> their whole business is litigating patents, so tying them up in court is not exactly a deterrent.

Lawyers can make a considerable amount of money, any disruption to the Patent Trolls' game will severely reduce their numbers. These Patent trolls, as you said, are often lawyers and by tying them up in court you are costing them profitability. Every non-settlement forces them to go to court and actually fight.

This is how the courts work: Out-of-court settlements are frequent and are preferred as it saves a Judge's time for real cases, however mass submission and withdrawal of suits are seen as a waste of the courts time and effort. A fair amount of cases end up in front of a judge before settlement, if these end up being withdrawn (and the judge will know it's withdrawn not settled) you're going to be in bad credence with a judge. Pissing off a judge is a big no-no. Patent Trolls will be pissing off multiple judges, which is like putting a shotgun in your career's mouth and pulling the trigger.

An insurance agency that fights patent trolls would be ideal. Offer $X amount of legal fee coverage for $Y a month, get a few talented lawyers and a fair amount of paralegals. Once word gets out, the majority of the work can be performed by paralegals (essentially everything up until you're standing in front of a jury-or-equivalent can be performed by paralegals) as no troll will want their time tied up. Eventually, only Trolls with a fair-case will fight, you then have 3 options to give your clients: offer them your lawyers' services/suggest settling/wish them good luck elsewhere. It would be a sort of no-case-no-extra-fee's deal.

Lawyer's typically bill by the 6-minute interval, and frequently charge 100's of dollars an hour. In fact 1-2 billed hours is typically a lawyers daily pay, and they may still have 6-8 hours on top of what they're billing clients. If you start consuming these lawyers time, their profitability drops, especially considering they require settlements for their pay. They're invested in the 'easy win', which isn't conducive to fighting a drawn out case. If companies actually start going through with the court battles, the payment is potentially delayed anywhere from 3-months to 3-years.

Part of my assumption is that if the patent should truly not have been granted, the start-up has a decent chance of winning if they have the resources to fight it. (I do know of some counter-examples, like RIM and D2L)

The trolls don't want to enter a losing battle because of the cost (even if they're in the business of litigation, time spent suing one company is time that could be spent suing another). Also, losing would damage the validity of the patent for future cases. (In my understanding, that is. IANAL)

There are a couple of other angles you could work here. In some jurisdictions, it's possible to seek damages for the cost of defending a failed lawsuit. I don't know if any of these jurisdictions extend this to patent litigation, or what tax and other implications might be of incorporating there (so that one might legitimately change venue to such a location). The two typical problems with this approach are getting into the right jurisdiction and being able to front the money for the defense in the first place.

Also, if the patent troll is suing a bunch of companies you're insuring based on the same patent, you could attempt to combine the cases, thus consolidating the defense costs. This works because patent trolls work by firing off lots of very similar lawsuits, each for a small enough amount that it's economical to settle. That's what keeps their costs down; such an insurance policy would let the defense leverage the same economy of scale.

The assumption is if you have few assets and the insurance then there would be little reason to litigate. Which would drive down the cost of the insurance.