I've been trying to say something similar for a long time, but I think I've finally found the words;
When something occurs, the greater the ratio between the total incentive to lie, and the total incentive to tell the truth for the witness, the greater the margin for error in one's opinion, if one believes their account, must be.
Unless one is nearby when something happens, one has to get their information from a trusted third party. And they in turn must get their information from a trusted third party, and so on, until eventually a witness is obtained. This witness may be a scientist(for example, researching global climate change), an eye-witness(as in someone present during the assassination of Bhutto), or anyone that has an immediate, personal, experience. How much trust one has in the chain between themself and the event must determine how much trust that one puts in one's belief in the account.
In addition, there is also the matter of how accurate one's trust conforms to how trustworthy the people between themself and the event in question actually is, and how trustworthy this assessment is, and how trustworthy this last assessment is, ad infinitum. This concern must too expand one's margin for error on any specific belief.
Although hopefully the margin of error on meta-inquiry converges, meaning that one can have an accurate estimate of some state of affairs, it is often not necessary, and it's entirely possible to survive a long time, just 'playing your cards randomly' so to speak, and not really having any kind of accurate conception about the world around you. Amoeba are often more or less fine with their lack of knowledge, and you could be too.