Proving Consumer Perception: What are the best ways to test what consumers and users perceive about a work and how it is being positioned in the market? Featuring David Bernstein (Debevoise & Plimpton), Graeme Dinwoodie (Chicago-Kent College of Law), and Johanna Schmitt (Kirkland & Ellis), moderated by Barton Beebe (NYU Law)
The full video of this panel is available on YouTube.
- All right, as we gather back in from lunch, I want to welcome all of you back. Thank you all for sticking around for this afternoon. The next panel is all about proving consumer perception. So much of what we talk about when we talk about intellectual property-related disputes has to do with what people out in the world are seeing when people are using various types of works or various kinds of marks, and so, this is a panel where we really focus in on how do you quantify, how do you describe, how do you think about what consumers are seeing in connection to the dispute? So, I will turn it over to the panel. Thank you so much.
- Thank you Michael. Good afternoon. My name's Barton Beebe. I teach here at the law school in intellectual property, and it's my great privilege to be sitting up here with this superstar set of panelists. I will introduce them very briefly, and then we will proceed to a discussion of the topic about proving similarity. Just to my immediate right is David Bernstein. David co-teaches advanced trademark law with me here at NYU. He's an extraordinarily good classroom teacher, very generous towards our students, and he also happens to be the head of the Intellectual Property Litigation Group at Debevoise and Plimpton. As I was looking at, just most recently, the latest list of the cases he has litigated or is currently involved in, they read like maybe 1/3 of the biggest cases in trademark law in the past 20 or 30 years. I teach many of them, as do many others. So it's a great privilege to have him on the panel. To David's right is Graeme Dinwoodie. I never miss a chance to say that Graeme is, in my view, the leading authority in the world on trademark law. He is certainly the only person I know of who has total expertise, both on U.S. trademark law and European trademark law. Many people can pose, such as myself, in one or the other, but Graeme is the real thing. So very excited to hear his comments. I should say, as a matter of institutional affiliation, Graeme is currently the global professor of intellectual property law at Chicago-Kent College of Law where he is also, more generally, a university professor there, which is a highly-distinguished position there. Before that, he was at Oxford and was the intellectual property law professor at Oxford, and also ran their Intellectual Property Research Centre. That explains, I think, to some extent, his great exposure to European trademark law and intellectual property law, more generally. Then, to Graeme's right is Johanna Schmitt, who comes to us from Kirkland & Ellis. I was just reading, in The American Lawyer of, I guess it's this month. It's like a couple of, well last month--
- Yeah.
- last month.
- Hot off the presses.
- There's this wonderful review of Kirkland's Bionic Women. It's a group of litigators. I think, if you're in intellectual property litigation, you know about Kirkland's intellectual property group and Johanna is one of the bionic women there. If you look at the list of cases she's litigated in copyright law, trademark law, you see the other portion of the leading cases in trademark law, at least in the U.S. And I recognize cases that are in our casebook. So, as I say, it truly is a great privilege for me to be up here with these three panelists and I hope you'll enjoy their commentary on the question of similarity. We'll begin with David. I've been instructed to time their remarks to about 10 minutes each. We'll see how that goes. Then we'll have some internal discussion and then turn things over to the people assembled here. So David, please, thanks.
- Sorry about that.
- That's okay.
- All right, well thank you Barton and to the Engelberg Center for having us.
- Here you go.
- Thank you very much. I'm going to assume some familiarity with surveys given the time restraints, so that we can really get into the issue I wanna talk about. That issue is, when we are looking at the different results of likelihood of confusion surveys, what should the threshold be? Now I think there's some conventional wisdom that you wanna get at least 15 to 20% confusion. If you have a survey that shows a net level of confusion of 15 to 20% or more, a lot of courts will say, yes, that's enough. That shows that an appreciable number of consumers are likely to be confused. And actually that's the key standard that we've heard from the Second Circuit and other courts, that you wanna show an appreciable number of consumers have been confused. And if you're below that, then the conventional wisdom is, wow, that's probably not enough. Where did this number come from? Why is 15% sort of the magic number? So I've gone back and I've looked at some of the early cases that established the 15% threshold. We can go back to the 1970s and I think, really, the first key case was the Beefeater case back in 1976 from the Seventh Circuit. In this case, the question was whether this restaurant which was selling roast beef dinners at the Sign of the Beefeater, infringed the famous Beefeater Gin mark. A survey was done and found that about 15% of consumers were confused into believing that this restaurant came from or had some affiliation with the famous gin brand. The district court said, well that's not enough. 15% is just too small. That's too small a number. But the Seventh Circuit reversed and the Seventh Circuit said, no, actually 15% is enough. It is not a small number. The fact that it was a reversal, I think is quite significant, because it's not as if there were fact-findings here that the court of appeals deferred to. Rather, the court of appeals said, look, I'm going to accept the finding that 15% of consumers are deceived here. And the court of appeals specifically said that number is sufficient. We've seen, just by historical accident, this 15% number appear in other cases. In the Hawaiian Punch case I have up here, the RJR Foods case, there was 15 to 20% confusion. And in the Texon case, it was also 15% confusion. So here are three cases in the sort of early days of trademark surveys, all of which found that 15% was enough. There was no discussion as to why that was enough. There's no academic consideration of what makes that threshold significant. It's not like genericism surveys where you wanna get over 50%, you wanna show that more than half of consumers perceived the mark as being a brand name as opposed to a generic term. I can understand, intellectually, why 50% might make sense there. Why 15%? Well one of the things that's really key about these cases that I wanna hone in on is that, in all three of these cases, there wasn't a control. This was really in the time before controls were regularly used in surveys. Indeed, in the McSleep Inn case, which was more recent, from 1988, there, the court also found that 16% was enough. This was another case where there was no control. The question here is, were consumers deceived into believing that McSleep Inn had some affiliation with, who do