TMSR Session 3: Private Actors…and their Machines

Introduction: Jeanne Fromer

Private actors pursue their own interests. Focused on Amazon: free riding on gov’t mechanisms, particularly TM law, to communicate to gov’t not to regulate it—product liability, intermediary liability. It’s also cheaper, and Amazon is notoriously cheap. Affecting TM law (as set forth in new paper with Mark McKenna, which is excellent). Can enshrine things like notice & takedown in © law, but that has effects on the ability to assert fair use. When things are happening en masse at a business, just as in the TM system, w/o individualized attention and with a different agenda than the law’s, you can see the law’s protective features be minimized. So big theme: dealing with masses of marks—eBay, domain names are another example—requires new thinking. You have to be wary of private actors’ incentives.

TMs are a proxy for all sorts of info and goodwill, but there are other things that are proxies for that and have always been—selling in a certain marketplace or mall gives information; being the first listing in the Yellow Pages (might be the first nonsense marks!). But we’re seeing more and different proxies for information and goodwill; started w/search engines but seems more prominent now. Amazon is used for more product search than Google. Product reviews: different info is available. AI modeling is also imagining what consumers want to see. We consider that TMs have costs as well as benefits; we should think about the costs as well as benefits of these proxies. TMs may not matter as much with these other proxies, but they’re also noisy proxies. (This reminds me of a classic Posner assertion where he claimed that Trisha Waldron’s photograph, or the location of her stores, would give people information about whether she was Native American.)  People will game them in all sorts of ways that dilute their information, whether that’s product review manipulation or otherwise.

People trust Amazon, not the individual “marks” on the site, including their ability to return products easily; TM’s work is done in the service of Amazon. A variant of separating the mark from the brand; not seeing it in all categories, and we may have a world in which some marks—maybe the 1%--are not divided from brands but the others are. Not that surprising that the luxury marks are not selling on these platforms and are often fighting them; luxury may require keeping the mark with the brand.

Human desire for TMs? We like how they sort us/how we get to identify with them? Even if we live in a world with other sources of info, maybe we still like TMs. How much of that is cultural?

Eric Goldman

SAD scheme and what it tells us: Efficacy/brittleness of our institutions—the real action is below the waterline, with implications for everything else we’re doing. Sealed complaint; ex parte TRO; used to freeze entire accounts and money. TRO is better than a notice & takedown—you get the seller’s entire account down and you get all their money; marketplaces are treating them as risk of being held liable for infringement by that merchant. Gives rights claimants things they couldn’t get from notice & takedown; will only increase until we fix it.

Standard notice doesn’t produce $, but if you can freeze $, they’ll pay you to get it back. Entire law firms tell rightsowners: let me have the right to enforce and I’ll cut you a check every month. Luke Combs outsourced his enforcement; he didn’t say that he was going to stop using the scheme, even though he tried to make the one person who got attention whole.

TRO freeze is usually left in place even after the TRO expires. We can start w/good faith presumption that the court thinks it’s doing something temporary, but it’s not given the nature of the TRO and the marketplace. Mistakes are common in the process and appeals are almost nonexistent. Thus there is little scrutiny and little pushback.

Main players have dubious rights—emojis, smileys—already at the border of what TM should protect. They shouldn’t have the power that marketplaces give them.

Increasingly seeing competitors using SAD scheme to knock them out during peak sales periods. TROs harm consumers.

Lessons for institutional thinking: The law of TM on the streets is very different from what it looks like in appellate cases. Ex parte: the law is whatever the judge will sign. Civil procedure critiques—basic joinder, jurisdiction questions where the rules are clear but are simply not matching what is taking place.

Third-party enumerated rules are also law on the street. Actual adjudication by services is what the law “really” is. Google’s TM policy for keyword ads is law for 99% of all TM disputes out there.

SAD scheme is swallowing up the rest of TM law. About 1000 SAD scheme cases are filed in the past year with 200 defendants each, that’s 200,000 defendants versus about 3000 traditional cases in the past year. It’s 99% of TM law and that will only grow as long as the scheme continues to work.

