Yggdrasil said:
If we leave the question of math and biology out of it, we can view it linguistically. 
Various theories in pragmatics, a branch of linguistics, have a theory of effective communication that started with Grice's conversational maxims, a set of observations about how people communicate. These are not laws on how to speak, but rather observations on how people communicate. From these maxims, you can posit that people will usually say the most they can to describe their point, but no more. The result of this supposition is that people usually assume that the conversation has followed these Gricean patterns.
For example, let's say I describe an actor as an Oscar nominee. This is %100 true even if that actor actually won the Oscar, but our Gricean assumptions say that if he had won, he would have been described as an Oscar winner not a nominee.

How is all this blather relevant, then? If someone says that they have two children, and one of them is a girl, we assume that they mean that only one of them is a girl, even though the sentence would be %100 true if they had two girls - just because people don't speak in a rational, logical manner - they speak in patterns that evolved to maximize information delivery while minimizing cognitive processing complexity.
Hey, do you know how I hate the fact of having to think bout this whenever I'm 'communicating'? I'm all the logic type of a person for me it'd be the 66% thing, then if I recall that stuff again I have to reconsider. That's all fine and all, but what bothers me is, I know wether if someone meant something the way they said it, or not, but they just say something without thinking about it, that's so confusing when you can have to read peoples minds... people only heavily think about what they say when they're lying or confused...

It just sucks, that when I'm talking to people they do just assume the stuff according to the communication rule, not considering the fact that other possibilities are there, even though their probability being less and that really bothers me.
Would there be a logically or programmatical way of solving such things in the full complexity? Sven's code would just stochastically calculate the probability from random values... that'd be interesting Smiley