Sunday, September 27, 2009

TPRC 2009: Copyright, DMCA and IPRs

Moderator: Sherwin Siy, Public Knowledge

Statutory Damages in Copyright Law: A Remedy in Need of Reform
Pamela Samuelson, University of California Berkeley, Tara Wheatland, Berkeley Center for Law & Technology

Tripartite structure of 1976; statutory damage awards in recent years have become sometimes grossly excessive, ignoring the initial statutory scheme. The paper articulates principles to make statutory damages better.

Some say the US has always had statutory damages, so they can’t be unconstitutional. But a 50-cent-per-sheet penalty from the Statute of Anne, 1790 Act, in copyright until 1909, was substantially different from today’s statutory damages. Explicitly penal, and thus narrowly interpreted; was a fixed fee, half going to the US government; rarely used because an equity court couldn’t grant, and plaintiffs usually wanted equity for an injunction and an accounting, which wasn’t available in a common-law court.

Legislative history of 1909 Act: considerable dissatisfaction with per-sheet penalty.

1909: For the first time, the law-equity split was ended—could get plaintiff’s lost profits plus defendant’s profits, plus injunction, costs and forfeiture, in one action. Statutory damages were created as an “in lieu” alternative to lost profits and defendant’s profits, not as a penalty. The goal was to compensate the plaintiff when damages/profits were difficult to prove. The perceived need for a penalty was broken off into a new criminal provision for willful infringement for profit. Caselaw: generally in keeping with compensatory goals. Courts would sometimes refuse to award statutory damages where actual damages were shown/approximated.

1976: New tripartite structure, creating a special rule for innocent infringement: could lower to $200 or even $0 nonprofit educational users. Ordinary infringement, $250 to $10,000 per infringed work. This was a change directed at making excessive awards less likely—per work, rather than per infringement. Didn’t foresee problems in modern era of multiple low-value works. Courts were expected to take actual damages into account in setting actual damages in cases of ordinary infringement. The reform that gave rise to the gravest modern problems: a new enhanced level of damages for willful infringement up to $50K. Congress expected that this award would only be available in exceptional cases like counterfeiting, not just every time someone should have known that a work was infringing.

Today, copyright owners who qualify can elect statutory damages any time before final judgment. 1909 Act: anyone could get statutory damages; 1976 Act: registration within three months of publication/before infringement began required. Means that major copyright owners are eligible, but not all victims of egregious infringement. The ordinary infringement minimum has now been raised to $750 and the maximum to $30,000; courts often treat ordinary infringement as willful, which goes up to $150,000. “Should have known” is now the standard, even where there were plausible, albeit unsuccessful, fair use and other defenses. This is inconsistent with what Congress intended. Courts haven’t developed a guiding jurisprudence. Congress expected courts to be the ones awarding statutory damages, but Feltner held that there was a right to a jury trial, and some of the worst awards come through juries.

Excessive awards: Jammie Thomas on 2nd trial, $80K per song: $1.92 million award. The jury was divided—some wanted to award $750 and others wanted to award larger sums, and they picked a number. Compare to $220K award in first trial, and $750 per song awarded by trial judges in 12 other music p2p cases. Also: $118 million Judge Rakoff was ready to award to for its “beam-it” service, despite lack of actual damages to RIAA firms and lack of profits. Rakoff explicitly said he was doing so in order to deter and also all those other people out there on the internet paying insufficient attention to copyright law. $19.7 million award against Legg Mason for photocopying articles form journals to which it subscribed, compared to $66K in actual damages. $1 million initial award in LA Times v. Free Republic, where a nonprofit conservative website posted news articles to show bias; eventually settled for $10K.

Other examples of arbitrary awards: $300K for posting two poems on a website, but $750 per work for Scientology texts. Peer cases involving similar sound recording infringements—vastly different amounts per work.

Huge damage awards and their potential have real chilling effects on many different sectors: documentary filmmakers, tech innovators—where the per infringed work rule leads to grossly excessive awards because each song that might be processed is potentially a new award. Google Book Search: another example where the exposure Google faced for scanning 8 million books is quite substantial.

Principles for courts: award the minimum when there’s no damage to plaintiff or profits to defendants, or when the plaintiff is unwilling to show approximate evidence of harm. Approximate actual damages when there was a plausible fair use or other defense. 2-3x actual damages/profits when reclkless or intentional or some other reprehensibility. 10x if highly willful. Wouldn’t need constitutional jurisprudence if we had these principles.

