RIM had to be huge before NTP bothered to shake them down form money. All of those examples above share the characteristics that they were huge successful companies when they were shaken down. The results were also fairly random.
I think it fits the model that once you are large and successful, the trolls will come out of the wood-work. The compliment is that you are big and successful enough to be worth a shakedown. Once you are the $250k to actually defend (its really more like $1 mil/year for up to 7 years) is chump-change to you at that point. The big problem is the more or less random outcomes of such trials.
Until you are making big money, you are not worth the time because a victory would be costly to the victor and you'd have nothing to actually turn over. If they won debt, you'd just close up shop, as there'd be no point in working to pay a troll.
When I say that "there's no such thing as infringement" I'm not talking in the made-up world of legal types. They could rule you infringing for tying your shoe if they had the right lawyers, I'm speaking from the realistic standpoint that no two software systems are ever really alike unless you outright steal a copy of the source. I've used the example before that 2 computers that factor large prime numbers look form the outside like a pair of beige boxes that do exactly the same thing, but inside they are likely worlds different. This confuses the lawyers, patent officers and uneducated outsiders and leads to the problems that the patent system now faces when it comes to software.
When you patent a machine, you have to show its inner workings and they have to be novel. When you patent software, you can pretend that the general purpose computer is the inner workings, and not the actual software. This lets you cheat and patent a concept directly, without proof that the inner workings are novel or even possible.