AALS Section on IP Law

Liability for Intermediaries Under Copyright and Trademark Law

Peter Menell: Indirect liability is part of a larger system: law, technology and markets. The goals of the system: promote creativity, free speech, privacy. Markets take care of it when possible—ringtones, for example, where copyright hasn’t given way at all; a cellphone provider can control ringtones even though people won’t pay 1/3 of the price to iTunes for the same song.

Politics and norms can also be used to control behavior. A levy would solve current problems but create new ones, like freedom of speech. The benefit of tort type remedies is their dynamism, evolution. Recent cases have helped markets figure out what to do: look at video (Hulu), Google Book Search: creative solutions prodded by risks of liability.

Mark Bartholomew: He starts from trademark, which is based in common law. Question: why do we want Iago punished more severely than Othello? Mental state; facilitation of murder; agency/trust relationship betrayed. The cases are sometimes divorced from a philosophical grounding: when is it just to hold someone accountable for another’s misdeeds? Basic concerns are with vicarious (relational) and contributory (blame/fault plus contribution to bad act) liability. These standards are applied differently depending on the tort; it’s much easier to win secondary liability in copyright than TM. For vicarious liability, relationships that can create liability are construed more narrowly in TM. As for contributory, courts are hesitant to infer knowledge in TM. The Supreme Court has endorsed this. But why?

Mark Lemley: The flipside: justification for safe harbors and the oddity that the safe harbors differ across fields. Safe harbors are justified because the internet turns retail interactions into wholesale. Vicarious/contributory liability were developed with the assumption of particular individualized acts. The internet involves billions of automated transactions among people the alleged contributory infringer never meets. Running the internet should not lead to liability, and safe harbors allow the internet to survive these doctrines. That said, by accident there’s a bizarre patchwork of safe harbors. If the case doesn’t involve IP/child pornography/housing discrimination, you’re absolutely immune under 230, even if you’re aware of and encouraging illegality, regardless of whether there’s an alternative identifiable defendant.

Then there’s the DMCA for copyright for those entities that existed in 1998 that Congress knew about. Viacom case: Viacom alleges you don’t get the safe harbor unless you are not secondarily liable, which is actually not an impossible reading of the statute. The DMCA is also vague about things like what a repeat infringer is, so YouTube errs on the side of banning users even when their content is unquestionably fair use but draws repeated (wrong) complaints.

There are statutory safe harbors for electronic information in TM, though very few people know this. No money damages; injunction only if it wouldn’t affect the operation of the intermediary’s business. For other IP, no safe harbors (patent, right of publicity in First Circuit).

This regime is nonsense. There is no good economic argument for different safe harbors for different regimes. Also it misleads people: ordinary folks tend to assume the DMCA applies to everything. Leads to strategic gaming, where people convert TM/defamation claims to copyright and abuse notice and takedown. It’s not a model other countries can follow. This is bad because the rest of the world doesn’t get the internet; they’ll hold ISPs liable for having unpopular speech or copyrighted materials on their systems. They’ll even criminally prosecute ISPs. For the US to have a moral position, we need to get our own house in order.

Lemley likes the TM model: less absolute than 230.

Me: I’m very interested in eBay v. Tiffany as an indicator of the way the wheels are coming off. The court says multiple times that it’s not engaging in cost-benefit analysis, but then pretty much does. It says that eBay isn’t required to have VeRo, but then uses VeRo to explain why eBay isn’t liable. Like many courts, as Barton Beebe has shown, it ultimately cares a lot about intent. Intent is probably here to stay. But in the past, good faith and efficacy were pretty tightly linked: massive infringement was a good sign that you were operating in bad faith, and bad faith has been considered a good sign that infringement is likely to take place. eBay marks the moment that the court has to decide which matters—if eBay is proceeding in good faith, but there’s still a lot of counterfeiting, which wins? Answer: good faith, filtered through some idea of best practices. But because we’re confused about things like what good faith means, it’s not explicitly on the table that this depends on a judgment that, if eBay can’t operate without a high level of counterfeiting of certain items, then the social benefit of eBay trumps the harm of counterfeiting.

Fred Yen: The conflict between cost-benefit analysis and intent/culpability—when the two diverge, which do we pick?

Bartholomew: The trend is towards more focus on fault, less on instrumentalism. eBay explicitly rejected a cheapest cost avoider standard. Grokster emphasized mental state. Vicarious liability is different, and doesn’t worry about blame. Rather, it’s imposed instrumentally because of some social value. Vicarious liability is becoming less important in TM—few successes in recent years—B thinks this is good. It’s dangerous for courts to handicap tech futures; courts are good at analyzing fault/blame.

