Isn't the argument in one false? If one applies bayes' theorem, with initial prob. 50% and new likelihood ratio of a billion to one, don't you get 500000000 to one chances?