you think? McDonald's, of course, because they have this family of Mc marks. Believe it or not, the survey showed 16%, again, without a control. So this is gross confusion, just like the three cases I showed in the prior slide. Yet the court said, look, I know there's some noise in this number. The courts were sophisticated enough to say, we know that this is not measuring the actual number. Some people guess. Some people are just giving me a brand reference. Some people might be led or biased by the way the questions are asked. But even if there is some amount of noise in that number, 16% is a significantly high number that it shows that there are an appreciable number of consumers deceived. So this is really key because the cases that are still regularly cited for the 15% threshold come from an era when we were talking largely about gross measures of confusion, not the net levels of confusion we see today. I will say, there are some cases from this era where less than 15% was accepted and I still try to cite some of these cases from time to time. Here, in New York, the Steinweg case is probably the one that confusion was found on the lowest numbers, 7.7% or 8.5%, depending on whether you were looking at, whether it was a business affiliation or actual confusion. But the court there did say, look, this level of confusion is enough. It shows an appreciable level of consumers are deceived. In the Jockey case that was involving whether JOCK SOCK was infringing of the Jockey trademark and the Goya case I have up there, we have, also, levels below 15% that were found sufficient. So we've got some history that suggests that some levels of confusion below this received wisdom of 15% can be enough. Yet, the courts have pretty much stuck to that 15% threshold and we've seen it time and again, including pretty recently in the 1-800 Contacts case in the Tenth Circuit, where the court specifically said, actually 7.5% confusion, which was the same as in the Steinweg case, that actually is evidence of non-confusion, that if you get a level of confusion like this in a survey, it actually is evidence that people are not deceived. It's not an appreciable number of consumers. So why does it matter and what's the key proposition I wanna leave you with today? It's this notion that the 15% number that a number of courts relied on were before controls. And controls are critically important. Here's one case we had in the Seventh Circuit involving these food replacement bars, South Beach Solutions. The question is, by using the South Beach name on the bar, were you deceiving people into believing that these bars came from or were affiliated with a Dr. Agatston who was the doctor who created the South Beach Diet? If you just did a survey, we might not know, are people deceived because they think all low-carb products must be associated with Dr. Agatston, or because the teal color of the bar is the same as the teal color of his book or the palm trees, for example. And he has a palm tree on the cover of his book. But we really wanted to focus in on, is the South Beach name causing the deception? So we created a control that mirrored virtually everything about that, but only took off the South Beach name. This control is as perfect a control as you can get. It allows us to measure, with great precision, to what degree are people deceived only by the inclusion of the name South Beach on the bar as opposed to anything else? Of course, we did find over 15% net confusion, so that was very supportive of our preliminary injunction application. But, in so many of these other cases that we've talked about today, where the 15% threshold was created, it was based on gross confusion. So, the proposition I'd like to leave us with is that, when courts today say the threshold is 15%, and they cite these old cases, they're talking about gross confusion, which is not the same as the net confusion that we are able to measure with much greater specificity today given the use of very creative controls. What should the right number be? Should it be 15%? Should it be something less than that? How should we evaluate what is truly an appreciable number of consumers? I think there's a few things that we can look at. I'd like to focus on the fact that if we have measured with great specificity, with a control, that seven or 8%, like in the Steinweg case, or 10 or 11%, like in some of the other cases we've looked at already today, are truly deceived, and we've measured the level of noise, we've measured mis-measurement error or bias or a leading nature of the questions, and we know that's a true number, then if we're talking about a large market, we are talking about a very large number of consumers. If you're talking about all consumers who go to fast food restaurants, which was the case in the Wendy's commercial here against Big Bite, so Big Bite had a television ad comparing its pita sandwiches to Wendy's hamburgers and McDonald's hamburgers and Burger King hamburgers, and saying that people should go to Big Bite instead, and in that case, the court found that 7% of consumers were deceived into believing that this was sponsored or approved or somehow affiliated with Wendy's. The court said that that small percentage nevertheless represents millions of consumers. I think we need to focus on the size of the market in thinking about what is an appreciable number. A really accurate measure of confusion at that level can still be millions of people. What are some of the other things we can look at? How proximate are the products competitively? If the products are directly competitive and you're showing a relatively smaller number of confusion, let's say in the seven to 10% range or 10 to 12% range, but the products are really sold in direct competition, that may also be something that would be relevant. So, I'd like to suggest that courts should be more open to focusing on what are the appreciable number of consumers who are being deceived here. And courts could look at certainly the quality of the survey design. So if the controls are giving us very precise measures, I think courts should be open to accepting that a smaller number than the conventional wisdom should be enough. There may be other reference points we should look at, so the market size or the competitive proximity. There are also other measures in these surveys. Look at the quality of the verbatims. Sometimes the verbatims, where you ask, why do you say that, they actually are very specific. I can show you that I'm deceived because it used the South Beach name. If you see that, it's actually telling you that it's not random, that it's the use of that name that was really the thing that was deceptive. Then the final point I'll make is that if you do find a survey that shows, let's say something in the seven to 10% range, even if a court might be reluctant to enjoin it, courts should be open to thinking about, are there other remedies that might be appropriate given that we still have a portion of the market that's deceived. Maybe we wouldn't enjoin it, but maybe we should think about disclaimers or other ways of trying to minimize that level of confusion that won't also harm the rest of the market. So I look forward to chatting with the panel and you all about this idea further. Thanks very much.
- Which one is yours?
