What are the valid points, and why are they easier to address with a client side solution? I'm a little bit of a data science guy, though I teach machine learning and not search algorithms :)
If an indexing service simply respected a server or account flag to be ignored, wouldn't that be enough? Or is it important that a specific account has different visibility to various searchers?