(Professor Michael Risch of the Villanova University School of Law is a prolific scholar who is fascinated by patent trolls. Given his interest in this subject, I invited him to write a short post on his current project examining conventional beliefs about patent trolls. - JAR)
Guest Post by Michael Risch
Few players in the patent system (maybe none) are more hated than patent trolls. There is a lot of debate about what it means to be a patent troll, but the commonly accepted definition is a plaintiff that asserts patents (usually acquired ones) but makes no product. More accurate terms are non-practicing entities (NPEs) or patent assertion entities (PAEs), but I’ll keep using trolls here if only to be mainstream.
Commentators (including me) have intuitions about the types of patents that trolls assert; the media is filled with claims. The problem with these intuitions is that they are all anecdotal – we think we know what we know from the limited reporting we see about cases that hit the news. But trolls have been around for a long time (two of the entities I studied brought their first suit in 1986) and they file a lot of lawsuits.
I wanted more evidence to support my intuitions, so I set out to comprehensively study a group of trolls – namely the ten most litigious trolls. I don’t have space here to discuss why I chose this group, but the paper discusses my reasons as well as why I believe that the findings here apply to less litigious trolls. In all, I studied 10 trolls, about 1000 cases, and about 350 patents over a 23 year period. The full paper examines the type of patents, case outcomes, patent quality, venture funding, and other aspects of the cases and patents.
It turns out that most of what I thought about trolls – good or bad – was wrong. But first, one area that fits conventional wisdom: the moniker “troll” is accurate to the extent that it means waiting before asserting a patent. The patents I studied were asserted for the first time on average more than seven years after issuance. That’s a lot of time for an industry to develop. That said, delay decreased as issue dates increased. In other words, older patents sat on the shelf much longer than new patents. Patents issuing in 2006 were asserted on average within 2 years. This could mean that trolls are acting more quickly, or it could be what we call a “selection effect” – only some recent patents were asserted early and there is a large group of patents yet to be asserted; 15 years from now, it may be that the average delay for patents issuing in 2006 is seven years, though I doubt it.
Perhaps the biggest surprise in the study was the provenance of patents. I thought most patents came from failed startups. While such patents were represented (about 14% of initial assignees were defunct), most came from companies still in business in 2010. Indeed, more than a third of the initial assignees were publicly traded, a subsidiary of a public company, or venture capital recipients. Only 21% were patent assertion entities at the time the patent issued, and many of those were inventor owned companies (like Katz) rather than acquisition entities (like Acacia). Further, about 26% of the patents were inventor owned, and a comparison with the percentage of individual inventors represented in litigation generally shows that trolls serve an important role in enforcing individual inventor patents. Of course, if you think individuals have no place in the patent system, this is not a good thing.
Another area of surprise was patent quality. While trolls almost never won their cases if they went to judgment (only three cases led to an infringement finding on the merits), the percentage of patents invalidated on the merits was lower than I expected. A total of 43 patents had validity adjudicated on the merits. Only 4 were found completely valid. Another 23 were held completely invalid, and the rest were partially valid. Thus, just over half were invalidated. This sounds high, and it is—but consider that most of these cases involve a group of related patents that stand or fall together. When viewed on a case by case basis, as most other studies do, 13 cases out of 46 cases that reached a final judgment (28%) fully invalidated a patent. Compare this with a study of all patent cases filed in 1995, 1997 and 2000, which found that 118 out of 584 summary judgment or trial rulings (20%) invalidated a patent. Viewed this way, the invalidation rate for patent troll cases is not that much higher than patent cases in general. The quality is worse, but not that much worse (and as discussed in the paper, methodology differences may mean that the difference is smaller than 8%). Of course, it is more costly to defend against multiple related patents, so litigation strategies still have a social cost.
These are just a few of the findings. A full draft of the paper, called Patent Troll Myths, is available on SSRN and will be published next year in the Seton Hall Law Review. I am also planning a followup that looks more closely at outcomes and patent quality over time. Comments are appreciated.