prevent the destruction of a major city.
I would assign an extremely-high negative consequence to that outcome. I would assign a low-ish probability to the outcome based on my inherit disbelief that you are Always-Right-All-Knower-and-All-Seer, and that you have not furnished any evidence.
If your proposed solution is to invade the country, I would assign a very-high negative consequence to the outcome. As such, I would ask you to provide quite compelling evidence that your claim was true, before supporting such a rash action.
If, instead, your proposed solution was to propose international monitoring to the country of Iraq I would assign a quite-low negative consequence to the outcome (probably even a slightly positive consequence). As such, before backing your proposed solution I would require a much lower bar of evidence.
But I believe you about the WMDs in either scenario to the same degree. Your proposal changes how much I need to accept your premise about the WMDs to accept your solution.
So, I think accepting decision D based on belief X is a far far more complicated question that "do you believe X"?
----
Let me also look at this from the other side. Let's say I, for some reason, conclusively believe you to be an All-Knower-and-All-Seer. My belief in this is 100%. Perhaps I am easily convinced, or perhaps you made 1,000,000,000,000 predictions of fully stochastic events, and I have assigned a 10-sigma confidence that you are some kind of All-Knower-and-All-Seer. Let's also assume that I have 100% confidence that you are honest, and that I have some way of gaining that confidence. However I got there, it's somewhat irrelevant at this point.
If I have 100% confidence that you are truly an All-Knower-and-All-Seer, and you suggested a particular course of action was strictly beneficial to me, I would almost certainly follow that course of action without requiring further evidence from you.
I would argue that, in such a scenario (and such a scenario is quite contrived, I believe, for anyone who is truly skeptical) it would be irrational for me to ask you for more evidence before acting. After all, at that point we've already moved by belief threshold high enough for me to take actions in the world to further my goals.
This all assumes that we are an actor in the world with goals (which, I again assert that goals cannot be defined rationally, but rather that rationality serves as a tool to achieve goals). If you have no goals, then I agree that pure reason would require you to ask for more evidence.
----
My return hypothetical is to ask you this question:
You are a key decision maker for a scientific body. You are deciding whether you should run an experiment code-named Alpha. There is a low-likelihood but not preposterous theory (say, accepted by 10-15% of the scientific community), but lacks any solid experimental support. This theory says that running the Alpha will destroy the planet. The gain from running the experiment will be furthering scientific knowledge in quite useful, but not ground-breaking ways.
Again, this "Destruction Hypothesis" seems like it's not super likely to be correct, but it's not a crackpot-theory.
Do you run Alpha today without delay, or do you order more investigation of the Destruction Hypothesis to conclusively rule it out before running Alpha?