- That's it, yup. Thank you, thank you Michael. Thank you Barton. Thank you, the organizers, for the opportunity to come talk to you. I'm gonna talk about the question of proving consumer confusion in trademark cases, so the same topic, to some extent, as David, but from a comparative perspective. The focus of my remarks, however, is actually gonna be provoked by a tension in the United States between the black-letter doctrine on the one hand and the actual practice as revealed in Barton's study from about 10, 12 years ago regarding the use of the multifactor confusion test in U.S. courts. Black-letter law suggests that a survey, as David sort of suggested there, I think, that a survey is the best evidence of confusion. However, when Barton did his study he found that, although many believed that to be the case, and the black-letter law says that, in fact the surveys were rarely presented by parties and very often not credited by courts. Obviously at the micro level, judges can be persuaded of the failings of any particular survey in a particular case, but given Barton's systemic finding across a variety of cases, is there, in fact, in the U.S. courts, a more generalized skepticism towards surveys? In that respect, it's intriguing that in the recent years, so the last seven or eight years, in some of the other more significant common law jurisdictions throughout the world, judges have almost uniformedly started to express skepticism about surveys as a means of showing confusions and surveys have started to recede as a means of proving consumer confusion. So you have a decision from the Canadian Supreme Court. You have a decision from senior courts in the federal system in Australia, and you have a decision, two decisions actually, from the English Court of Appeal that I'll talk about in a minute, which are pretty much at one on this point and on this skepticism. So I'm gonna focus on this Interflora decision that's on the bottom of the last slide, and the two brief points that I wanna highlight from the British experience are, first of all, why have the British courts turned their backs on surveys and do those reasons apply with as much force in the United States? Secondly, once you start moving away from surveys as a way of proving confusion, what comes up to fill that particular gap? Interflora is a keyword advertising case, a really significant case that went on for 10 years, still going on. Marks & Spencer's had purchased the keyword interflora, such that an ad for Marks & Spencer's online flower delivery orders came up in response to a search for interflora, standard litigation. Interflora conducted a survey which involved taking respondents who were initially screened into a hall with laptop computers. They were asked to type the word interflora into a search box on the Google home page. The result they got was not actually a real search. They got the search come up that actually was at issue in the litigation and then they were asked a series of questions, not unlike the kind of questions that a U.S. survey would pose. Interflora didn't actually use the survey numbers, however, in the litigation. This was actually what the British courts called a witness-gathering exercise or witness-collection exercise. The U.K. courts have previously expressed the view that it was much more valuable to them to hear evidence from real consumers than try and extrapolate from statistics in surveys. So it became not uncommon for a survey to be accompanied by testimony from some of the people who were surveyed. So they were actually brought into court during trial. And, in fact, in some instances, the witness gave the testimony and then survey wasn't even introduced, so that all that the survey was actually doing was actually collecting the particular witnesses. So since 1984, under sort of case management practices, parties who wished to be able to admit a survey, and therefore be able to recover costs if they won the case, under the English rule, actually needed advance permission from the U.K. courts to conduct the survey. That was true also if the survey was only being used for a witness-collection exercise. So, in the Interflora case, when the trial judge, Mr Justice Arnold, admitted witness collection evidence, and that issue was appealed to the Court of Appeal, the whole issue of whether the surveys were admissible was put on the table. So Lord Justice Lewison, interestingly for an IP case not an IP judge, wrote the opinion for the court and suggested that survey-based evidence was generally of little or no value. He said, "Sometimes it does no more "than confirm the conclusion "that the judge would have reached without the evidence." So he held that, even if the survey or evidence from witnesses emanating from the survey is technically admissible, the judge shouldn't let it in unless satisfied, first of all, that it would be valuable, and secondly, that the likely utility of the evidence justifies the cost involved. And he considered, in this case, that, in fact, Interflora hadn't shown that there was real value to the particular evidence. Mr Justice Arnold was not deterred by the Court of Appeal and this time he, it went back down to him and he allowed in additional survey evidence on a slightly strained interpretation, maybe of the Court of Appeal's decision, claiming that this survey evidence was actually spontaneous, actual confusion evidence. Back to the Court of Appeal, and Lord Justice Lewison was having none of it. He said, "With the benefit of hindsight, "I perhaps, "I did not make my message clear enough in Interflora 1. "Let me say it again, but more loudly. "A judge should not let in evidence of this kind "unless the party seeking to call that evidence "satisfies him that it's likely to be of" bold and upper caps "real value, "and that the likely value of the evidence "justifies the cost." So the upper caps, very Trumpian, and bold, apparently conveys the force of the view. Indeed Curley, which is the living practitioners' bible for trademarks in the United Kingdom, has actually taken to repeating that presentational form to convey the force of it when it refers to the test. So this is actually one of the pages that throughout their treatise, they actually use the real value test in upper caps just so you get the message that Lord Justice Lewison was trying to convey. The court said, and this is true both of Lewison and also Robin Jacob, in the concurring opinion, that it was not absolutely precluding the admission of surveys. It was lying. It has. I am not aware of any single case since Interflora, in six years, where a survey has been admitted on confusion. It has been occasionally on distinctiveness and secondary meaning but not on confusion. Effectively, they have killed surveys in the U.K. courts. So what are the reasons for this? First they have great doubts about whether surveys are reliable in their methodological rigor, in partial implementation, and accuracy of the conclusions drawn. Essentially, they think of surveys as second best constructed evidence, both because of the artificial environment in which they are asked, differing from the actual purchasing experience and because of the hired-gun dimension to dueling party-adduced surveys. Secondly, surveys can be expensive to design, to conduct and critique. And concern over the costs of litigation has generally become more prominent in the United Kingdom in recent years. And this was actually referenced by Lord Justice Lewison in