By Gus Campbell. Gus Campbell is a 3L student at Sandra Day O’Connor College of Law. He is the Student Co-Executive Director of the McCarthy Institute. As a musician, Gus is interested in the world of intellectual property, specifically copyright, trademarks, and right of publicity. Gus is interested in pursuing a career in intellectual property law and entertainment law.
Introduction
In the world of intellectual property law, survey evidence has emerged as a highly influential and persuasive tool. In trademark disputes, survey evidence is practically a necessity, as properly conducted surveys offer courts essential insights into consumer perception.[1] In patent infringement matters, various survey methodologies have been developed, from non-infringing alternative surveys to apportionment surveys.[2] Even in right of publicity matters, surveys provide consumer evidence necessary to establish damages.[3] However, there is one area of IP in which survey evidence is not just rare but oftentimes discouraged: copyright law.
In copyright infringement cases, expert testimony and related surveys are typically disfavored as evidence to support an infringement analysis.[4] Despite this reluctance, surveys could play a pivotal role in proving certain elements of copyright infringement analysis, particularly whether two works are substantially similar.[5] This essay aims to recap the copyright infringement analysis, highlight the potential use of survey evidence to measure substantial similarity, critique why courts generally afford little to no weight to surveys that seek to measure substantial similarity, and suggest that courts’ general disfavor of survey evidence in copyright infringement matters may be unwarranted.
Analysis
In determining copyright infringement, courts generally engage in a two-factor analysis: (1) factual copying, or copying in fact, and (2) improper appropriation.[6] While circuits may differ in their approach, the aims of the analysis remain relatively consistent— to gauge substantial similarity between the works.[7] For the first factor, most courts consider either (a) direct evidence of copying by the alleged infringer, or (b) circumstantial evidence showing either (i) access to the copyrighted work and probative similarity or (ii) such striking similarity between the two works that independent creation can be precluded.[8] For the second factor, courts find improper appropriation if the allegedly infringing work both copies protectable elements of the copyrighted work and is substantially similar to the copyrighted work from the perspective of an “ordinary observer.”[9]
While surveys could serve a useful evidentiary purpose for various portions of the copyright infringement analysis (and, indeed, expert testimony has been held to be appropriate in some instances),[10] the second factor—and its inquiry into substantial similarity through the lens of an ordinary observer—presents a prime opportunity for survey evidence.[11]
A survey on “substantial similarity” could adopt methodologies from fame, secondary meaning, and likelihood of confusion surveys in trademark matters. For example, screening questions could ensure respondents reflect the U.S. Census on key demographics such as age, gender, race, and geography.[12] Following these, the survey could use formats like Eveready[13] or Squirt[14] survey methodologies, which are commonly used in likelihood of confusion matters. The Squirt format, in particular, would allow side-by-side[15] comparisons of the works, with respondents indicating whether they believe the works are “substantially similar” through open or narrowly directed questions aimed at the penultimate matter of substantial similarity.[16] The line of questioning in determining substantial similarity could have answer selections that are similar to a Likert-type scale (i.e., “Strongly Approve,” “Approve,” “Undecided,” “Disapprove,” and “Strongly Disapprove”). [17] For example, respondents could indicate two works are “substantially similar,” “similar,” “undecided,” “not very similar,” and “not at all similar.” Respondents who select “substantially similar” would yield a percentage of “test-cell” respondents who find the works substantially similar.”[18]
Surveys related to the “substantial similarity” inquiry can be highly customizable depending on a circuit’s analysis.[19] For instance, under the Ninth Circuit’s extrinsic/intrinsic test,[20] a survey could focus on protectable elements of the copyrighted work using control cells and control stimuli.[21] Much like the use of control cells and stimuli in a secondary meaning or genericism surveys common in trademark litigation,[22] control stimuli could remove unprotected elements from the copyrighted work, and a control cell of respondents would compare this to the allegedly infringing work. A control cell of respondents who see the control stimuli and the allegedly infringing work would then indicate if they believe these works are “substantially similar.” The resulting control cell percentage, indicating perceived substantial similarity, could be subtracted from the test cell percentage to create a “net” figure of individuals who find the works substantially similar based on only protected elements, and eliminating “noise” attributable to non-protectable elements.[23]
The use of survey evidence in copyright cases to establish substantial similarity is not novel.[24] Similar surveys have been conducted[25] and submitted as evidence, often facing judicial criticism.[26] First, courts argue that the “lay observer” is in reference to the trier of fact, and doesn’t need to bear any relation to either the plaintiff or the defendant’s likely market for the copyrighted material,[27] nor does the trier of fact need to resemble the general population at large.[28] Second, courts have argued that “substantial similarity” is a fact-intensive inquiry that can only be determined upon reviewing significant evidence during trial.[29] Additionally, surveys may fail to properly establish “substantial similarity” if they fail to properly define the concept to respondents, as seen in some cases where surveys used trademark law analogies.[30]
However, these critiques can be addressed through proper survey design. First, the idea that the jury or judge can properly apply the “lay observer” standard itself has been met with criticism. Jurors (and judges) may not properly reflect the populace at large, and jurors (and judges) are not properly lay observers, as they have been made aware of facts and analyses during the course of litigation that may lead to their over-analysis of the works.[31] Survey evidence could provide a more accurate analysis of the lay observer’s perspective by avoiding any danger of over-analysis.
