Thursday, October 20, 2016

IP and Climate Change

My colleague and friend Josh Sarnoff (DePaul) sent me a review copy of the book he edited: Intellectual Property and Climate Change, even though I told him I wouldn't have much time to look at it. Wouldn't you know, on a quick skim I found it pretty interesting, and thought I would talk about it a bit.

The book is part of the Elgar Research Handbook series. I wrote a chapter that I really like (who am I kidding, I just love that book chapter) in the Research Handbook on Trade Secret Law. But because it's in an expensive book, nobody seems to know about it (and my colleagues in trade secret law will attest that I remind them whenever I review one of their drafts that is remotely in the area of trade secrets and incentives).

So, I thought I would flag this book, so readers would know this is out there. IP will have a growing role in climate change, as this cool story from this week illustrates. The book is comprehensive - it has 26 chapters from a variety of different authors. Some of the topics:

  • International law and TRIPS
  • Enforcement
  • Technology transfer
  • Innovation funding and university research
  • Antitrust, patents, copyrights, trade secrets, trademarks
  • Rights in climate data
  • Privacy (this one surprised me)
  • Standards
  • Energy, transportation, food, natural resources
There is something for everyone in this book. Though it is focused on climate change, much of the discussion can be generalized to other emerging areas of law. In that sense, it does present a little bit like the law of the horse, but given that this is a research handbook, I'm not so sure that's a bad thing.

Tuesday, October 11, 2016

Apple v. Samsung Oral Argument Thoughts

The Supreme Court heard Apple v. Samsung (technically Samsung v. Apple) today. The transcript is here. This post is short and assumes some knowledge of the issues: the tea leaves appear to be that many justices are uncomfortable with the current rule that treats the "article of manufacture" as the entire product. Even Apple seemed to give ground (surprisingly) and admit that what the article of manufacture is a factual determination (but that Apple satisfies those facts in this case). This was an interesting concession, quite frankly, as it is not a given that the Court would accept its view that no new trial is needed. The Court did not seem to even want to hear the "don't remand" argument.

But it also seems clear that the Court has no idea how a rule would operate in practice. After all, it's been 140 years and as far as I can tell, one important case has ever denied profits to a design that was not sold separately (on a refrigerator latch). And Justice Kennedy, in particular, was worried about how one would distinguish a method of determining profits of the patented article of manufacture (OK) as compared to the apportioning profits associated with the patented design (Not OK).

I've made no secret that I favor Samsung's view here. I signed on to an amicus that said as much, and I believe that there is case law mentioned briefly by counsel for Apple (and cited in the amicus) that allows for assessing profits only associated with the infringement, not the entire product. For example, in Westinghouse, the Court held that even if all profits were to be assessed with no apportionment, the defendant could still provide evidence that separated profits unrelated to the infringement.

But that said, I think the argument highlights the central tension here. In Dobson v. Hartford Carpet (1885), the Court considered the very same arguments made today: that people buy carpet for a variety of reasons, that design might drive sales, but there are also other inventions that affect quality in the carding, spinning, dying, and weaving (not to mention the backing). Thus, the Court held, any profits must be allocated based on the patented design. It would be unfair, the Court ruled, to force defendants to pay out their entire profits twice if they infringed another patent.

It was this ruling that Congress promptly intended to reverse by making a rule that all profits on the article of manufacturer would be owed.  If the proposals discussed today were in place then, Dobson would not have received all of the defendant's profits; I cannot believe that Congress intended this outcome then, and so I wonder how one can read the statute differently now.

And that's where I'm left. The utilitarian in me says this statute is wrong in many ways (and I've looked for ways to achieve that result through statutory interpretation, like Westinghouse). But the statutory intepreter in me finds it hard to believe that we've somehow misunderstood this statute for the last 140 years, and that the solutions today would leave us with the same outcome that Congress quite clearly intended to reverse. If the Court does decide on narrowing damages, I wish it good luck in finding a happy medium.

Friday, October 7, 2016

Apple v. Samsung, Part ?? (not the Supreme Court case)

I've lost count of the rounds back and forth in Apple v. Samsung. But another opinion issued today, and it was a doozy. When we last left our intrepid litigants in late March, the panel had reversed the $120m verdict in favor of Apple, ruling a) non-infringement of a patent, and b) obviousness of two patents (slide to unlock and autocorrect). These rulings were as a matter of law - that is, they reversed jury findings to the contrary.

Apple filed for an en banc hearing, and we never heard anything again...until today. The en banc Federal Circuit (except Judge Taranto, who did not participate) vacated the panel opinion. The decision came without briefing, and it was unanimous...except for the three members of the original panel, who all dissented.

The opinion begins with a statement (rebuke?) about what appellate review should do: take the facts as found by the factfinder, and then review them for substantial evidence. The opinion takes umbrage at the fact that substantial evidence is not really addressed by the original panel at all. The opinion also takes issue with "extra record" material being considered for claim construction on appeal. The opinion then goes through each patent and shows the substantial evidence that would support a verdict, even if the appeals court would disagree with it.