the Interflora case, and it's clearly the cost issue that is driving a lot of the debate. So my question that I was asking in the beginning there was, do those concerns transfer to the context of the United States such that they might be informing the unspoken U.S. resistance that Barton detected in his study 12 years ago? One might expect the first concern to apply in the U.S., but that manifests itself, the concern over reliability, in case-by-case critiquing of the weight of surveys, rather than by blanket rule decrying them. It might also be that there is an appropriate difference between the U.S. and the U.K. on this because the value side of the cost-benefit ledger might be less weighty in the United Kingdom because the U.K. courts, starting actually in the Interflora case, have taken pains to stress that the concept of the average consumer, who is the person in EU law language who has to be confused, is a normative legal fiction. So in an earlier iteration of the case Lord Justice Kitchen characterized the average consumer in trademark law as a, quote, "person who has been created," doesn't exist, "has been created to strike the right balance "between various competing interests "including, on the one hand, the need to protect consumers, "and, on the other hand, the promotion of free trade "in an openly competitive market," unquote. So a court's legal conclusion about whether a defendant gets too close is more than a mere empirical assessment that one would get from a survey. And that may or may not be true. We could have a debate about that in the United States, but it's certainly no U.S. court has articulated it in those particular forum as the essence of trademark infringement. The second concern of costs might also transfer to the United States, but a completely different calculus might be in order whereas here we, of course, we don't have a loser pays system. So the English cost regime under threat of having to pay the other sides costs arguably works as a disincentive to vindicate rights. So making litigation more expensive might play out with even more pernicious effects in the United Kingdom than it has in the United States. Second point, and I'll finish with this, as I see I'm running out of time, what has happened in the U.K. to fill the gap of surveys? Has expert evidence and empirical analysis gone away? The answer is, no. First of all, we still have expert evidence admissible on the conditions of market and consumer behavior generally which judges then use to inform their analysis in particular cases. So, for example, in Interflora itself, when it eventually gets to trial before Mr Justice Arnold in take three, he allowed expert evidence regarding searching habits of people using searching engines. That included an eye-tracking study, which Google had commissions, would produce heat maps to show where people looked at particular parts of a search results page. It included a study by Ben Edelman and Duncan Gilchrist from Harvard Business School on the effect of a labeling ad. So that expert evidence still gets in. Secondly, and perhaps more worryingly, litigants have started to scour for more direct evidence of actual confusion now that they can't use surveys. So the plaintiffs in a number of cases have sought to rely on completely random third-party online uses of the marks as evidence of confusion, and given the mass of online information obviously, it's not hard to find some crazy examples of usage. One has to ask, I think, whether, in fact, admitting that kind of evidence is worse or better by the same standards to which the courts have been holding what are much more scientific surveys. So, in sum, if Barton is right, as he suggests in his survey, survey, sorry, in his study, 12 years ago, that U.S. courts are less persuaded by surveys than the black-letter law suggests, the U.K. experience, I think, suggests that it's worth thinking hard about the alternatives for jettisoning the existing approach that American courts use to survey evidence. Instead, it might be that the focus should be to address the reasons why courts have concerns about existing surveys by improving reliability and reducing costs, and this is an irony in that those two may actually cut against each other, but also, I think, thinking about other ways in which expert evidence can help courts make informed judgments about the way they regulate the competitive environment. Thank you.
- Great, thank you.
- Thanks, I'm gonna turn the tables now away from trademark law and similarity between trademarks and likelihood of confusion and talk about issues of similarity in the copyright context. In doing so, I'm gonna talk about the Ninth Circuit somewhat recent decision in Rentmeester versus Nike. That case is near and dear to me, as I worked on it. I represented Nike, along with my partner, Dale, who will be up on the next panel, and it's playoff season and everybody loves basketball and thinking about Michael Jordan. By way of background, Rentmeester is a photographer. He created this photograph of Michael Jordan on the left in the early 80s. At the time, Mr. Jordan, or Michael, was a student in college. He was on the U.S. Olympic team. He was a bright and hopeful star. This photograph was published in Life magazine as part of a piece on U.S. Olympic hopefuls in the early 80s. Flash forward a couple years, late 80s. Michael Jordan's on the Chicago Bulls. He is hired by Nike. He's a superstar. Nike creates this photograph in the middle which it used in ad campaigns at the time for its then new Air Jordan brand. Rentmeester sued Nike just a few years ago for copyright infringement, claiming that Nike's photograph infringed his photograph, and he also claimed that Nike's Jumpman logo, which has been printed on a gajillion pairs of sneakers over the years, also infringed his photograph. I remember, when I first got this case, looking at the complaint in my living room. My husband walked by and said, "Oh boy, you're in trouble," like, those look the same to me. Now if it was a trademark case, I'd be worried, 'cause he's a consumer of sneakers. But, I said, "Oh, you know nothing about the law or copyright principles, so, you know, and thank goodness, the court agreed, spoiler alert, and the district court dismissed the case finding no similarity, and the Ninth Circuit agreed. What did the Ninth Circuit do? First of all, in copyright cases, it's important to remember that similarity comes up in different parts of the test. This is a flow chart that I created sort of to visualize the things you have to prove to win a copyright case in the Ninth Circuit. It's similar among other circuits. But, you have to prove you own a copyright and that there's been copying of protected aspects of the work. And that breaks down further. For that factor, you need to prove two things. One is actual copying, that somebody actually copied the work because independent creation is a defense even if they are identical, With actual copying, sometimes you can prove that with direct evidence. People admit it. There's a photograph of them at the Xerox machine. I don't now, but, that's a little less common. More commonly, people prove that through circumstantial evidence, and that's the, defendant had access plus the two works share similarities that are probative of copying. So that's one place where similarity comes up. In addition to actual copying, you have to prove that there's been illicit copying or unlawful appropriation. So you could copy