I don't think you can "remove" information from A probability problem in this way. And even if you could there's no reason to assume it would zero-sum with the later information gain.

Happy to be wrong though, been a long time since i dabbled in probably.