Peter Detkin sent me the following commentary on James Bessen's comments:

I read Mr. Bessen's comments on my presentation with interest. While he and I can argue at the margins about some of the data (in fact, I understand he and my partner Nathan Myhrvold have had a number of such discussions), at bottom we are in fundamental agreement: the data shows clearly that trolls do not represent a substantial portion of litigation (contrary to the wild and unsupported assertions of many in DC). They do exist -- which is why I developed the term "patent troll" in the first place -- and are an issue to be dealt with, but we need to be careful not to throw out the baby with the bath water.

The patent system as its exists can improve performance so long as the PTO has the resources to do its job. Ending fee diversion will go a long way to addressing many of the patent quality, pendency, and examiner retention issues. As for patent litigation -- while there are some anomalies that can be cleared up; the law does not need to be overhauled completely. For example, the current law on wilfullness can produce anomalous results, and can expose companies to enhanced damages in unfair situations. The latest coalition proposal under consideration in Congress does a good job of balancing interests; let's hope it can move reasonable reform successfully.

Mr. Bessen is correct that the settlement data I discuss is for settlement after a suit has been filed. I admit that this misses those matters that are settled without filing; I dispute, though, that this is significant. My experience is that matters settled without litigation tend to be for those patents for which a strong case has been presented in negotiations by the licensor, and thus litigation would not be appropriate. "Nuisance value"settlements are relatively rare. I'm not denying that they happen, of course, but I don't believe they happen on a regular basis. Thus, if anything, I believe an analysis of settlements that occur in the absence of litigation would yield data that favors the patent system, not detracts from it, as it would show that there are many patents that are quite strong and respected by the industry.

Peter Detkin

So we have consensus that "trolls" are not "the" problem; which is just as well, since any attempt to enact a law to "do" anything about trolls is likely to be a definitional nightmare, and also foreclose important options for new business models in innovation.

There are still problems.

The settlement question continues leaves me uneasy. The "tech" position in the battle to make it harder to get an injunction in the case of a finding of infringement has been that any time a major firm (say, IBM or Intel or what have you) makes an announcement that they are going into a new line of business, they get 50 letters insisting that they cannot do so without infringing the patent of someone they never heard of. If they are infringing, the problem is that it is hard to do tech/software without doing so given the sheer number of patents, problems with prior art searches and so on. If they are not infringing, the problem is nuisance settlements. So does anyone have the data? How rare are they? If they are indeed rare, it seems that tech will have trouble making a case that this is indeed a problem they cannot handle; the best solution may be loser pays. If on the other hand they are in fact infringing, then on the one hand the "tech" side is less sympathetic--unless it is indeed truly impossible to make progress without infringement, in which case the solution is taking patent quality up a notch.

Getting the data on settlements would help.