Tulane IP conference, part 3

Michael Grynberg, Things are Worse than We Think: Trademark Defenses in a Formalistic Age

Underlying premise: more TM defenses would be good. But courts have less leeway than you might think. People who don’t do TM say that the Lanham Act specifies defenses and you can’t just add them. People who do TM refer to TM’s rich common-law history and point out that courts make stuff up all the time (e.g., DMCA elements of circumvention like “assisting somehow with infringement”).

Congress allowed courts to run with ambiguous language in the Lanham Act and amended it to endorse broad interpretations of TM rights. Meanwhile, formalism has become dominant as a matter of judicial, especially Supreme Court, interpretation. Moseley and KP Permanent relied on the statutory language; Dastar went about defining “origin” with reference to the statute.

Well, what about Wal-Mart? Isn’t that the epitome of results-oriented reasoning? This is a harder case. But it can be fit into a formalist narrative as a contextualist ruling: modern textualists are no longer strict constructionists; it’s always appropriate to look to a broader legal context. They derive a rule based on the surrounding context of §43(a), from the treatment of non-inherently distinctive marks. Then they make a judgment call, after having taken the text as far as they can.

Incontestability defenses are largely closed; Congress has ratified that understanding by amending §33(b) to add defenses.

More interesting: §43(a) defenses. As a matter of practice, new defenses are occasionally recognized. But how? When courts apply interstitial reasoning, they borrow from state law or use principles of general law. You rarely see federal courts just making stuff up.

So what’s the result? Definitional moves—nominative fair use is about uses that are definitionally unlikely to cause confusion; product designs are about uses that are presumptively unlikely to identify source. It’s largely a move to work within the likelihood of confusion standard.

But then the 3rd Circuit gets nominative fair use in Century 21, where they say that nominative fair use is a true defense. But on what ground did they say that? There’s no basis for them to say that it is a true defense. The Third Circuit took conduct that, in its terms, was statutorily enjoinable but refused to find liability. That’s the difficulty with any attempt to create a new affirmative defense.

He likes materiality as a possible defense. There might be no Article III standing if there’s no injury in fact from materiality.

Inherent problem with his approach: it’s very hard to get around a materiality claim on pleadings or summary judgment. You can imagine that good lawyers can construct arguments that the confusion does matter (the claim would be that a use harms brand value—e.g., Balducci, where the court credits consumer survey evidence stating that consumers might be less likely to purchase Michelob). Plaintiffs can get around defensive doctrines based on likelihood of confusion.

Rebecca Tushnet, Running the Gamut from A to B: Federal Trademark and False Advertising Law

I gave a bit of a retread of an earlier presentation. I added in some discussion of what TM should learn from false advertising, in particular that implications matter. In nominative fair use and other situations, courts and commentators often argue that they are making an empirical judgment: consumers can’t be confused by a truthful statement of the relationship between the parties.

This is a conceptual mistake, neglecting implicature. Example from Richard Craswell: you ask me for the nearest gas station. I give you an address. My answer implies, because we assume that I’m being truthful, complete, and helpful, that (I believe that) the station there is actually open, because unless you have specified otherwise I ought to infer that you would like to buy some gas for your car.

There are good reasons to deny the relevance of some implications, especially when they will not be received by or important to many members of the audience, but it is analytically insufficient to stop at the dictionary meanings of a string of words. False advertising law says that clever use of innuendo doesn’t protect a false advertiser; we protect consumers precisely when they need it most, when it would require a lot of effort to discern that what appears to be promised isn’t. Examples: save “up to 85%” on home heating when that would basically only be true if your house had big holes knocked in the side. Mylanta Night Time Strength, which didn’t explicitly say that it lasted all night but sold millions of bottles because that’s how people understood it; etc.

We can endorse nominative fair use and other defenses, but not because of the formalist divide between explicit and implicit meaning. We should think of them in terms of policy objectives and because of error costs: certain kinds of implications are so unlikely, and so outweighed by the value of other explicit or implicit information conveyed by the same statement, that we should just have a rule allowing such statements and not conduct an individualized inquiry into confusion.

Eric Goldman, Economics of Reputational Information (Oversupply Problems)

Reputational information is info about an actor’s past performance that helps predict future performance: unmediated (recommendation letters, student evaluations, word of mouth) and mediated (credit scores, GPAs, investment ratings, consumer ratings on Amazon). We’re awash in reputational info, and the trend line is up.

Unmediated reputation is idiosyncratic: my taste in movies will affect how likely it is I’ll like a particular movie; it’s often based on small data sets (I’ve seen a relatively small number of movies); it has high transaction costs to find and transmit. And it’s hard to police against shills. There are few boundaries for how we write recommendation letters. Most job reference letters are negotiated with employees; we don’t know what deal underlies each one.