TM is a terrible subject matter for ex parte proceedings. When the defense doesn’t show up, it’s too error-prone. It’s like with CPB. Riddled with errors. Adjudicators need to hear about risks of overenforcement, overclaiming; defenses—he gets emails from sellers of used or grey market goods snared in a robotic search. The judge isn’t hearing that they’re not counterfeiters.

Other examples: design patent, as Sarah Burstein has written. Putting pictures side by side often isn’t enough. Trade secret: its scope is the whole ball game.

Because of the lower stakes, secretiveness, lack of access to court that Ds have, these cases just don’t get corrected.

Research project: Go and look at ex parte proceedings and whether there were subsequent proceedings and whether the court changed its mind—we could literally identify the error rate.

This needs to be fixed, but next steps are not clear. We haven’t found a good way to intervene in the cases. Even when judges blast specific plaintiffs, they still aren’t taking the corrective action—they aren’t applying their reasoning to other cases or other Ds, they aren’t imposing sufficient penalties to deter.

Mary LaFrance

Focusing on counterfeits doesn’t address other forms of infringement. Birkenstock & Nike left Amazon b/c of counterfeit concerns; that was Amazon’s focus. But private systems have limited ability to process context. High risk of false positives as well as false negatives. Focusing on counterfeiting doesn’t protect consumers from other types of non-genuine goods, which exist on Amazon & eBay—gray goods, expired products, repackaged, mishandled in supply chain. SHOP SAFE wouldn’t change that b/c it is only about counterfeits. We know that automated systems designed to flag infringements get lots of false positives b/c they can’t recognize fair use; may also be false negatives, though that’s not the focus. There probably aren’t other major complications than fair use for © screening, though they can also miss licensed uses; but it’s much harder for TM to determine infringement.

Nongovernmental mechanisms that work well: Dave Fagundes on derby girl names, using norms. But the context is really simple: 1 name, 1 skater.

UDRP: context gets considered.

But you need a full LOC analysis if you really want to ID infringement; instead, automated systems aim for counterfeits. False positives: probably miss descriptive fair use. Platforms may not have incentives to help sellers who are wrongly identified get back online; not willing to face potential contributory liability. Platforms don’t disclose how many invalid notices they get.

Brands can choose to “gate” on Amazon with authorization letters, but how does the consumer know the seller will stick to selling genuine goods? Brick & mortar sellers have more protections in their supply chains as a rule—relations w/distributors over time, though there are exceptions with discount retailers. [I don’t think this is describing the situation of small brick & mortar sellers.] Their potential liability for counterfeits gives them more incentives to be careful, whereas ecommerce platforms don’t face that potential now with third party sellers.

SHOP SAFE would impose contributory liability with [RT: impossible to comply with] safe harbors. Would lead to more false positives. Seller could appeal, but there’s no requirement that be expeditious. And proof that goods aren’t counterfeit could be challenging, b/c there’s no way to have Amazon actually inspect the products; there’s also no requirement that the TM owner cooperate with the process, so even if the seller provides product codes, the TM owner doesn’t have to verify them even if they’re accurate.  SHOP SAFE would also presage a move to require filtering for © to replace the DMCA.

Stacey Dogan: role of human agency in private actions in an increasingly AI-driven world. We think of TMs as informational vessels that allow consumers to make decisions for themselves based on what they’ve learned either personally or through other means about the goods bearing the mark. But as we move increasingly to algorithmically chosen or differentiated tools to be used by third parties to shape our preferences, that notion of TMs as having info value to humans who then exercise their agency and make choices go away. Private actors will act in profit-maximizing ways; TM owners have always done this, and intermediaries have existed, but when they’re doing all this informational intermediation for their own economic incentives, that’s very troubling.

Jeremy Sheff: we have one to one disputes; one to many where a TM owner tries to enforce against a lot of counterfeiters; then many to many disputes like low quality marks flooding the system in ways that raise costs for everyone. Crystals and mud story about this: in many to many, an attractive case for more crystalline, bright line, formalist rules for cancelling registrations. One to one is more attractive for standards. One to many: hard to figure out whether to worry more about over or underenforcement. If we were to try to adapt different rules, we have to ID what category we’re in, and how someone might be allowed to contest their categorization. DMCA has a formalist process: if you flip the switch three times, you end up in federal court. [You’re supposed to do that, but platform policies don’t work that way—they often don’t restore struck material—and neither do infringers, who may just reupload without filing a counternotice.]

