IPSC Panel 15 – Trademark Law II

Trademark Fame and Corpus Linguistics, Jake Linford, FSU College of Law and Kyra Nelson, Independent

Dilution protects famous marks as if they were monosemous: having same source meaning no matter what goods/services applied to. Proposal: use data about TM use from large corpora to better ensure that fame is properly granted and perhaps reduce the costs of finding fame/decrease reliance on things like sales numbers. Famous=unique commercial signifier (consistent with Schecter’s original formulation) and widely recognized (household name). Different databases with different sources, date ranges, numbers of words. Could use these databases to measure recognition via frequency of appearance in corpus and to measure level of singularity. [That’s measuring production not recognition, which is probably worth distinguishing.]

Question: can we set a threshold for appearaances?

Diachronic changes/change over time: could a mark lose fame the same way it loses distinctiveness? What are the words that appear more frequently than you’d expect next to this word, and do they point to singular meanings or multiple meanings?

Looked at frequency in database, compared to litigated cases. Lots of marks held not famous are relatively not frequent: Buck Rogers, Blue Man Group, Field of Screams. Some high frequency marks are found famous: Microsoft is an outlier [that is likely because the corpus draws from articles that mention using Excel or another Microsoft product that does the work]. But there are also ones with mismatches. Was this FTDA problem only? FTDA cases do have slightly lower mean frequency v. TDRA, but not statistically significant. One per million is relatively high frequency so that could be a frequency, but leaves out Citibank, Tylenol, Rolex, and Victoria’s Secret.  Nearly 2/3 of marks found famous are at 0.83 per million, so maybe that’s right. Non famous .042/million. 3 standard deviations above mean is .375 per million. Hotmail also changes in frequency a lot over time.

Collocation: Microsoft’s adjacent terms all point to Microsoft the entity (or its competitors).

What about Coach? Collacates in top 20 are all sports terms. Apple: all the brand use is computer-related but all the non brand use is much bigger and is apple/fruit. What’s the best treatment?

Could we do a better job predicting how cases should go? Choose Your Own Adventure: court seems to presume CYOA famous. In 1979, when first book comes out, only one hit. .06 frequency, highest .12 over time, never gets there. Concordance: some TM use, some not.

Mark Lemley: useful tool, but separate utility of tool from the particular vision of dilution which might not be widely accepted. Could be valuable even if you don’t need singularity.

Mark McKenna: singularity might not be about fame, but it is about dilutability: if the mark isn’t singular, it’s already diluted. Also, does frequency really correlate with fame? It’s not surprising that “coach” would have a lot of non-brand use b/c it’s an English word. Are you really measuring anything other than fancifulness?

A: Panavision is distinctive, but not famous: that’s why frequency matters. Apple is the question: does Apple swamp commercial terminology and (more importantly) should that commercial dominance matter?
Jeanne Fromer: need to have an explicit methodology for measuring fame for nonfanciful marks. Recognition and production are distinct: apple might be easily recognized as a fruit without a lot of use in the corpus. So we need better definitions.

A: is the concern overcounting or undercounting famous marks?

Fromer: probably both! It has to have frequency for you to care, but if there’s any mention of a non brand appearance, how does that factor in? Develop a methodology at the outset=very helpful.

Lisa Ramsey: agrees re fanciful or even suggestive marks (Microsoft) & separate analysis of those from what’s not in the dictionary.

Barton Beebe: exciting idea. Hand coding can account for nonmark use but you should feature that in the intro/discussion.

A: note that Hotmail looks very strong in one year—the corpus doesn’t always give you a snapshot over time and databases vary. Another concern: scholars seem to assume that a judge can be turned loose w/ this toolkit and reach right answers, and that might not be true. This could just be a tool to stick a number on an intuition, like with stampeding the factors.

RT: non word marks: it matters that the corpus is for words! Bigger picture: there is a hierarchy of goodness of marks; word marks are better at doing the things we want done with fewer side effects and we should start admitting it more readily. [Alt take: we maximize what we can measure and reliance on the corpus will make it harder for us to think about what non-word-mark fame would look like. That could be good or bad!]

Ramsey: does the corpus distinguish b/t contexts where brand use is more likely (casual discussions) v. where it’s not (academic journal articles)?

Fandom is Nonexcludable, Betsy Rosenblatt, University of Tulsa School of Law

Warner sued Potterhead running club, which raises money for charity by having runners run on their own. “We solemnly swear that we are up to no good.” WB claims: Ds designed merch & business model to appeal to and attract fans of the HP franchise, but that’s a privilege reserved to WB and its licensees.  But that’s not the law: New Kids on the Block says you can raise money by appealing to fans. There are other limitations as well, including the First Amendment.

