I think in these cases it'd be even simpler: the letters don't actually spell out what part of the patent is infringed or how the recipient infringes it. It'd be the equivalent of sending a letter saying the recipient's violated a contract and has to pay penalties or face a lawsuit, without saying what contract, with who, or how the recipient's violated it. The state should be able to deal with that aspect of it without going anywhere near the patent itself, that kind of behavior should constitute bad faith even if the underlying patent's valid.
A lot of trolls do send out letters without any basis, because a lot of people will take a cheap settlement rather than spend the money to fight it, go through discovery and all it's costs, and get the suit dismissed. Take a look at the SCO v. IBM lawsuit, where the SCO executives were fairly explicit about not caring whether their claims would stand up or not because it'd cost IBM more to fight and win than to settle so they figured IBM would just settle. Anti-patent-troll laws aimed at this sort of vague accusation help because it forces trolls to be explicit up front about what they're accusing their victims of which gives their victims more opportunity to knock the accusations down early on before it gets expensive.