Has the difficulty of ascertaining freedom to operate led to a patents arms race?

Reading Boldrin and Levine's "Against Intellectual Monopoly" (Cambidge University Press, 2008, and here), I am struck by a thought that isn't directly related to the authors' arguments against the patent and copyright systems (I expect they will get round to trade marks in due course, although that will have to be on a different basis). Perhaps it is such a banal thought, it isn't even worth putting out here, but I don't recall it being expressed in these terms before so maybe I have something original to say ...

There are so many granted patents in existence (why on earth did I waste time typing "granted"? There are so many patents in existence ...) that "freedom to operate" searches must be little more than a lottery. This may explain why patent owners (in some fields at least) now put their efforts into collecting extensive portfolios so that when (not if) a competitor sues them for infringement they will be able to find a patent which the competitor is infringing in their own portfolio. Then the problem can be dealt with by cross-licensing rather than fighting out the infringement claim in court.

There must also be a problem with unpublished applications (not quite submarine patents but similar) which form an undiscoverable part of the state of the art - they would not show up in "freedom to operate" searches but could prove fatally damaging. The accelerating pace of technological change must make these a bigger and bigger problem, which in turn makes freedom to operate searches even less reliable, and gives further encouragement to alternative defence mechanisms.