Perhaps most telling of the deferential approach is the slide to unlock patent, which I think is the weakest of the bunch. The prior art, when combined, clearly has all the elements. But in finding non-obviousness, the jury found that the person with skill in the art would not have combined the references. That might be wrong, but evidence was submitted to support it, and thus the claim is non-obvious. The dissent takes issue with this, saying that KSR loosened up the combination standard. More on this later.

This opinion has a lot of important aspects:
1. It is another en banc opinion without briefing. I am sure the litigants (especially the losing ones) hate that. As an observer, I'm not so bothered in this case. The briefing was full and complete, and there was little to add in the way of analysis.

2. Why is the en banc circuit showing up only now to defend substantial evidence? I can think of at least two prominent cases in which juries made non-obvious findings of fact that a Federal Circuit panel disregarded to find a patent obvious and the en banc request was denied. Why now?

3. Just what is the obviousness standard of review and who is supposed to make these decisions? In general, the final obviousness determination is one of law, based on underlying findings of fact. Many judges have juries decide those underlying findings of fact, such as the scope of the prior art or the motivations to combine references. Some jury instructions ask in detail, and some just say "is it obvious?" If it is the latter case, then any jury finding is entitled to all inferences on appeal - if the jury said non-obvious, then it must have found no motivation to combine. The problem with this approach (and even the specific question approach) is that it makes it hard to "loosen" a standard as KSR v. Teleflex says we should. KSR affirmed a grant of summary judgment by the district court - in other words, it affirmed a finding that references could be combined as a matter of law. It is unclear why an appellate court could not have made the same determination here. At the same time, why have trials and findings of fact if we are simply going to ignore them? Deciding how obviousness should get decided is almost as important as the obviousness standard itself.

4. This opinion shows the importance of which panel you draw at the Federal Circuit. Apple had the bad fortune to draw the only three judges to disagree here. Of course, we don't know how the en banc dynamics work, and perhaps some in the majority here would have concurred in the original panel opinion. To generalize, though, judicial preferences may drive the disparity in opinions in Section 101 right now. A couple cases that have just issued are ripe targets for en banc review as well to aid this.

This is my final takeaway - if any part of this case is to make it to the Supreme Court, it will be the slide to unlock patent. This is a patent where all the elements are in the prior art, and there is a real dispute about the procedure for determining obviousness. This question has been presented to the Court before, but perhaps this version will take hold.

Thursday, October 6, 2016

The Long Awaited FTC Study on Patent Assertion and Nuisance Litigation

After months of speculation that the FTC's long awaited FTC study on patent assertion entities was going to issue any day now, the study has finally issued. The press release is here, and the full PDF is here.There is a lot to learn from this study - the FTC had subpoena power to obtain data unavailable to mere mortals like the scholars who study this area. I thought that the study was generally balanced and well researched. A scan of the footnotes alone will make a great literature review for anyone new to this area (even if it is missing a couple of my articles...).

Some of the highlights of the study come right out of the press release:
The report found two types of PAEs that use distinctly different business models. One type, referred to in the report as Portfolio PAEs, were strongly capitalized and purchased patents outright. They negotiated broad licenses, covering large patent portfolios, frequently worth more than $1 million. The second, more common, type, referred to in the report as Litigation PAEs, frequently relied on revenue sharing agreements to acquire patents. They overwhelmingly filed infringement lawsuits before securing licenses, which covered a small number of patents and were generally less valuable.
The report found that, among the PAEs in the study, Litigation PAEs accounted for 96 percent of all patent infringement lawsuits, but generated only about 20 percent of all reported PAE revenues. The report also found that 93 percent of the patent licensing agreements held by Litigation PAEs resulted from litigation, while for Portfolio PAEs that figure was 29 percent.
The separation of portfolio versus litigation business models was an important one - something I discuss in Patent Portfolios as Securities and Lemley & Melamed discuss in Missing the Forest for the Trolls.

The study also debunks the notion of widespread demand letter abuse, but does show that PAEs tend to sue first and demand later. It shows about $4B in revenues for all study respondents over a 6 year period, 80% of which came from portfolio companies (and mostly not after litigation). The study is unclear what this $4B represents if extended to all PAEs, but provides some statistics that might aid in a ball park calculation for the whole market.

There's more -- a lot more -- to this report, which runs 150 pages before the appendices. The study concludes with some reasonable and mostly uncontroversial ways the system can be made better, such as limiting expensive discovery at the early stages of a case. That said, I do take issue with one point of the study - relating to nuisance litigation. More on this after the jump.

Wednesday, October 5, 2016

Helsinn v. Teva Oral Argument Recap

In March, I posted about an amicus brief filed by 42 IP profs in Helsinn v. Teva, which argued that contrary to the district court's opinion and position taken by the USPTO, the America Invents Act (AIA) did not change the meaning of "on sale" and "public use" in 35 U.S.C. § 102(a)(1). The case was argued yesterday before the Federal Circuit, and the panel (Judge Dyk, Judge Mayer, and Judge O'Malley) didn't seem eager to conclude that the AIA wrought a significant change.