something, but if it's in the public domain, there's no claim. So you have to prove both. And that's what courts often refer to as substantial similarity. In the Ninth Circuit, there's a two-part test, the extrinsic test, which assesses objective similarities between the two works after filtering out unprotectable element, and the intrinsic test, which is sort of a holistic total concept and feel comparison. The extrinsic test can be decided by a court as a matter of law. We, or Nike, moved to dismiss Rentmeester's complaint on an early 12 motion on the grounds that the works, when you look at them, are too dissimilar. He can't state a claim that it satisfies the extrinsic test, as a matter of law. That's just a good practice point for copyright cases. For trademark cases, I'm not a patent lawyer, but maybe it's the same, it's hard to get cases dismissed early. It's very factual. People allege confusion, whatnot, and then you have to go through discovery, have experts, survey experts, all that, to sort through it. With a copyright case, sometimes, if it's books or movies or something the judge can look at side by side, as a defendant, you can make a claim and get it dismissed. If it's complicated music or source code or something that you really need someone to talk about what's protectable or not, it might not work. But, in this case, it did, and the Ninth Circuit affirmed dismissal. Another thing that the Ninth Circuit did in this case was clarify the application of the inverse ratio rule. Rentmeester seized on, sort of, I don't wanna criticize the Ninth Circuit, but there was imprecise language in some older decisions which said things like, if you have strong proof of access, then the standard for substantial similarity is lower. Rentmeester seized on that and said, "Under the inverse, I've alleged access by Nike. "They had my photograph. "Ergo, the standard for substantial similarity is lower "and the works don't have to be as similar "and my claim should survive." What we argued and what the Ninth Circuit confirmed is that the inverse ratio rule only applies to this similarity test under the actual copying. It has nothing to do with this test. So, that didn't apply here and they rejected that argument. So in comparing the works, the courts obviously had the works side by side, and, in the Ninth Circuit, and in most courts, is, the first thing they have to do is decide what is the idea of the work 'cause that's not protectable. Here we argued, and Rentmeester said, in articles and in his complaint, that the idea was having Michael Jordan dunk a ball in a pose inspired by ballet's grand jeté, and the court found that, okay, that's the idea, not protectable, granted it broad protection because there's a number of ways you could do that, but here found that, even though Nike's photograph also was basely loosed, perhaps, on a grand jeté, that they looked very different. In Rentmeester's photograph, he's in a scissor split, he's looking like he's moving horizontally. In Nike's photograph, it looks like he's moving up. His legs are sort of straddled. The Ninth Circuit clarified that you can't have copyright protection in a pose, just the pose, meaning I could pose one way and you could have a marionette pose in the same way. That's not copyright infringement. It's how you've express that pose. Here, again, the poses were different. The limbs were placed differently, arms not quite the same. The Ninth Circuit recognized that, okay, both have the same concept of shooting him outside, not in a traditional basketball court. But that's where it ended. The expression of that was very different. Rentmeester, he's outside, there's a grassy knoll, the sun is pictured. In the Nike photograph, it's the Chicago skyline. The colors are different. The mood is different. Over here, in Rentmeester's photograph, Michael Jordan is a hopeful. Is he gonna make it? You don't know if he's gonna make that shot. Everybody's rooting for him, but it's just not clear. Nike's photograph is all about triumph. Here I am. I'm in Chicago and I rule the world. And went on and on about the different arrangement of the basketball hoop, et cetera. In the end, the court found that the differences meant that Rentmeester couldn't state copyright infringement as a matter of law. And as a sort of final practice pointer, we've done this before and had success and obviously courts have dismissed early on, at that point, proving the differences as a defendant, you don't have a fact witness, you don't have an expert explaining it, so in your briefs, you need to really make charts and bullets and really walk the court through all the differences. You can't just say, oh look at them. They're different. See. You have to, even if it seems painfully obvious, the blue sky versus red sky and sort of juxtaposing all the differences can be very powerful to a court and sort of prove that the works aren't similar.
- Thank you.
- That's it.
- Thanks to all the panelists for really interesting presentations, which I'm sure raise a lot of questions among the audience. But first, as was our plan, I'd want to turn it over to the panelists to see if they have comments for each other or questions for each other, and I'll also maybe throw in a question or two. But first, maybe some discussion of the value of survey evidence might be in order.
- Yeah, so Graeme, actually, your remarks made me think about the way some U.S. courts are thinking about surveys and false advertising cases these days because we're seeing, increasingly, courts being very skeptical of what does the survey show. A couple of cases that come to mind, one from the Seventh Circuit, was the 1st Choice of Doctors claim where the survey showed confusion and the court of appeals said, those are regular words. I don't need a survey to know what these words mean. So, I don't care what the survey shows. I'm going to find that this is truthful advertising. Or, in the Havana Club case where a survey showed net confusion, about 15% of consumers, believed that Havana Club Rum, sold by Bacardi here in the United States, made in Puerto Rico, came from Cuba because of the Havana Club name in the name of the rum. The court said, I actually won't consider the survey at all because everyone knows that this is Puerto Rican rum. It says Puerto Rican rum right on the label, and so people can't be confused. And yet, the survey showed that people were confused because they saw the label with the words Puerto Rican Rum and it was very well-controlled because the control took the word Havana and changed it to Silver Club, and so we were, in that case, one was able to see that the reason for people's deception was not 'cause they thought all rum must come from Cuba, but rather the specific use of the Havana name. I think it's troubling when judges substitute their own judgment for those of consumers. The last point you made, I think really resonates with me, because if the concern is, maybe the survey is not measuring what's really happening, then let's talk about what's wrong with the survey. Is the methodology right? Was the control good? Did we survey the right audience? Did we ask the right questions? But if the survey is accepted, and if it really shows that people are deceived at that level, maybe they're numbskulls, maybe they don't realize that there's an embargo with Cuba, but if the survey really shows that, I think it's improper for judges to substitute their own judgment for that of what the surveys show.