Therefore, there’s a trend for mediated reputation, which solves a number of these problems. A mediator can aggregate opinions and individual ones will average out. Editorial filtering can manage the information glut. And the mediator may have incentives to manage the database and engage in data-correction. (I’m not sure Goldman is right about this—it might not be worth it to sort it out if so much of the pile is okay. In fact, that’s often the premise of aggregation, see Everything is Miscellaneous. Google does not go back and fix badly scanned pages from its Book Project.)

Problems with mediated reputation: People may misinterpret data as possessing faux precision/may ignore the margin of error. Outputs may be insufficiently granular, one-size-fits-all. US News & World Report scores offer generic rankings, but they don’t help determine what’s right for a particular student. We have no idea what’s under the hood—the editorial filtering/algorithm may be opaque: Google’s PageRank; credit score calculations. Data sources may not be credible, or credibility may be hard to assess, as when users are shills. People try to game known algorithms, like people merging their credit reports with others with higher scores. People attack review systems, e.g. Spore on Amazon. There’s a risk the mediator may move to pay-for-play, selling more favorable treatment. And error-correction can distort the database.

There are wildly divergent regulatory approaches: sometimes people can complain and remove information (credit reports) and sometimes there’s no way to force a mediator to act (§230). If errors are easy to report, negative information will be wrongly purged; if they’re hard to remove, then negative information will wrongly remain.

Which of these weaknesses can be left to market solutions, and which deserve regulatory intervention? Goldman is struck by regulatory divergence, and asks whether we can learn from differences. His bias is always the market, because of incentives to correct bad data.

Bill McGeveran, Endorsement, Identity, and Social Marketing

Background: word of mouth is the holy grail of marketing. Social networks are big and exciting. They still don’t make any money, though, and nobody knows how they will make the expected giant buckets.

Thus, the introduction of Facebook ads, particularly Beacon. People didn’t like the disclosure; it’s creepy and even seriously invasive for friends to learn about your purchasing/rating information. A guy whose significant other found out via Facebook feed that he’d bought a diamond ring on Overstock is the lead plaintiff in one of two class action suits. But these suits are hard to plead.

Information quality is a consideration: we don’t know why people buy products (for themselves, for others). Also spamification: each individual message may have an impact, but an ocean compromises the usefulness of all such marketing messages and even swamps the good stuff from the wealth of networks. Third, a Warren & Brandeis discomfort over loss of control. But American privacy law is so focused on sensitive information that mostly it doesn’t help here.

What about TM/false advertising claims? There’s an advertising message of endorsement. But that’s also hard to sustain. Most people have no TM rights. Consumer protection/regulatory responses might be appropriate. What about right of publicity? Seems like there’s a claim. The chief privacy officer of Facebook disagrees; and neither pending lawsuit makes an appropriation claim. Why not? Because Facebook argued consent. (I don’t think that would help with NY law, at least, and I’d think a NY class would be appropriate.)

Within 5-6 weeks, Facebook moved to opt-in: “tell your friends” option.

Without a European data privacy model, privacy won’t help. But publicity rights might be useful with a careful enough understanding of consent. We do want “tell a friend” to be available.

Counterarguments against regulation: is it too much propertization? Are there free speech objections? It’s commercial speech, so that shouldn’t be a problem. (Comment: I am dubious. If it’s truthful speech, it has a lot of protection unless we appeal to concepts of property; privacy isn’t going to cut it these days.)

Bernstein: Beacon was visible, which allowed the market to react. But regulation may be more important when the effects don’t slap you in the face.

McGeveran: agrees, but there are many things that Facebook could do that would be less transparent.

Comment for Goldman: there’s also an adverse selection problem: people who care lots may be more likely to comment (whether lovers or haters). With recommendation letters, it may be that nice professors get more recommendation requests.

Goldman: Statistics show that love/hate are more common than the nuanced middle, but it’s not clear whether love is a better or worse spur than hate.

Q for Goldman: How do you count individual blogs, etc.—is that mediated/unmediated? They don’t necessarily expect people they don’t know to be interested.

Goldman: Blogs are a great example of where a search engine may act as a mediator, but for other people the interaction may be unmediated (subscribers). Facebook and MySpace are aggressive editors of their sites, like credit reporting agencies manage their databases. They may allow lots of choices but they’re not passive technology providers.

Q for Goldman: Some situations you discuss don’t fit neatly into your description of reputation info (predicting past performance). Movie reviews aren’t about future performance, but about potential reactions to encountering that very same product. Evidence law struggles with the extent to which past performance can indicate future action. Would like to see clearer definition of the categories; movie reviews are easier to use because we know the variables and the performance is over. There’s more uncertainty in evaluating future performance based on grades, or a grade.

Seltzer: The information may not be transparent, because consumers aren’t thinking about motivations for ratings. Could we try to flush out more information about ratings before regulation?

Goldman: eBay has its own incentives to make the feedback form useful, and those incentives have worked really well. People trust the data enough to make purchases.