Defenses tend to break down over things that look like “use as a mark”: Potterheads name, Packers Fan podcast. Names identify source of group/its goods or services and describe relationship to underlying object of fannish attention. They are descriptive in her categorization b/c they are referring descriptively to what they’re fans of rather in a TM sense to the creator of that underlying work.

Claims against fan organizations look like irrational overclaims, but that’s not a helpful argument. Even when there’s competition w/ the “authorized” user, there is no economic benefit in allowing control over these kinds of descriptive uses. Fanmarks have characteristics of aesthetic functionality: a non reputation related value. Anti-monopoly principles: you get shoddy fan goods instead of really nice ones.

Ramsey: use as a mark by the plaintiff? False association as a countervailing risk.

A: Among other things we’re willing to tolerate risks of confusion in Rogers, nominative fair use, descriptive fair use, other contexts; we should be willing to tolerate some risks here.

Jessica Litman: need a pretty crisp definition of fan activity/fanworks, or any number of competitors will be able to claim to fit.

A: not distinguishing fan from non fan, but distinguishing referential mark from that which refers to the producer of the underlying work in a Dastar sense.

Linford: people are raising money and that seems nice, but how do we distinguish that from a universal 5K run? We are in a place where there is a presumption of licensing for runs & theme parks. Not persuaded yet that New Kids will get you there.

A: agree, which is why she’s developing the argument.

Linford: but why is referential not indicating source/sponsorship? Can you say something about grassroots/astroturf, is that what we care about?

A: reference is to a type of good not to a source. We shouldn’t be calibrating rights based on maximizing TM owner wealthy; these are positive spillovers.

Elizabeth Townsend Gard: is this a commercial/noncommercial distinction?

A: Not necessarily.

Lisa Macklem: Concerned about people taking advantage of fans/feeding at the trough?

A: think it doesn’t matter b/c we still get the benefits [that happens all the time in traditional expressive media: biographies or magazine articles taking advantage of something popular, and we think that’s a good thing].

Lemley: very easy to say that money changes everything. And very easy to find money somewhere in a transaction, especially online. So you will have to go broad.

A: doesn’t care whether people make money. Under her standard: Why can’t someone just call themselves Warner Brothers & make movies? Because she’s basing this on a genericism [I’d say functionality] and referential use framework, we can use the same framework we use for house brands.

Certification (and) Marks – Understanding Usage and Practices Among Standards Organizations, Brad Biddle, Arizona State University, Jorge Contreras, University of Utah, and Vigdis Bronder, Biddle Law

Would expect to find certification marks going along w/standards orgs. Just 17K of 36 million TM registrations are certification marks, but hard to distinguish ICT standard setting organizations from others.

Findings: testing and certification is an important activity for SSOs, though some large organizations don’t do the testing. Over 60% did do their own testing, and most had an associated logo. Use traditional TMs much more than certification marks. In 94 different organizations: 122 certifications across 9 jurisdictions, but that’s just 2% of their marks. Why? Initial registration is challenging; providing a copy of the specification; more flexibility and control over use of a traditional TM; tend to have sophisticated licensing programs. Certification marks can’t be used by certifying entity, and as a pragmatic matter most orgs had similarity b/t mark and organization name. International issues. Implications: conventional wisdom doesn’t seem to match real world practice. Is this a misuse of marks, dodging the consumer protective elements of certification marks? For policymakers, clear that certification marks aren’t using them as intended. For reformers, tightening rules for certification marks won’t matter if most certification happens outside certification mark system: would have to focus on the process of certification instead.

Margaret Chon: TM too are supposed to be about quality, and your findings reinforce that.

Jorge Contreras: practically it seems harder for certification mark user to police use

Vigdis Bronder: no certification cancellations we found; even organizations that had no formal licensing structure for a long time were fine.

Chon: was revenue stream an issue?

A: interviewees were practitioners who might not know and probably would not have said. The people who run these orgs aren’t lawyers and don’t have in house counsel. Figuring out who has the info is tricky.

Rosenblatt: should we let people use non-certification marks in this way? Part of her issue with fandom marks is that there’s a similar blurring.

A: it seems there’s some need that the certification mark isn’t fulfilling.

Rosenblatt: but maybe that’s a bad need, like discrimination or self-dealing for oneself.

A: doesn’t think most of these orgs are being nefarious. International treatment issues are huge and they’re trying to be global organizations. Even trying to file certification marks in other countries can be very difficult w/different rules.

Bronder: since interoperability and sometimes even safety is a priority, it’s a different scenario than w/t shirts.

Ed Timberlake: are there new data you were able to get?

Biddle: did a mini project before w/USPTO and it was enormous amount of work. TMNow is really great. Could just go in and scarf up huge amount of data. Different than using USPTO data. Can find countries and classes.

Contreras: international data are also there and usable, which is great.