- So that's interesting. Two things you said there which I pick up on. One is, when the Court of Appeal in Interflora articulated it, it technically was only talking about instances where we're talking about ordinary consumer goods where the judges were in no worse a position than the survey respondents to come up with an assessment. In fact they've, since then, not managed to find any circumstance where they're not perfectly well able, including cases where understanding Hindi TV in the United Kingdom, but British judges who didn't actually speak the language was regarded as context in which they were still able to understand the nature of the goods. So I think a lot of it does, there is an impetus in cases where they think I'm just as much of a consumer as the consumer is answering I think to sort of take it on. The other thing in terms of substituting judgment, let me be a little perverse on this, which may come back to the point about the role of the judge in and the notion of confusion and European law being a little bit different. I think, and certainly European scholars have said this, that European judges have said that why they don't like surveys is they think that, in fact, their role as the judge is being substituted by a bunch of scientists out there. So, I'm the one who's making the decision and if you give surveys a tremendous amount of weight, essentially that power is being taken away from me. All I'm doing is sort of administering this factual determination by scientists, and if they are making what is a normative decision, which is, again, the different conception, maybe, of European trademark law, then, in fact, they want to be in control of that. And interestingly, you don't see a lot of that in American confusion discussion by judges, but I remember in the Samara case way back, John Newman dissented at the Second Circuit on the distinctiveness point. One of the concerns that he had was, in fact, the distinctiveness had a lot of in-built normative concerns about competition, and by turning it all over to the jury, he was reneging, in some way, or the Second Circuit was reneging, in some way, from making a sort of legal judgment. So, some of the concern about who has control of a case I think might be at play in some of the attitudes you see on both sides of aisle.
- Ah, interesting.
- I'd like to jump in with a question about the Michael Jordan case. I have a thousand questions about the case but that's for another time. I was very much struck by your comment that when your husband looked at these photos, he said, "Oh boy, you're in trouble."
- Right.
- And, you said, "Hold on. "This is not a trademark case. "This is a copyright case." I'll take advantage of that because I thought it was such a lovely expression of some of the distinctions that one is hearing on the panel between a trademark analysis of similarity versus a copyright analysis of similarity. So, on the copyright side, I felt like I was back in art history class or something where you're sort of saying, well see, that's vertical and that's horizontal and if you had told me that's horizontal and that's vertical I might have believed you as well. And, the inside word of mouth, the Nike case apparently is that Nike benefited from the bionic women at Kirkland.
- They had extremely good lawyers and that really helped them a lot, and the other side, maybe, was not as proficient in this particular case, on these issues, so to speak. So, I suppose my question is, first, did you come up with the horizontal versus vertical thing your own or did the judge come up with that? More generally, is it, as a sort of strategic matter, is the plaintiff in a copyright case trying to persuade everyone that this is a copyright case, whereas the, oh sorry, a trademark case, whereas the defendant wants to persuade everyone the similarity standard is a copyright standard? To say that better--
- What do you mean?
- 'cause you're looking at me like, what did I just say?
- There were no trademark claims in the case. So sometimes you have both.
- But the plaintiff wants to, essentially, show to the judge, look at these two photos. Obviously, there's a problem here. It's almost like a bad faith issue. Avoid all of this fancy lawyer-made decision tree and just to ask, do you think these are similar or not, regardless of all of this filtering and everything else, which begins to sound a little bit more like a trademark question than a copyright question. I mean, is the standard of similarity would you would say is quite different in those two situations?
- Yeah because, again, trademark law is all about the consumer and confusion and it's not, it doesn't, one factor is comparing the two marks at issue. I mean that's just one of many. So even if you have identical marks, there might be other factors that show there wouldn't be confusion. It's different businesses. The consumers are educated doctors that always would know the difference. So, while you might look at something and, at first say, oh my God, it's the same word, we're in trouble, it's much more nuanced. And in copyright, the same thing. There, there doesn't have to be any confusion. It can be two different works. They don't have to be competing. But, again, it's not just looking at it and saying, oh yeah, my God, it's Michael Jordan. He sort of looks the same. You can't own Michael Jordan's image. You can't own a basketball hoop or him holding a basketball. Nobody can have a monopoly on that. That's what I think I meant. Like, you don't know copyright. Like, you don't know about filtering out and it has to be protectable expression.
- Maybe I could ask a question of David's, which we've talked about this ourselves and he knows I disagree with him on his approach. So I'll ask the question that I've asked him before.
- This is why our class is so much fun. We disagree on almost everything.
- I'm very much in favor of using a percentage rather than absolute number of consumers who are confused because the percentage takes into account the percentage of consumers who are not confused. Michael Grynberg who is a trademark scholar who published in NYU Law Review, an article years ago, on consumers who are not confused essentially really made the point that, okay, you've shown a million people are confused, but the other side has shown that 15 million people are not confused, and those 15 million people benefit from the information that is gained by the defendant's use of a similar and, to them, non-confusing mark. If we just look at absolute numbers, we don't take into account that cost-benefit analysis of the percentage of the consumer population however, that might be very small that is confused versus the enormous proportion that is not confused. So that's the argument.