Second, while “substantial similarity” is a fact-intensive inquiry, so is determining a likelihood of confusion in trademark law.[32] If it works in trademark, why not in copyright? Additionally, the goal in the use of survey evidence would not be to supplant the perceptions of the jury or judges, but to provide additional real-world perspectives. Third, concerns about improperly defining “substantial similarity” can be mitigated by designing surveys that define key terms for respondents with specificity according to a jurisdiction’s prior precedents, thus being particularly relevant to the questions at hand.
Conclusion
The use of survey evidence on the matter of substantial similarity is likely to be an uphill battle. However, in breaking down this barrier, survey evidence in copyright infringement matters could prove helpful in pre-litigation in assessing a claim’s likelihood of success. For instance, surveys could assist in determining “striking similarity” so as to preclude independent creation[33] at the summary judgment stage to determine whether a reasonable jury could find infringement.[34] For a concept so criticized by the legal community, academics, and creators, the use of surveys to provide evidence that two works are substantially similar could provide much needed clarity in copyright litigation for judges, juries, and litigants alike.[35]
[1] See generally J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition, § 32:158 (5th ed. 2024) (“In fact, lawyers and judges are in most cases not representative at all of the reasonably prudent buyer for the goods and services in question. Lawyers and judges are too analytical, too likely to think the questions through logically. That’s not the way most reasonably prudent buyers react to trademarks they see in advertising and sales.”).
[2] See generally David Franklyn & Adam Kuhn, The Problem of Mop Heads in the Era of Apps: Toward More Rigorous Standards of Value Apportionment in Contemporary Patent Law, 98 J. Pat. & Trademark Off. Soc’y 182 (2016).
[3] Malla Pollack, Litigating the Right of Publicity: Your Client’s Face Was on the News, Now It’s on T-Shirts and Video Games, 119 Am. Jur. Trials 343, § 66 (2024).
[4] See Arnstein v. Porter, 154 F.2d 464, 468 (2d Cir. 1946) (in determining unlawful appropriation, “the test is the response of the ordinary lay hearer; accordingly, on that issue, ‘dissection’ and expert testimony are irrelevant.”); see William F. Patry, Patry on Copyright, § 9:83 (3d ed. 2024) (“While there is no per se rule barring survey evidence in copyright infringement actions, their admissibility has been rejected.”); but see Navarro v. Procter & Gamble Co., 501 F. Supp. 3d 482, 486 (S.D. Ohio 2020) (finding expert witness’ survey and testimony was relevant to the issue of damages in a copyright infringement matter).
[5] See Jamie Lund, Fixing Music Copyright, 79 Brook. L. Rev. 61 (2013) (arguing that use of tailored survey evidence in regards to substantial similarity of musical compositions “would be more statistically sound than administering the Lay Listener Test to a jury or panel of judges.”).
[6] Clark D. Asay, An Empirical Study of Copyright’s Substantial Similarity Test, 13 UC Irvine L. Rev.35, 43–50 (2022).
[7] Elements of copyright infringement claim–Substantial similarity, 6A Fed. Proc. Forms§ 17:71 (2024) (“To gauge substantial similarity between a copyrighted work and an allegedly infringing work, courts consider whether the accused work is so similar to the plaintiff’s work that an ordinary reasonable person would conclude that the defendant unlawfully appropriated the plaintiff’s protectible expression by taking material of substance and value.”).