Barton: Do European perspectives change depending on tech sector? Might be different in how they treat agriculture versus tech where they might not diverge as much. Alexandra Mogyoros: Work on pseudo-certification marks: connect w/her. Antitrust/competition law issues, also in Fromer’s work.

A: yes there are nefarious actors, also non-nefarious actors doing good work.

Contreras: so many different organizations independently decided not to use certification marks. Doesn’t seem to be law firm driven/coordinated.

RT: Say more about international issues. Sounds a bit like they’re using trademarks just b/c it’s easier, which is not meaningless but suggests different policy levers.

Biddle: yes, there’s a global system for managing TM, but not for certification marks; only a handful of countries have such a regime, sometimes you have to show gov’t authorization. Can be expensive and even impossible in some jurisdictions. Even if you assumed that away, there are still relevant aspects, like discrimination. Do they have good reasons to discriminate that are policed via antitrust and not via certification marks?

Jessica Silbey: motivation—why one instead of the other—can be very hard to find in these studies. One possibility: catalog possible reasons and evaluate. Or focus instead on repercussions.

Chon: Jeffrey Bell’s book discusses breach of implied warranty from certification – maybe TM doesn’t raise that issue.

Biddle: many orgs call what they do certification, but a handful were very intentional about avoiding that word explicitly for that reason.

Portmanteaumarks, Brian L. Frye, University of Kentucky College of Law

Popular marks b/c they combine novelty (distinctiveness?) and familiar meaning. Looking for a framework for when they should be treated as distinctive v. descriptive/generic. Can mush words together, combine w/o blending syllables (brunch), combine w/blending (smog), orthographic/puns where there’s a homophonic element (shampagne); etc. May not help much w/evaluating distinctiveness but can be useful.

Framing: some portmanteau words are good, easily intelligible, and improve on existing words, thus enter the vernacular. [Is this distinguishable from embiggen?] Bad ones sound awkward/fail to convey information. Brunch is good, lunchfast and linner are not.

TMEP: portmanteau is subset of unitary mark, which is a combination of unregistrable terms that has the potential to be registrable. Compounds. Telescoped mark: closer to portmanteau, with blending.

Concept: failure to function. A good portmanteau creates a new word and thus should be generic. A bad one does not. Telling the difference: scores using a linguist’s test for predicting whether newly generated words are likely to become real words in a language. Formalized heuristic: Frequency of use; unobtrusiveness (does it feel natural or awkward); diversity of users/situations; generation of other forms/meanings; endurance of the concept. Each factor is scored 0-2 and if sum is 8 or higher, word is likely succes. 5-7 chance of success; less than 5 almost certain to fail. May not be directly portable to this situation but could be useful in helping think.

Linford: is this test really predictive? It seems post-use, which will have limited use to the PTO if the application comes early in the life of the new word. Work on sound symbolism may also help.

A: test is trying to predict whether a current word will survive, but needs tweaking.

Ramsey: consider genericide: what’s the process for a word or its components? Does it matter whether the initial components are clearly generic? Doctrine of foreign equivalents comparison: how likely are people to combine these words?

Rosenblatt: maybe we don’t care at all about portmanteau status, only about function: how does brunch function in language? The test you offer might work for all words.

A: does think that different words might need different analysis, but not fully committed to that. With spork, alternatives were offered, but they never worked well/caught on. Says there’s something special about that particular kind of combination.

Fromer: thinks these are a fruitful source of (claimed) marks to focus on. Consider the whole spectrum: could the word be descriptive even if not generic? There is some work on this in AI: working to take concepts and words and find best portmanteaus.

McKenna: similar question about categorization: how much do we want new categories that feed into or separate from Abercrombie? Takes Booking.com to be a statement that we will ask only one question about genericness: consumer meaning. The paper doesn’t have to be descriptive, but curious about the broader utility of subcategorization. Maybe easier to get purchase on that in registration than infringement, since TMEP is full of rules of thumb.

Lemley: it’s not right to say good = unprotectable/bad=protectable, though there might be some correlation; what do consumers think about it? To him the interesting placement is between descriptive and suggestive. You can see the etymology (if the new word is good enough) and then the question is how do you think about it. Also, everything here seems plausibly true of any combination of words put together to generate a new concept, e.g., Pretzel Crisps.

Sarnoff: neologisms can’t be generic until received into language [I am not sure this is true: any new invention will need a generic identification]. Time sequence is key.

A: trying to distinguish b/t way in which portmanteau is new and way in which other words are new words. How novelty works.

RT: we unfortunately have seen some conflation in TM b/t © concepts and TM concepts and this area could be a way to explore that. Compare the treatment given to HONEY BADGER DON’T CARE where the court conflated the creator’s authorship interests w/his TM interests v. LETTUCE TURNIP THE BEET, where at least at the district court level it may have been relevant that there was evidence that the TM claimant didn’t “invent” the pun.