- First of all, that actually does nothing to justify why 15% is the right number. One could talk about what should the right balance be? But just because there are some people who are not confused doesn't mean that that means that the defendant should be allowed to continue doing what they're doing. You could say, well, even if it's 30% confusion, still the vast majority, 70%, are not confused.
- Yeah, why not?
- I think the reason is that trademark law already internalizes a lot of protections for the other consumers. We don't allow marks that are generic to be protected. We don't allow marks that are merely descriptive to be protected. We don't allow marks that are functional to be protected. We have a nominative fair use defense. We have a descriptive fair use defense. And trademark law is a lot about competition. So, because I feel like most of these protections that protect the other consumers are built in, that's one reason why we can say, if despite all of that, there's an appreciable number of people who are deceived, shouldn't we stop that? But I think the other way that I come at that is the very last point I tried to make in my remarks, which is, maybe this issue should impact what kind of remedy we look at. And I do think that when we're plaintiffs, too often we go and say, Judge, I want an injunction. But we are seeing courts of appeals say to district courts more often, you need to think about, is the remedy properly measured for what the harm is? Although I disagree with the way in which eBay against MercExchange, which was a patent case, is being applied in the trademark laws, and Mark Lemley and I have both written about why we think the presumption of irreparable harm should still apply, what the Supreme Court did tell us in eBay and remind us, is that equitable rules still should be considered in thinking about what kind of relief you grant. If we could show, even in, if it's toothpaste, that presumably almost everybody in the United States buys, so we're talking about the whole U.S. population, and if 5% of people are deceived by Crost Toothpaste, well, sure, maybe 95% recognize that the colors are different or they're sophisticated enough to see that Crost is not Crest, but can we come up with a remedy that will actually not harm those 95%? Because, if you sell Crost, but you have to change the font or change the colors or have a large disclaimer to reduce the confusion, I don't think that that remedy actually harms that 95%. If we're saying, you can't sell this at all anymore, well yes, maybe we are harming the 95%. But more often than not, even if it's an injunction, it doesn't take a product off the market, it simply says, can we reduce the level of confusion? And, I think courts should think about whether an injunction is automatically the right answer.
- I see we have some good questions coming up so we'll start with them.
- [Mark] In the spirit of an academic conference, I have a comment and then a question, and also, in the spirit of academic, Barton stole my question, so I'm gonna ask a different one that's related to that. The comment is, sort of about the colloquy between David and Graeme and I think, Graeme, you'll sort of appreciate this. It seems to me that what's happening in a lotta cases when judges say things like, I don't really care what the survey shows in this circumstance, what they're really saying is just, whatever level of confusion here is just not, it's not reasonable confusion or there's nothing that I would require of the defendants beyond what they're doing that is reasonable. So they're just sort of substituting a normative judgment about what kinds of confusion should count. So, maybe that's illegitimate, but I would just, I think, point out that that's actually pervasive through the doctrine and that there's all kinds of places where courts just sort of say, as a matter of fiat, that there's not gonna be confusion caused by this. So you think of in the private label goods cases where they sort of just don't accept the idea that there's any empirical evidence that's relevant to the question. They just sort of announce it. So, it doesn't strike me as unusual. I think it's just it would be better if the courts were more transparent in what they were saying about them. So the question is this, is that I, along the lines of what Barton asked is, in those cases, if the argument, in the cases where there's a relatively small percentage of consumers who are confused as the survey shows, should there be some obligation, and I appreciate the comments about the remedy 'cause I think there's been some impetus in that direction, but, if the proponent of the argument about confusion is essentially suggesting we can come up with a remedy that doesn't harm this vast majority of consumers, should the proponent of the survey have to demonstrate that empirically? Should you have to actually do a sort of an alternative survey that shows, we can actually make these particular kinds of changes and we need to stand that up alongside the survey that's actually being introduced to the court?
- Yeah, it's a great question, Mark, and the Second Circuit, in the HBO-Showtime case actually suggested, that if you're relying on a disclaimer, it's your burden to come forward to show the disclaimer's effective. I think we can take that idea and expand it and one of the things Shari Diamond and I have been discussing is whether we should have multiple controls to show, actually, which part of the trade dress, for example, is actually the thing that's causing the confusion. 'cause when you look at the whole trade dress, you may not be aware of which is the element that's causing confusion. I think it would be very interesting for parties to actually come to court and say, what I want them to do is, don't just enjoin them, make them change it this way, Judge, because this would reduce the level of confusion. I think a smart plaintiff should be thinking about doing surveys that would show that and I think it would give courts more comfort in issuing an injunction because they would feel like this really is going to be effective in minimizing that level of confusion.
- Dale Cendali and then Rob.
- [Dale] I'm very interested in this issue of appreciable number of consumers being confused. I think, David, you suggested that, well, you know, if you have a popular product, even if it's some number less than 10%, that's still a lot of people. You're making a quantitative argument. My question for you and for the panel is, what about a qualitative argument? Have you seen anyone arguing, yeah, it's only 3%, but this is about a drug that would kill people and we gotta protect those people even if it's 1%?
