James Bessen sent me a thoughtful note following my links to Peter Detkin's data, as follows:

Regarding your piece on patent litigation data, I suggest that you check out a summary of academic research on litigation data at http://www.researchoninnovation.org/WordPress/?p=65. Some of the conclusions are quite different than what Detkin and Myhrvold have been presenting, including:

- The rate of litigation per patent is NOT decreasing. Several careful studies have come to this conclusion. Note that litigation per patent needs to be measured over the life of the patent. Judging from the magnitudes reported by Detkin, they are not doing this (they report a rate of .0025 or so for litigation per patent; the other studies report more like 1-2%,), although it is not clear to me what exactly they are measuring.

- The rate of litigation per R&D dollar is increasing rapidly. This may be a more meaningful measure of the "tax" that litigation imposes on R&D. It is NOT the case that all the "defensive" patents that have been acquired are used purely defensively. In fact, Mike Meurer and I find that the more patents a firm has, the MORE likely it is to be sued and the more likely it is to sue, all else equal.

- The comparison to trademark and copyright suits is not particularly meaningful--patent litigation is far more expensive.

- I believe that the data they report on settlement is settlement AFTER a suit has been filed. This likely misses the majority of disputes that occur without a filing. Moreover, even suits that are settled before a full trial can be quite expensive. Many suits settle after the Markman hearing at which the judge decides how to interpret the patent claims (some legal scholars interpret the Markman hearing as the first stage of a trial, but this is not recorded in the statistics as such). The discovery costs preceding settlement can be large. Moreover, these are not quick settlements by and large--Jay Kesan finds the median duration to take about 10 months. Studies based on stock market reactions to lawsuit filings suggest that business opportunity costs such as lost sales and management distraction can be quite large even in cases that settle before trial.

I do believe that the point Detkin and Myhrvold make about "trolls" is correct--trolls (however they are defined) do not represent a substantial portion of litigation. On the other hand, trolls may represent the "canaries in the coal mine"--their significance is not in their absolute numbers as much as an indicator that patent property rights may be poorly defined.

---"

Best
--Jim