[8] Lydia Pallas Loren & R. Anthony Reese, Proving Infringement: Burdens of Proof in Copyright Infringement Litigation, 23 Lewis & Clark L. Rev. 621, 641–42 (2019).
[9] Id. at 643–45.
[10] See Swirsky v. Carey, 376 F.3d 841, 845 (9th Cir. 2004) (quoting Three Boys Music Corp. v. Bolton, 212 F.3d 477, 485 (9th Cir. 2000)) (“The extrinsic test requires ‘analytical dissection of a work and expert testimony.’”).
[11] Narell v. Freeman, 872 F.2d 907, 910 (9th Cir. 1989) (“Because in most copyright cases direct evidence of copying is not available, a plaintiff may establish copying by showing that the infringer had access to the work and that the two works are substantially similar.”); John Bourdeau et al., Substantial similarity, 18 Am. Jur. 2d Copyright and Literary Property § 259 (2d ed.); Loren, supra note 8, at 638–48.
[12] This approach is similar to those utilized in Fame surveys to establish whether a mark is famous by “the general consuming public” as to be entitled protections under the Federal Trademark Dilution Act, 15 U.S.C. § 1125(c). See Shari Siedman Diamond, Surveys in Dilution Cases, Trademark and Deceptive Advertising Surveys 163, 164-66 (2d. ed. 2022).
[13] See generally McCarthy, supra note 1, at § 32:174.
[14] See generally id. at § 32:174.50.
[15] The use of Squirt surveys in trademark litigation is often controversial. See 24 Hour Fitness USA, Inc. v. 24/7 Tribeca Fitness, LLC., 447 F. Supp. 2d 266, 279 (S.D. N.Y. 2006). It could be argued that the survey should take other forms, such as a “two-room” Squirt survey (in which the two works are shown in separate “rooms,” creating more temporal space between the works) or an Eveready survey (in which only one of the works- the copyrighted or the allegedly infringing work – are shown to respondents). Additionally, as is common in the use of a Squirt, it could be argued that the use of multiple stimuli (i.e., various related third-party examples) would be necessary to remove the spotlight from the two works and, thus, be considered less leading. See Jerre B. Swann, Likelihood-of-Confusion Surveys, Trademark and Deceptive Advertising Surveys 59, 66–67 (2d. ed. 2022). However, the Squirt inquiry with no additional stimuli would seem to fairly replicate the analysis conducted by fact-finders in the determination of whether two works are “substantially similar.”
[16] See Warner Bros. Inc. v. Am. Broad. Companies, Inc., 720 F.2d 231, 244 (2d Cir. 1983) (criticizing the use of surveys on the issue of substantial similarity, and a particular survey’s inability to establish substantial similarity) (“We appreciate the problem plaintiffs confronted in designing the survey. If they asked too pointed a question, e.g., ‘Do you think Hinkley is substantially similar to Superman?’, they would have been faulted for using a leading question. On the other hand, their open-ended questions caused Chief Judge Motley to rule the entire survey inadmissible because the responses were ‘too general.’”).
[17] See Rensis Likert, A technique for the measurement of attitudes, Archives of Psychology 140, 15-20 (1932).
[18] The utilization of a test-cell and a control-cell would likely be important in an experiment establishing whether two works are substantially similar to remove several categories of “noise.” See Shari Siedman Diamond, Control Foundations: Rationales and Approaches, Trademark and Deceptive Advertising Surveys 239, 241–44 (2d. ed. 2022).
[19] John Gladstone Mills III et al., Exploiting copyrights—Recourse against infringement—What constitutes copyright infringement—Direct infringement—Objective aspect—”Substantial similarity,” the test of copyright infringement, 2 Pat. L. Fundamentals § 6:70 (2d ed. 2024) (“Several approaches, some interrelated, are taken by courts in analyzing works for substantial similarity. These approaches include applying: an ‘ordinary observer’ test, ‘a more discerning ordinary observer” test, total content and feel, an extrinsic/intrinsic test, and abstraction/filtration/comparison analysis.’”).
[20] Cavalier v. Random House, Inc., 297 F.3d 815, 822 (9th Cir. 2002) (“[W]hen applying the extrinsic test, a court must filter out and disregard the non-protectible [sic] elements in making its substantial similarity determination.”).
