Saturday, October 03, 2009

WIPIP at Seton Hall part 4

Mark Lemley, Stanford (with John R. Allison, University of Texas)

Extreme Values or Trolls on Top? Evidence from the Most-Litigated Patents

Fight over patent reform: protecting innovation or litigation abuse? Trying to replace assumptions with data from the Stanford IP Litigation Clearinghouse, with over 25,000 patent cases.

Collected every patent litigated 8 or more times since 2000, 106 in total, 2987 suits, over 10,000 defendants. Compared to a matched set litigated once over this period. Data on patents, prosecution, industry, nature of technology (can overlap with or differ from industry—a computer program used in the auto industry), and entity status (12 classes from Lemley & Myrhold).

Finding: by all economic measures, these are extremely valuable patents. Earlier finding: litigated patents are more valuable—take more continuations, cite more prior art, have more forward citation, have bigger families—than nonlitigated patents. The most litigated patents are far more valuable than the once-litigated patents. Number of continuations/non-provisional US applications in the chain: mean for nonlitigated, 1.5, for litigated 2 (median 1.5), much-litigated 4.3 (median 3)—these patents spend a long time in the patent office. Total number of forward citations for most-litigated: mean 32, median 15.5, compared to smaller numbers for once-litigated. Also more self-citations—if someone is working an industry, they’ll file a bunch of patents to create a fence.

Same result with number of claims, and references—almost 3x as many citations to US and foreign patents in the most litigated set, and number of references to nonpatent prior art is almost 10x (52.7 compared to 5.6) compared to once-litigated patents. This is a sign of validity/value according to the literature. Average nonlitigated patent cites no prior part. Much more invested prosecution process than less-litigated patents, though one can ask about correlation/causation.

Major differences in tech area and industry. Software patents are more litigated, electronics/mechanics tend to be litigated only once. More than 70% of the most litigated patents are software patents, compared to 30% of once-litigated. Mechanical patents are more than 50% of once-litigated, but less than 10% of much-litigated. Computer and communications industry dominate the much-litigated set, with pharma patents also more likely to be much-litigated (15% of set) than once-litigated.

Entity status: much-litigated are more likely to be issued to large entities, but also more likely to have been assigned before lawsuit. Substantially more likely to be asserted by NPEs. Result on NPEs even more dramatic for number of lawsuits rather than number of patents. More than 60% large entities compared to 45% in once-litigated set.

Who’s a troll? Of his 12 classes, only 1 is a traditional practicing entity. Class 10 is undetermined after research—probably NPE. Also: acquired patents, university heritage/tie, failed startup, corporate heritage, individual inventor who started company, university/govt/NGO, startup pre-product, individuals, industry consortiums/patent pool, and IP spinoffs from bigger companies.

Once-litigated patents overwhelmingly (over 80%) come from the class of traditional practicing entities, product companies. That falls to 50% for most-litigated patents. Nearly all the rest of the most-litigated are from an individual-owned company that asserts patents. What if you look at number of lawsuits, not patents? Then individual-owned companies start really to dominate.

What actually happens? Two hypotheses: patent owners of most-litigated patents ought to be risk averse and willing to settle to avoid offensive collateral estoppel. But the most litigated patents should do better in litigation when they do go to judgment because they’ve been selected for litigation and have high value characteristics. Bulletproof: went through multiple continuances to get the claims right; cited a bunch of prior art; etc.

Sadly, neither hypothesis is true. A bunch of cases are pending/transferred—a lot of MDL cases. 40 cases to judgment, 15% plaintiff win rate (6). And the settlement rate is about the same/a little less than the once-litigated patent settlement rate. (about 50%). Once-litigated plaintiff win rate is 50%, but the numbers are so small that they may not be statistically significant.

We have pretty good evidence of which patents are likely to be litigated, and which are likely to be litigated a lot. So if you want to tailor patent office examination, to spend more time on the most important patents that we’ll likely care about in the future, then you’d spend more time on the ones with more continuances, claims, prior art citations and that allocation would likely be rewarded.

Disproportionally NPEs are multiple/serial litigators. Probably thus don’t need to worry about restrictions on litigation having big bad effects on product-making entities.

Outcome data: should we question the value connection with litigation? If you have to sue 1000 defendants, maybe your patent is too broad, or maybe your patent is obvious. Or if I have to sue 1000 instead of licensing, maybe the industry has assessed the patent and found it wanting, though Lemley’s skeptical that there’s that much intention going on. Or maybe we’re just dealing with outliers, and people willing to sue 1000 companies are just different in ways that mean our rational models of litigation behavior simply don’t apply.

Chris Newman, George Mason University

Copyright Infringement as Nuisance

Copyright as property with metes and bounds, like a piece of land or a chattel with its own specific spatial boundaries. The argument: infringement is trespass, conversion, encroachment: you drove my car without my permission. But he says not all property interests are possessory. If we take property seriously, infringement is more like nuisance: you’re interfering with the IP owner’s use and enjoyment of its property, which leads to a different mindset both on remedies and on whether infringement actually exists.

