I'm not a lawyer but in the case of patent infringement I expect the plaintiff to be able to show said infringement. It is a technical matter that can be deduced from the product by studying its functionality. The defendant practically cannot hide infringement.
There is a balance to be struck here, I agree, but in one example [0] it cost a troll an estimated $450 to file a completely baseless lawsuit which would have cost the defendant almost $200k to have dismissed ("would have", but didn't, only because the attorney worked pro bono). That doesn't seem to me to indicate that the balance is currently struck at about the right place.
It seems to me that before the court allows the dogs of discovery to be loosed upon the defendant, it is more than fair to require the plaintiff to be specific about what claims it is alleging infringment of and what the defendant is doing that constitutes infringement.
[0] http://www.techdirt.com/articles/20130814/02270724171/massiv...
Think of it this way. You have a patent. You're pretty sure, but not positive, someone is infringing on it. Without the ability to inspect the supposed infringer's facilities, how do you plead with particularity factual allegations sufficient to support an infringment claim? If the infringment is blatent, sure this is easy. But in more nuanced cases (I can't think of a technical exmaple) it is possible to end up in a siuation where actual infringment is occuring but because of an inability to access the right information there is not enough available to sufficiently plead a claim.