Goldman: that assumes that someone can access court enough to push back.

Sheff: yes. If everything’s done by platform w/no recourse, there’s no way to reverse that.

Goldman: the rule of law is in place but is inaccessible.

Sheff: DMCA says counternotify and you’ll end up in federal court at the end of the day; that might be too expensive, but that’s the cost of muddy standards. Nothing like that for TM.

Lemley: Prior rounds of trolling, something that helped: targeting the lawyers for sanctions or even state bar work. Are these the sorts of representations a lawyer can ethically make to a court? State bars don’t have standing requirements and you don’t have to litigate in the case itself. The appetite for doing it might dry up if there are disciplinary consequences for lawyers about joinder, jurisdiction, etc.

RT: On LaFrance’s points about gray, expired goods, etc: I’m pretty sure TM owners are actually happy to argue that all those things qualify as counterfeits. I also don’t think used, grey market, expired/near expired, repackaged goods, mishandled goods are infringing or counterfeit; the problem if any is that they’re falsely advertised.

Fagundes wrote follow-up article suggesting that system was breaking down. Revisiting Roller Derby’s Master Roster, in CREATIVITY WITHOUT LAW (Kate Darling & Adam Perzanowski eds., NYU Press, 2016).

Marketa Trimble: A little worried about following up TRO success rates—in the trade show context, should not compare to online case results. In the trade show context, there’s a lot of information that preexists, so there is some relationship between the two parties. Judges are pretty strict on issuing bonds in those cases. And also there is a robust world of social norms surrounding Ps in these cases—they do have offline reputation to protect; they’re repeat players. Online plaintiffs are less constrained to care about their reputation among sellers.

Even in trade show cases, we were shocked how little the judges knew about the problem of bad TROs. What helped was meeting judges from other countries, so they weren’t just hearing from local lawyers about how the world worked. Protective letters in other countries: ability to alert court in advance about a potential problem with a case. Might seem burdensome but courts love it (they collect fees for it).

Mark McKenna:

Important to recognize that the quality of online shopping has gotten really bad. Enshittification is real. Labeling has gotten much more confusing. Amazon is also decentering brands because you can’t necessarily trust that you’ll get the right thing just by ordering it by name. The motivation to respond to counterfeiting is increasingly understandable even if we have skepticism about methods. What’s the best way to do that? Schedule A was an invisible problem, but why is that happening at the same time as platforms are creating dispute resolution policies? Maybe this is all really about lawyers and not brands. But it’s worth thinking about which levers are being pulled on brand value/what is depressing the informational value of TMs.

Linford: externalization of costs—Amazon just tells you to throw the product away if it’s not what you ordered. Free shipping makes it easy to start & stay with Amazon.

One response is to get off the platform. We’ve externalized to consumers the cost of protecting themselves—reading reviews.

Another cost of AI specifically: energy. Generating one image cost as much as one charge of your phone. Should that weigh into the fair use analysis? [No, for so many reasons, including that the alternative to fair use is licensing.]

Doesn’t handing over right to sue to lawyers create a problem like for Righthaven in ©? Some sort of naked licensing/transfer in gross?

Lisa Ramsey: this might drive us not to grant TMs in inherently valuable terms like LIFEGUARD for clothing.

Could algorithmic practices affect what counts as TM use? Brand name, product source (manufacturer, third party, etc.), description all show up on Amazon. If Amazon adopted a policy of taking down content only when it appeared in brand name or product source and not in description, that could affect how courts think about it too. User names on social media may or may not include trademarks; user bios give more information.

Bill McGeveran: Enshittification is about size, size is about antitrust: you can enshittify when you have market power. Scale has various manifestations.

Burrell: Monopsony is a better description b/c the economic evidence is that lower market concentration is necessary to have a monopsony.

McGeveran: that’s definitely Wal-Mart. Links to larger conversation about antitrust. You can’t tell a full story about causes or potential responses without looping in antitrust concerns.

There are so many different harms, including monopsony, that are getting blurred together. Useful to pull them apart. Having a listing taken down is not the same thing as being kicked off a platform entirely and losing your reputation accrued on the platform [and having your account seized including the money]. Some enshittification is nuisance to consumers, but it’s not the same thing as getting expired medicine that Amazon wouldn’t replace quickly.

