One reason you're getting different answers from different people is that the question can be interpreted in different ways, and it isn't clear what you mean by "probability" here. One way to make sense of the question is to assign priors and reason using Bayes' rule as in Matthew's answer.

Another interpretation is the following. In setting up the question you're ascertaining that the color of the car is well defined and that it *is* blue. Given that context, the "probability" that the car is blue, is one. The person asked of course doesn't know this, so can at most say that the probability is either zero or one, depending on whether the car is actually blue or not. The fact that 1000 people were asked about the color and couldn't agree doesn't affect the fact that the car has a fixed, but unknown (to the person asked), color.