However, as things are, due process issues arise. When awards are grossly excessive, they violate due process under BMW v. Gore. Juries in particular may be insufficiently constrained, requiring de novo review. Don’t punish this defendant for the wrongs to other parties who are strangers to the litigation. Three guideposts: reprehensibility of defendant’s conduct—some are worse than others in copyright as in tort law. Higher ratios are ok when the conduct is more reprehensible. Disparity between the harm to the plaintiff and the punitive award—few awards exceeding a single digit ratio will satisfy due process. Ben Sheffner thinks this guidepost shouldn’t apply in copyright, but she thinks that approximating actual damages is generally possible. Third: comparison of award to ordinary civil penalties, which is a little more troublesome. Part of what we should look at is other awards for similar kinds of infringement.

Two district court cases have rejected the idea that BMW applies to statutory damages, but she disagrees. Part of the reason: courts increasingly talk about statutory damages for willful infringement as punitive, which makes due process review appropriate even when there are statutory caps.

We need statutory reform as well as judicial action. The problem is the melding of compensatory/moderately deterrent functions and the penal function. Canada has statutory damages, but only for ordinary infringement and only where actual damages are hard to prove, and then there’s a separate punitive award for exemplary purposes. Consider patent or TM-like awards of up to 3x damages as an alternative to statutory damages. The US is the only country with this whacked a statutory damages regime. Out of 120 WIPO countries, less than 20 have a statutory damages award scheme at all, and most of those were forced to do so by US FTA agreements.

Undue Process: Challenges for Rightsholders and Service Providers in Implementing Section 512’s Notice and Takedown Provisions
Jennifer Urban, Laura Quilter

Empirical perspective on the DMCA, which can give insight into other proposals, for example to modify §230 to provide DMCA-like procedures. For this paper, they looked at a random sample of 451 §512 notices from The Planet, a hosting service provider in Texas, out of a set of 6366 from 2004-2007. They coded the set for parties, dates, claims, DMCA section, etc.

The Google dataset from their first paper mostly concerned search, as you’d expect: people were using the takedown to get things out of search indexes—competitors trying to change search rank. The Planet was a more standard ISP at the heart of the DMCA’s concerns about hosting.

Providers of connectivity got a straight safe harbor from infringement under §512(a) in 1998. In 1999, Napster showed up. The DMCA presumed that infringing content was hosted, but the major copyright owners quickly grew concerned over p2p, which is a transitory network communication for the ISP subject to §512(a). So what happened? Connectivity providers say: large copyright holders responded by sending takedown notices anyway and telling the ISPs that they needed to terminate repeat infringers’ access entirely. Research question: is that happening? Answer: 21% of their dataset, an extra burden on ISPs.

As expected, the people who send 512(a) notices tend to be large copyright industry players. Over 90% sent by third parties—BayTSP and other p2p focused contractors/rights enforcement agencies, trade associations, and others. Tend to be small shops, but they send a ton of notices. Though courts haven’t bought this, copyright owners argue that ISPs are obligated to investigate and terminate users.

Third parties send a lot of notices overall—55% of the sample. Range of techniques. A lot are well done, but their investigations aren’t always accurate—identifying IP addresses that aren’t on the Planet’s network, etc.

Who sends 512(c) notices? The third parties; large industries—shows that there were still a bunch of allegedly infringing files being hosted, not just shared p2p, so 512(c) is providing a benefit to copyright owners.

Qualitatively, how are people dealing with this complex statute—is it really cheap and easy? Apparent: the complexity of the statute and the underlying law was a real challenge. The Planet’s information about 512 is pretty clear to a lawyer, but to a nonlawyer it’s not. 20% of notices were returned with “does not substantially comply” notes. That doesn’t include a substantive evaluation of whether the claim sounds in copyright or anything else, just the statutory information: did you identify the copyright owner, the location of the material, etc.? (16% of notices were bounced when you excluded one particularly troubled and active copyright agent.) So this doesn’t include the problematic §512(a) notices—combined, that’s a pretty good chunk of activity.

Example: someone trying very hard to comply with the statute, but can’t figure out what he’s doing wrong; The Planet’s standard reply is that the notice is deficient and the complainant should visit their website. He says he’s done so and asks, angrily/plaintively, what he’s done wrong, but they just refer him back to the website. The authors think he probably just gave up.

Some people who seemed to have real claims were just unable to comply—misunderstood the DMCA. People were really bewildered.

They also got anonymous tipsters and people complaining about non-copyright content (game cheats, trademark, product keys), or people invoking non-US law. Some senders may have been confused, but some may also have been using §512 more strategically. Why not try to fit things into copyright if it means you can easily get what you need? Examples: unauthorized product resellers (Sony claimed copyright in photos of the products); game cheats (game companies complaining about gold farming, which implicates TM and contract but not copyright).