[21] See Diamond, supra note 18, at 240 (“A similar need for a standard for comparison in the form of a control cell or cells or in the form of a control stimulus arises in surveys assessing genericism or secondary meaning. We need to assess whether consumer responses in the survey reflect the construct of interest (e.g., for secondary meaning, that the consumer believes that the mark indicates that the product comes from a single source; for genericism that the consumer believes that the mark as applied to the product indicates that it is a brand rather than a “common” name) or whether they have been produced by noise due to guessing or some other source (e.g., a belief by some consumers that all vodka or tomato sauce is made or authorized by the same company).”).
[22] See id.
[23] See id. at 247.
[24] See Lund, supra note 5, at 96; see Ideal Toy Corp. v. Kenner Prods. Div. of Gen. Mills Fun Grp., Inc., 443 F. Supp. 291, 304 (S.D.N.Y. 1977) (“The defendants argue that the Court should look to the MISI survey as an indication that lay observers find a ‘substantial similarity’ between the Ideal figures and the characters in ‘Star Wars.’ Court ultimately found several flaws in the survey, and concluded that the survey’s inquiry into ‘substantial similarity’ through rates of ‘association’ were insufficient to establish substantial similarity.”).
[25] See Aaron Basha Corp. v. Felix B. Vollman, Inc., 88 F. Supp. 2d 226 (S.D.N.Y. 2000); see also Ideal Toy Corp., 443 F.Supp. 291.
[26] See Aaron Basha Corp., 88 F. Supp. 2d at 232 (“Therefore, because Basha’s survey relied on an imprecise definition of substantial similarity, this Court declines to consider this survey evidence in resolving Vollman’s motion for partial summary judgment.”).
[27] See Patry, supra note 4 (“Copyright infringement asks the trier of fact to determine whether the defendant’s work actually copied a material amount of expression; it is not the likely reaction of the marketplace, but rather the actual reaction of the flesh and blood trier of fact that matters in copyright infringement actions.”).
[28] See id. (“The trier of fact need not be representative of the market for the parties’ works, and indeed may have no familiarity with them at all. The fact finder may be wildly atypical of the population in general: no matter.”).
[29] See Warner Bros. Inc., 720 F.2d at 245.
[30] See Aaron Basha Corp., 88 F. Supp. 2d at 232 (“Therefore, because Basha’s survey relied on an imprecise definition of substantial similarity, this Court declines to consider this survey evidence in resolving Vollman’s motion for partial summary judgment.”); see Patry, supra note 4.
[31] See Edward Lee & Andrew Moshirnia, Do Experts Matter? A Study of the Effect of Musicologist Testimony in Music Cases, 2022 U. Ill. L. Rev. 707, 803 (2022); see also Arnstein, 154 F.2d at 468 (Clark, J., dissenting).
[32] Daryl Lim, Trademark Confusion Revealed: An Empirical Analysis, 114 Trademark Rep.. 799, 814–15 (2024) (based on an empirical analysis, most courts apply what are known as the Polaroid factors, proffered by the court in Polaroid Corp. v. Polaroid Electronics Corp., 287 F.2d 492, 495 (2d Cir. 1961), which include evaluating “(1) strength of the plaintiff’s mark; (2) similarity of plaintiff’s and defendant’s marks; (3) competitive proximity of the products; (4) likelihood that plaintiff will “bridge the gap” and offer a product like a defendant’s; (5) actual confusion between products; (6) good faith on the defendant’s part; (7) quality of defendant’s product; and (8) sophistication of the buyers.”).
[33] See Repp v. Webber, 132 F.3d 882, 889 (2d Cir. 1997).
[34] See teamLab Inc. v. Museum of Dream Space, LLC, 650 F. Supp. 3d 934, 951 (C.D. Cal. 2023) (citing Unicolors, Inc. v. Urban Outfitters, Inc., 853 F.3d 980, 985 (9th Cir. 2017)) (“If the Court finds that the extrinsic test is met, then the matter survives summary judgment and proceeds to trial where the jury applies the intrinsic test, which looks to whether the ordinary, reasonable person would find “the total concept and feel of the works to be substantially similar.”).
[35] For criticisms of a jury’s ability to properly assess substantial similarity, see Zahr K. Said, Jury-Related Errors in Copyright, 98 Ind. L.J 749 (2023); see Lee, supra note 31.