Here is my favourite parable illustrating a serious problem with typed theories of truth, and drinking too much when having serious academic discussions, from Leon Horsten’s book The Tarskian Turn (MIT, 2011).
[S]uppose that you are convinced that on the subject of the history of Mayan civilization, Anna is extremely reliable. Suppose that even though you were not in Anna’s company yesterday evening, you know that the conversation in which she was engaged was restricted to the subject of the history of Mayan civilization.
Then you may say:
Everything Anna said yesterday evening is true.
Because you do not know what specific sentences Anna asserted yesterday evening, you do not know how far she got carried away after a few drinks in nesting occurrences of the truth predicate. Therefore, you have no means of determining how high you have to crank up the level of your use of the truth predicate in the sentence you assert about Anna. (p. 56)