(As Mark Lemley was about to point out, I think this elides knowledge/intent: what counts as bad faith? Knowledge alone can be use to infer fault.)

Lemley: Disagrees fundamentally with the conclusion. Courts are going to fault, but that’s wrong. Bad intent is a concept that makes no sense divorced from a view of what we want to prohibit: intent to do what? We use intent to substitute for a principle of legal fault. Infinitely malleable: everybody’s got an intent to free ride. Look at Grokster: the Court says there are three components to its bad intent: (1) Grokster hoped to make money (but a footnote says that’s not enough); (2) it didn’t filter (but a footnote says that’s not enough); (3) it internally considered, but on advice of counsel did not run, an ad touting the availability of infringing material. The exact same company with the same business plan, but initially better advised, wouldn’t be subject to liability (comment: if you believe the Court—maybe Grokster was going down one way or another—but that still proves that the current law’s a mess). That makes no sense. We can ban the business model or accept it, but subjective state of mind and effects are different. (Comment: again, I think that tech may have separated state of mind and effects further than in the past.)

Menell: Grokster is braindead because the Court was so concerned about avoiding Sony. Safe harbors are a political and democratic way of dealing with tech, but legislatures are not good at ongoing oversight. Different safe harbors may be appropriate for different IP regimes; patent’s staple article doctrine deals with patent-specific antitrust concerns.

As for eBay, VeRo seems pretty good, different from Napster. We want ISPs to try hard: this seems to be what’s going on in the recent Veoh case. eBay and Veoh are going beyond the call of duty, so courts are pragmatically evaluating good behavior even in the safe harbor context.

Me: But that’s scary! It means the safe harbor isn’t safe. VeRo and video screening become the new baseline.

On Lemley’s point: I agree that “intent to do what?” is the key question, but I believe it could actually be answered better. If we did answer that in a way other than “free ride,” which I agree is a terrible answer, then we could manage a fault-based standard. Aimster’s cost-benefit analysis isn’t necessarily better, because courts aren’t all that good at it.

Lemley: You can’t run an internet business without reckless indifference to the possibility that someone’s rights are being violated somewhere. So you need some cost-benefit analysis to figure out what an acceptable state of mind would be.

The panel’s consensus seemed to be that doing the analysis in big chunks, not granular, as in Aimster, is probably the best way to protect innovation/future tech.

Menell: Punt to institutional choice. Who will be making the cost benefit analysis? In tort courts assess technology all the time. (Comment: I see a connection here to preemption arguments increasingly made by defendants in cases where the FDA, FDIC, etc. have regulatory authority over the subject matter.)

The challenge now is layering safe harbors on top of the common law. YouTube is trying to avoid a safe harbor loss and thus overscreens. Third party organizations like the UDRP might do human screening. Effective screening leaves Viacom happy. The devil is in the details.

Lemley says this is expensive, but big rewards require big risks. You’ll need a significant legal staff. This is just a cost of doing business, and Google has the resources for it.

(Comment: Holy open source catastrophe, Batman! Bram Cohen does not have a significant legal staff, so bye-bye Bittorrent and whatever might have come after.)

We should also change the measure of damages so the risk is lower. Every company needs to find a balance with its lawyers.

Bartholomew: Intent in criminal law is a very high standard. We don’t want that in secondary liability; it’s impractical/manipulable. Nimmer gives a quick checklist of how to avoid Grokster liability. But we can still pay attention to knowledge, objectively. The more we pin down what direct infringement is, the easier it will be to decide secondary liability.

In criminal law, the intent standard requires the purpose of causing the commission of the crime, but there’s also a very low contribution standard: any act that helps/approves, like applauding or babysitting for the criminal. Bartholomew proposes using knowledge with a higher material contribution standard. Courts haven’t looked enough at causation.

Sharon Sandeen: There’s an existing model in trade secret in the UTSA: third parties are liable based on actual knowledge/reason to know of infringement. The nature of the IP affects what people can know. People are less likely to know trade secret status versus patent status, which they can search.

Lemley: Outside IP, it’s considered bizarre that knowledge of the legal status of the property is an element; intent to do an act is usually key.

Bartholomew: Likes the idea of continuum, depending on tech. An exact reproduction has a more concrete relationship to knowledge, but if you rely on remix contributions, the situation is different. (Comment: and which is YouTube?)

Kathy Strandburg: Is this one of the internet differences? We’re clear on what we don’t want elsewhere, but we are less sure about the underlying harm online.

