Present discourse is replete with rogue actors. People publish and promote statements with no evidence or grounds, but the simple act of saying them seems to introduce the possibility of their truth into the epistemic considerations of the listener. If a listener judges that there is a very small but non-zero probability that what is being claimed could be true, then the model of possible worlds being considered by the listener seems to have been expanded to include those claims, with only unreliable testimony as the support for that expansion.
While it would seem as though independently the listener deems that what is being claimed is unlikely, in the possible world where the statement were true, then this testimony would seem to support it. Given that it was in fact unlikely but true, and our testifier stated it correctly, it seems as though one’s estimation of the testifier’s reliability should be revised upwards in that situation, and their testimony as included among items of supporting evidence for believing the claim.
There’s a kind of “boy who cried wolf” principle at work here. The boy (who I'd mistakenly named Peter before) spoke truthfully the third time, but by then his epistemic credibility had been blown. Someone who took [Peter]’s third claim seriously would be in a much stronger position relative to the facts in the case where the wolf really was coming, and trust in [Peter] would be to some extent restored, but there is a sense in which this act of epistemic trust would be ill-founded - if there had been no wolf the third time, we are inclined to say the belief was groundless in the first place.
But was it? Are there accounts of testimony that say that even unreliable, false testimony provide support for the beliefs about the reported propositions? And if so, how do those models tackle wilful disinformation?