- Yeah. We actually, we had a case for Bayer involving a flea and tick remedy and there's a particular product that's very effective in killing fleas and ticks on dogs, but it also is very effective at killing cats. If you are a house that has dogs and cats, you shouldn't use this product on your dog, 'cause it's not gonna go over very well in the rest of the family. So, there, we actually didn't even have to go to surveys because, just by pointing out that the risk of people being deceived and using this product without it being clear that this would kill their cats, it's so draconian that it really built right into our remedy. And I will say it goes to a point Graeme mentioned, which is, surveys should not be the end-all. Surveys are one element in the multifactor elements that should be considered, but I do think courts should consider more. The reason we haven't seen the exact argument you've just made, is that no plaintiff in this environment would put forward a survey that has under 10% because you are walking into such a trap given what the courts have said about surveys under 10%. Then the really interesting issue is, when I'm a defendant and I do a survey to try to prove an absence of confusion, and I get something like seven or 8%, I have to say, as a defendant, I get very nervous. I'm probably not gonna use that survey 'cause I'm worried about these speeches all coming back to haunt me, and people saying, oh yes, 7% should be enough. It may be those cases where someone might try to argue that. The qualitative part, where I think we also can focus on it, is the verbatims, because I really think the quantitative number of X%, that's one thing. But I really like, in my briefs and in my arguments to courts, to show, this is what consumers said were the reasons. Why did you say that? If you don't get anyone in response to the why did you say that question feeding back the trade dress aspects, or the trademark, South Beach, you have a real problem. In the South Beach case, for example, a number of people, in their responses, their verbatim responses, said things like, it says South Beach on the packet. Of course it comes from Dr. Agatston. Or, duh, it's a no-brainer. I saw the same book in the bookstore. And when you get the quality of verbatim answers, that, I think, can bolster a smaller quantitative number, 'cause it shows the court that the very thing you're concerned about is actually the thing that's causing the deception.
- And another thing is, and Dale, I know you know this well, you taught me, but, another thing to do is, it's not always just surveys and that's the end of the discussion. We use other types of experts on these issues all the time in trademark cases. So you can have somebody talking about whether a mark, a linguist or somebody in the industry saying, what's the meaning of a mark? We use someone who opined on the meaning of pin and was pin generic for pinning on the internet? You can have a marketing expert who can talk about the marks, how consumers are perceiving it in the marketplace, how the brand images, overall, are different or the same. They can't opine on the ultimate issue, but they can opine on different factors that go to secondary meaning, third-party use, all that sort of thing. Sometimes you might have a fact witness at your client who has some knowledge, but often not, and they don't wanna do all that heavy lifting. So it's nice to have these sort of industry or marketing experts who can buttress a survey. I do agree survey evidence is probably the most powerful now, but it's always good having other types of experts to help your arguments.
- And that goes to what Graeme was saying about the European model or the British model where they appear to be moving in, have moved in that direction.
- Right.
- And in the other part of, to pick up on David in that response, is on the verbatims. It always struck me as illogical, as someone who was arguing for more use of surveys in the U.K. courts, that they were willing to say that the survey is utterly unrepresentative, but if you could just bring in four people out of what's an unrepresentative sample, and we'll talk to them, we will get to the bottom of what's going on. But actually that the impulse behind that, even though it seems quite illogical, was that the ability to have an ongoing dialog that followed the bread trail a little bit more, would give you a better sense exactly, the nature of the type of the confusion, how deep the confusion was, what were the consequences for the consumer of being confused, that would allow the court to get a better sense of what was at stake in the case.
- It brings to mind the old story, the true story, of, was it Judge Hand or Judge Friendly, who, in the Seventeen magazine case decades ago, before surveys were common, actually just went around the courthouse and asked people, asked young girls, showed them Seventeen for girdles versus magazines and then reported this in his opinion. Very unscientific approach to just--
- And totally in appropriate for the court of appeals.
- I was gonna say, a little creepy.
- And inappropriate for other reasons too.
- Yeah, sorry.
- Rob.
- [Rob] If we talk about surveys and confusion doesn't the word likelihood carry the imprimatur of, say 51%, as opposed to possibility of, or appreciable, or those other concepts? So if we're going to use likelihood of confusion, why don't we say 51%? And the other part of this question is, aren't there studies that show there are going to be X number of people fooled all the time no matter the similarity? You can fool some of the people all of the time. So, what are the control levels of ordinary confusion that we get all the time?
- The second part of your question's actually very easy to respond to because I do think controls totally protect you against the some people are always gonna be confused. I will say I've done surveys where control levels were two or 3%, but I've done surveys where the control levels have been 20, 30%, and it really depends on the product. But the point is, if the control is well-designed, it's really telling you what's going on here. It may be that you have to ask a very focused question. For example, in a Squirt-style survey, where you're showing the products side by side, the very nature of the survey suggests a relationship. Why else would you be asking me this question? So in Squirt surveys, we tend to see much higher levels of control noise than we do in Eveready surveys. On your first question, I don't think the word likelihood really carries that much significance. I think likelihood is as compared to actual. In other words, the point is, and the whole idea of likelihood of confusion very often comes up in preliminary injunction motions, you don't have to wait until the harm has occurred. If we allow this product to be sold, and we let it go out into the market, is it likely to cause harm, and if so, that harm should be remedied now. We don't wanna actually wait 'til the harm has occurred, and then say, oh yes, you're right, this actually has caused lots of confusion, because then, the irreparable harm is irreparable, if you truly believe it's irreparable. I just don't think likelihood can do as much work as suggesting that it means 50% should be the threshold. And indeed, from a policy perspective, I would find it very troubling if someone was selling a product that confused 45 or 40% of all consumers and a court would say, well, it's not a majority so we'll let that happen. That certainly is an appreciable number of consumers.
- [Rob] Yeah, then shouldn't we rephrase the test to be appreciable confusion?
- I think it's likelihood of confusion among a appreciable number. Like it's not more likely than not. That's how I always thought of it. But.
- I'm getting a signal from Michael Weinberg, our executive director that's
- [Michael] I'll play the heavy here.
- Yeah, so we're out of time, but I
- Mr. Sandman.
- hope you enjoyed this panel. I certainly did and let's thank the panelists.