Me: I think the internet makes more salient that we don’t always know what the harm is. It is simply more credible to believe that people would copy mp3s for free that they’d never, ever buy—a zero price is different in its effects on behavior from a pirate’s lower-than-legit price—so we can debate the magnitude of the harm. Likewise, the dirty little secret of eBay was that a big reason that a lot of “counterfeiting” was left over even after VeRo was that the customers were perfectly happy to get “Tiffany” jewelry on the cheap. Tiffany says, and the law agrees, that this is a harm, but there’s a big group of people who disagree.

Lemley: this is a version of the retail/wholesale problem.

Menell: Cablevision tried to separate the two—the copyright owners claimed they didn’t want to investigate whether it was fair use for individuals to use the remote recording service. But that’s part of copyright’s path dependence. O’Connor wrote in a memo for Sony that timeshifting wouldn’t be fair use if she thought people were skipping commercials. Menell thinks this is sensible, because that’s how we funded the business model. But now DVRs are morally acceptable. We’ve lost the ability to see the direct relation between institutions and creativity. Maybe we should tax DVRs.

Lydia Loren: She worries about raising the bar. If we talk about weighing Google’s screening in its favor, that creates barriers to entry: are we sure we want to impose them? Menell’s comfortable with pretty high barriers.

Lemley: Yes. In theory, if entry imposes social costs, they should be internalized, but people tend to overstate costs and understate benefits of new tech/markets. This is why safe harbors are better than Posner evaluating Aimster.

Menell: Is more agnostic. He can imagine very different institutional arrangements with ongoing evaluation of tech. Google Book Search creates an extreme barrier to entry, but it’s doing quickly what ASCAP took 20-30 years to do, and ASCAP created a lot of wealth. Copyright maybe worked better in an earlier era.

Markets surprise you—see ringtones. Guitar Hero/Rockband: kids pirate like crazy, but pay $3/4 to play along with a song.

Mark McKenna: DVR moves us from ads to product placement. It’s a question of whether content owners ought to have to adapt. Usually without force they won’t. TV won’t go away; it will change.

Menell: We’re creating distance between creator and consumer because of artificial price support with product placement. That’s not the art of a democratic society.

Me: At this point I have to say something: that’s not the only way you get creative content! There’s lots of creativity going on now, enabled by the very technologies Menell wants to shut down, and it’s odd to hear him talk about these technologies as anti-democratic.

Menell: Copyright and patent are about spurring investment. Some types of content are best supported by copyright—this was true of music until the last decade and it’s still true of movies (comment: for how long?). Professional creators are still the main influences on our kids and in culture. A guaranteed revenue stream is a way to get quality. (Comment: please define quality! I’m actually pretty sympathetic to this argument—I love plenty of Hollywood product—but that’s a preference, and not one unshaped by my surrounding culture; it is not an absolute measure of quality.)

Yen: Should there be uniformity across torts?

Bartholomew: Based on a conception of fault: is it worse to infringe a trademark or a copyright? Neither—we should take care of differences at the remedial stage. Vicarious liability is different: we should ask when a social problem exists such that we need to expand liability.

Me: Eugene Volokh makes an interesting point: there are variations in the knowledge required for underlying torts, such as whether the original speaker knew (or was recklessly indifferent) about falsity of a defamatory statement. The ISP is going to have a real problem assessing some of these, so the pressure will be to over-remove even if the safe harbor is supposed to be pretty nuanced.

Lemley: Vicarious liability has definitely expanded in copyright over the past two decades. Direct financial benefit has become indirect, or even a hope of future profit, and the right and ability to control has come to mean the ability to shut down the entire system. This is a big move from master/servant.

Uniform statutory safe harbors are different from the underlying standard. We should definitely ask what the importance is of being able to sue the secondary transmitter. If you can find the defamer, the need to sue the ISP may be reduced. Also, ease of implementation is an issue that needs working through.

McKenna: Courts have paid no attention to contributory versus vicarious; they’ve bled into each other. There is a danger of siloing IP from other torts: we’ve taken doctrines from tort wihtout their nuances.

Lemley: As a patent person, he’s noticed that when a case makes it to the Supreme Court, the fact that there’s a specialized rule for IP doesn’t seem to mean much.

Bartholomew: the expansion of vicarious liability in copyright is bad and was done in ignorance of its pedigree in tort law.

Lisa Ramsey: Copyright works may also have TM protection: this is an argument for uniformity.

Strandburg: Uniformity in IP might be the wrong question. Where do we need professional involvement? That might lead to nonuniform rules.

Lemley: If we build it, he worries we’ll reify existing categories. In 1995, we would have thought that the recording industry required professionals.

Me: I would take that in a different direction: eBay is subject to different employment laws, etc. than My Auction Site and its one employee. We might consider that, or we might reject that because one person can do so much more damage on the internet.