Grinvald: Amazon’s disregard of supply chain is part of what’s eroding connection between product and source—signaling to purchaser that it doesn’t matter who the seller is.

Note that shifting to Target by mail won’t help b/c Target/Wal-Mart has also allowed third-party sellers to ship from their site. That’s the Amazon effect!

Discussion of various attempts to game liability for selling defective products—if Amazon is advertising, storing the product in its warehouses and shipping it, why should it matter whether it’s a “third party seller” fulfilled by Amazon?

Mid-Point Discussants: Lisa Ramsey

The problem of anticompetitive takedowns: DMCA, UDRP. Parody accounts: use noncommercial speech s a bright line, whether as defense or otherwise? If we had bright lines in the statute, maybe platforms would be more willing to keep speech up when challenged.

Marketa Trimble

UDRP as potential model for alternative dispute resolution/shaping rules that can be skewed by private ordering. UDRP is fast, cheap, doesn’t require legal representation. But also has flaws that we can learn from. Flattened the TM world. Designed for alphanumerical strings; treats TMs as word marks and ignores other types of TMs. Ignores classes of goods and services, which is incredible. It also ignores national law. Drafted w/the assumption that there is something beyond the written words that can be drawn from to interpret many of its concepts: what is a TM? Is registration required? What is substantial similarity? What is fair use? Many things were left unexplained, and there was no choice of law provision. The idea that you create a new body of law that will on its own create all this case law and interpretation to fill in the blanks is rather naïve. Common law takes centuries. So naturally they started w/national law, and given the dominance of US, Australian, UK panelists, we saw a lot of that influence. Might work well for the US, but might have problems scaling internationally. The panels may give victory to one who had a mark anywhere in the world first.

Need appellate process. The alternative is court—so how do you treat the UDRP in courts? It’s possible that courts turn to their own national laws, as we saw in the Barcelona.com case and others. Filling in the blanks is a big challenge.

Despite all the changes in the world, there is still a lot of reliance on TMs. [Tide is not a “brand” in the pumped-up way that some now are; its primary function is to guarantee and make you feel good about your detergent, not anything else.]

Note that UK IPO won’t register a mark in all non Latin characters on the theory that no British person would be able to understand/remember it—compare to “nonsense marks.”

Just as we needed AirBNB to understand why a hotel is a good idea, Uber to see what taxis are good for, we may need algorithmic recommendations to see what TMs are good for.

Linford: There’s been a presumption that Amazon isn’t working right now for buying a brand I want to buy b/c search results are not reliable. If I put Tide in I don’t necessarily get Tide out. If I’m uncertain about whether I’ll get Tide and how long the search will continue to do that, that’s a problem. MTM v. Amazon bothers him: would like Amazon to do the one thing of telling him “we don’t have MTM.” Danger is that intervention has problematic downstream effects.

We are talking about these brands as if they once had a guarantee—Philip Morris has bought brands that I may have strong affection for, but I don’t have strong affection for Philip Morris. We no longer have strong rules against selling marks. But does that mean we presume too much consistency from brand sellers? If that was already false, the Amazon problem is less acute/might get us to correct our priors in the proper direction.

Lemley: MTM is a Rorschach test to see if people’s attitudes have shifted to see how we’ve been affected by the techlash. Would it be great if you could automate communicating only and all correct information at scale? Sure. But if you just give liability for not doing so, we are likely to do damage, and end up with sites that warn “we don’t have what you want” followed by a list of things you want. That doesn’t solve the problem of whether Amazon is reliable. But MTM is a case of Amazon being perfectly reliable: here are the watches we have available to you, the closest to what you were looking for. A standard higher than that is trouble.

Enshittification is not “there’s stuff on Amazon whose provenance I don’t know.” It is that Amazon got dominant by giving you the products you wanted; once it achieves that position, it has a strong incentive to sell the space between you and the stuff you want. Accurately identifying third party alternatives is not a problem, but making it hard for me to find what I want through the clutter of advertising is a problem. Amazon makes more profit selling ads than it does selling products, which is not great.

Dinwoodie: Lush won against Amazon in the UK, had to label “we don’t have it.”