Conclusion: 512(c) remains useful for a wide range of copyright holders, but the senders also have trouble complying/understanding, which causes various problems—chaff for the ISP. These and 512(a) notices impose a genuine cost on ISP, though it’s hard to gauge how problematic this is. The greater risk: to targets. Different ISPs have different reactions—what they do in response to 512(a) notices may vary, and the authors feel that disabling internet access is a disproportionate response. Latest attempts: more draconian, three-strikes law like that passed in France. More moves towards filtering, too.

We need to understand more about rights enforcement agents and trade associations, prominent senders with incentives to find infringement and show results.

Reaffirm other recommendations: mostly tweaks to build back due process, like don’t do a takedown until the target has a chance to respond. Modest calls to help copyright owners who are having trouble using the statute—the Copyright Office or someone else should create educational materials on using the statute appropriately. Part of what makes §512 hard to understand is that it’s private, so more transparency would also help. Notices should be public. Also suggest caution before replacing §230 with notice and takedown—we see so much chaff with the wheat, and notice/takedown lacks any substantive/judicial-type review.

My reaction: The paper argues that copyright is one of the few claims where ISPs are actually required to take action to avoid liability, since so much is protected by §230, so there’s an incentive to send DMCA claims regardless of whether the conduct at issue is actually copyright infringement. This doesn’t entirely explain why lawyers would use the DMCA for trademark claims, where secondary liability would apply. Lack of familiarity with IP law or the hope that the ISP will just go along, coupled with uncertainty over whether the ISP would actually be secondarily liable for trademark infringement under the governing standards, could explain more of why all IP claims are funneled into DMCA notices.

A: TM private ordering: ISPs are beginning to follow similar procedures to the DMCA—eBay has essentially notice and takedown for TM. Sometimes people send mixed claims—copyright and TM together. But the ISP procedure tends to be copyright-based.

Incentives to Lead, Follow, or Compete: Comparative National Choices of International Copyright
Michael Yuan, Roger Williams University

England led with the Statute of Anne in 1709; US followed with the Copyright Act in 1790. Berne Convention led in 1886; the US followed in 1976 and 1988. The EU led in 1993 by extending duration for 20 years; the US ultimately followed.

When do countries have the incentive to lead or follow? Is this system a good idea?

Results of paper: it is desirable to have a lead-follow model except when the economic life of information goods in the leading country’s market is very short. But this is an unlikely case because the leading country is unlikely to act. In the ordinary case, not everyone is likely to follow—small countries can improve welfare by not following and free-riding—and that can be bad.

Model: two countries, each with a creative industry that competes domestically and internationally. Each creator decides pricing and volume of market participation. Each market provides national treatment to foreign works. In the lead-follow market, a leading country sets policy to maximize local welfare, and expects that follower will peg its copyright policy on the leader. In the competitive model, by contrast, countries act separately to maximize individual national welfare. There may be a need to incentivize small countries to lead/follow and to induce countries with short economic lives for information products to follow.

Question for Samuelson: how do you measure damages?

A: It’s clear, in general, that when a defendant sells items, there are lost profits/sales—compensating plaintiff for lost sales/lost license fees makes sense. Defendant’s profits are more about deterring infringement, but also often has a compensatory effect. People who don’t promptly register can have trouble getting enough of an award to cover costs, but costs are also an available remedy. We have statutory damages to make a small lawsuit possible—e.g., a photographer whose usual reproduction fee is $150. $750 can make detecting infringement/bringing a suit worthwhile. The legislative history suggests that one purpose is to make it possible to sue, but that’s more evident in 1909 because ordinary copyright owners who don’t promptly file, and most don’t, are ineligible now—wrong design.

Q for Urban: How often does the recipient counternotify? How often then does the notice-giver file/not file suit?

A: For Google’s search notices, the ISP doesn’t have to and can’t notify the target (no service relationship); Google said it essentially never got counternotification. Quilter: ISPs say they don’t get many. Didn’t see any in the study. The Planet gave them all correspondence attached to the original notice, but not the correspondence to the alleged infringer. We think that if they had gotten counternotices, we might not have picked them up, but we should have seen at least some of them in correspondence with the sender. We think people don’t know they can counternotify, and they might just give up. Lawsuit: another unknown, but sense is that it’s uncommon.

Q: Are there risks to the counternotice that might not be present under §512(f)?

A: §512(f) allows suit against knowing and material misrepresentations in the chain; people may be dissuaded, but the standard for liability is so high that it’s unlikely that anyone who understood this would have their decisions affected. (Understanding, of course, is precisely the problem the authors have identified in the study. Laypeople are really scared to claim rights because of all the legal and lawsuit-based language that accompanies a notice; I’ve seen this with YouTube users.)

Kathy Strandberg for Yuan: In your model, the cost of creation is independent of the length of protection—so that doesn’t take into account the effect of bigger rights on cost of creation, right?

A: Right.

No comments: