Julian Sanchez has a typically excellent post on the problem of bad patents:
Mike Masnick writes that a surprising number of online music vendors are choosing to settle rather than fight lawsuits from a company called Sharing Sound which managed to secure a ridiculous patent on, essentially, the very idea of selling music online. They are, one assumes, settling because Sharing Sound is asking less than what it would cost in legal fees to fight the patent, even if a victory were all but assured.
It strikes me that we can think of the invalidation of bad patents as a public good: If someone does fight back and get the patent declared bogus, that's game over for Sharing Sound--and good news for all their other potential victims, who are now spared the threat of litigation (or the cost of settling) without themselves having to pony up for legal fees. For just this reason, of course, Sharing Sound is probably careful to limit their settlement demands to less than the cost of beating them in court. The patent is sufficiently plainly bogus that if anyone were expected to fight, the lawsuit would probably not have been worth filing in the first place, since it would just result in a costly defeat.
It seems entirely possible that the combined cost of all those settlements is actually quite a bit more than the cost of mounting one successful counterattack in court. But, of course, if anyone proposed that the target companies pool their resources to mount a unified defense, each would have an incentive to free ride. You could have some kind of insurance-like setup where a consortium of tech firms keep a legal team on retainer to defend any members targeted for a suit over a patent determined by some independent panel to be sufficiently frivolous, where companies that don't join make more attractive targets for patent trolls, which reduces the incentive to free-ride. If the companies together can credibly commit to fight such suits, then assuming sufficient transparency, the existence of the mechanism should create enough of a deterrent threat that they'd seldom actually have to. But there are plenty of reasons such a system might be unworkably complex and have its own incentive problems.