The appeal involves Teva's challenge to Helsinn's post-AIA patent on the nausea drug palonosetron, which was filed over a year after a secret licensing and supply contract for the drug. In Pfaff v. Wells Electronics (1998), the Supreme Court held that the on-sale bar applies when a product is (1) "the subject of a commercial offer for sale" and (2) "ready for patenting" as of the critical date (one year before filing). Both issues are contested here, as the district court said that the drug was neither ready for patenting nor on sale within the meaning of the post-AIA § 102. I'll focus here just on the AIA issue, but note that Judge O'Malley asked about remanding for further factfinding and whether it is necessary to reach the AIA issue.

The only line of questioning on the AIA issue for Teva was Judge Dyk's criticism of the dueling canons of statutory interpretation for figuring out what "or otherwise available to the public" means in the new § 102. Teva argued that under the "last antecedent" canon, "to the public" modifies only "otherwise available"; Helsinn countered that under the "series qualifier" canon, the concluding phrase "otherwise available to the public" qualifies everything in the series, including "on sale." But Judge Dyk stated that neither canon can apply because the modifier would be "available to the public," leaving just the word "otherwise," which doesn't make sense. Teva pivoted to its argument that "or otherwise available to the public" is a catchall category for new technologies, which the panel seemed comfortable with; Judge Dyk suggested "an oral description at a conference" as something that might fall into this bucket.

Tuesday, October 4, 2016

A Comprehensive Study of Trade Secret Damages

Elizabeth Rowe (Florida) has shared a draft of "Unpacking Trade Secret Damages" on SSRN. The paper is an ambitious one, examining all of the federal trade secret verdicts she could find (which she believes is a reasonably complete set based on her methods) that issued between 2000 and 2014. The abstract is:
This study is the first to conduct an in-depth empirical analysis of damages in trade secret cases in the U.S. From an original data set of cases in federal courts from 2000 to 2014, I assess the damages awarded on trade secret claims. In addition, a wide range of other variables are incorporated into the analysis, including those related to background court and jurisdiction information, the kinds of trade secrets at issue, background details about the parties, the related causes of action included with claims of trade secret misappropriation, and details about the damages awarded.
Analysis of this data and the relationship between and among the variables yields insightful observations and answers fundamental questions about the patterns and the nature of damages in trade secret misappropriation cases. For instance, I find average trade secret damage awards comparable to those in patent cases and much larger than trademark cases, very positive overall outcomes for plaintiffs, and higher damages on business information than other types of trade secrets. The results make significant contributions in providing deeper context and understanding for trade secret litigation and IP litigation generally, especially now that we enter a new era of trade secret litigation in federal courts under the Defend Trade Secrets Act of 2016.
I think this study has a lot to offer. Although it doesn't include state court cases, it provides a detailed look at trade secret cases in the first part of this century. Of course, the verdicts, which were about 6% of all trade secret cases filed, are subject to the same selection effects as any other verdict analysis - there is a whole array of cases (more than 2000 of them in the federal system alone) that never made it this far, and we don't know what the tried cases tells us about the shorter-lived cases.

The study offers a lot of details: amounts of awards, states with the highest awards, states with the most litigation, judge v. jury, attorneys' fees, punitive damages, the effect of NDAs on damages, etc. It goes a step further and offers information about the types of information at issue, and even the types of information that garner different sizes of awards. It's really useful information, and I recommend this study to anyone interested in the state of trade secret litigation today.

There are, however, a couple ways I think the information could have been presented differently. First, the study has some percentile information which was great, but most of it focuses on averages. This is a concern because the data is highly skewed; one nearly billion dollar verdict drives much of the relevant totals. Thus, it is difficult to get a real sense for how the verdicts look and there is no standard deviation reported.

Of course, the median award according to the paper is zero, so reporting medians is a problem. I particularly liked the percentile table and discussion, and I wonder whether a 25/50/75 presentation would work. Speaking of zero dollar awards, though, I thought the paper could be improved by clarifying what is calculated in the average. Is it the average of all verdicts? All verdicts where the plaintiff wins? All non-zero verdicts? Related to this, I thought that clearly disaggregating defendant verdicts would be helpful. The paper reports how many plaintiffs won, but this is not reflected in either the median or average award data (that I can tell - only total cases are reported). At one point the paper discusses the average verdict for defendants (more than $800,000) which is confusing since defendants shouldn't win any damages. Are these part of the averages? Are they calculated as a negative value? If these are fee awards, they should be reported separately, I would think.

Though I would like more data resolution, I should note that this really is just a presentation issue. The hard part is done, and the data is clearly available to slice and dice in a variety of ways, and I look forward to further reporting of it.