Possessory uses are likely to conflict, likely enough that we’re willing to assume categorically that physical presence/possession is going to preclude other actions with respect to the property. Thus an exclusion rule may be the best way to deal with the problem—Henry Smith: easy to administer for the state, which lets the property owner make the fine-grained distinctions; easy for third parties to comply with rules—I know I simply can’t joyride in someone else’s car without permission. Reduces information costs! We enforce this right to exclude even when it gives rise to holdout problems.

We do have exceptions: accession. If you cut down someone’s trees and make them into barrels, and the barrels are much more valuable than the trees, the law sometimes allocates the laborer some of the value s/he put in, as long as the laborer acted in good faith and as long as the value created is disproportionately high in contrast to the value taken. This is a distributive justice norm. Good faith needs to be carefully police to avoid wanton nonconsensual improvements.

What about nonpossessory use conflicts, such as nuisance? Even Epstein says that a huge disparity of value between a use causing damage and value of the underlying land justifies a resort to liability rules. Even if I’m not possessing farmland, I may be in essence “using” it with the pollution I emit from my factory, except that’s indissolubly mixed by the value I’ve added by doing something else. I acquire a right to pollute (upon payment, sometimes) by accession.

Why? Maybe uncertainty—if you’re building a railroad, you know in advance you need land, and can plan which parcels you need. But with a nuisance, you don’t know exactly how valuable your enterprise will turn out to be, or what your effects on the surrounding land will be, or what the value of the surrounding land will be, and so on. (I am not convinced by the uncertainty argument. All these things are uncertain when you build a railroad as well, and there are plenty of times when you know your likely effects pretty well. And this ties in to the question of foreseeability and nuisance; foreseeable negative effects aren’t always nuisances.)

Copyright protects not tangible objects but underlying works, existing apart from embodiments. And it’s easy to wind up copying without even meaning to—as George Harrison discovered with My Sweet Lord. There’s also a boundary problem—defining what’s taking too much is not all that easy.

Before you decide whether there’s a nuisance, you actually have to assess the use. The Restatement test looks a lot like the fair use factors—you have to balance the harms and benefits. With nuisance, there’s also a significant harm requirement as a threshold before you even start to balance, which has huge relevance for fair use. Before even asking about balancing, you’d need to find significantly grave harm—harm to property that would be suffered by property in normal use/condition. So instead of asking whether a license fee could be charged, you’d ask whether there was significant harm that would be suffered by the typical copyright owner of works of that type.

My initial thought: One consideration of nuisance law is: we want to provide incentives to nuisance creators to improve their tech/change their procedures so they don’t cause the harm any more. How would this play out in copyright? This consideration ties into foreseeability as well. It could also affect fair use positions on commerciality/method of distribution, or whether a user needs to offer to take a license, or maybe this consideration is just inapplicable and weakens the analogy.

A: Then we look at what the copyright owner is actually doing to make money, and if there’s an interference with that value.

Pasquale: How do courts decide what’s not enough to be a nuisance? A survey of such cases might help linedrawing in copyright.

Katharina De La Durantaye, St. John’s University School of Law

International Solutions to the Problem of Orphan Works

Google Book Search settlement makes this problem very immediate. US proposals limit remedies on completion of a good search. Relies on private parties to aid in search, and has no collective/compulsory licensing provisions.

EU has been trying to deal with orphan works too. Commission proposal in 2006 recommended creation of interlinked national databases where users of orphan works could post ownership information that they have and describe the use they want to make; establish national rights clearinghouses that can grant collective licenses; and require diligent searches before anyone can take advantage of such licenses. Since that proposal, very little has been done. 2007: approved funding for ARROW, a database sharing information between collective organizations, national libraries, and the like.

Current digital library Europeana has 4.6 million entries covering books, audiovisual works, and others, but only a fraction of works overall, and national participation varies greatly—almost half of the works have come from France. Europeana has also struggled to find ways to include in-copyright works, for legal and financial reasons (2.5 million euros in annual budget), so the collection is almost exclusively public domain, not out of print or orphan works.

Compare (1) limitation on remedies to (2) creation of databases/collective organizations. They clearly reflect different legal approaches. US: depends on individual actions; if owner appears, parties have to negotiate a payment. Relatively inexpensive to implement and use, avoids payment of large sums that will never be claimed by copyright owners to a costly organization. Europe: relies on public ordering, establishing a publicly funded national database to reduce the problem and rights clearance centers to collect the money. Preservation of culture shouldn’t be governed by market forces. Interesting because they were both drafted in the digital age, with digitization in mind, and still didn’t take each other’s approaches into account.

Google: pressure from a potent market player (revenue exceeds the GDP of some EU members) may change the mindset on both sides of the Atlantic. France is now willing to negotiate with Google—can digitization be done with a public budget? France wants to stop envying other continents and take a hard look at European copyright to see if portions of it are no longer working. Google Book Rights Registry looks a lot like the systems (proposed) in the EU, and there’s been limited opposition to that portion of the settlement. The two legal systems are looking more similar.

The Commission knows that nothing will happen unless it scares members into action—it has been able to get members to act pretty fast on issues like consumer protection by proposing rigorous requirements, breaking down past opposition that had blocked national reform before the Commission got involved. So orphan works might be able to go that path.

No comments: