Fourth Circuit’s Rosetta Stone v. Google Opinion Pushes Back Resolution of Keyword Advertising Legality Another 5-10 Years

By Eric Goldman

Rosetta Stone Ltd. v. Google, Inc.,, 2012 WL 1155143 (4th Cir. April 9, 2012). I’m listed in the caption as a limited intervenor in the efforts led by Paul Levy to crack open the redacted material from the Joint Appendix. Paul Levy’s explanation of that effort.


Lawsuits over the sales of keyword advertising seem so…2005. In an era where everyone is freaking out about behavioral advertising, lawsuits over the comparatively innocuous practices of triggering ads based on search terms feel archaic at best. After the district court opinion in the Rosetta Stone v. Google case, which slammed the door hard on the most determined plaintiff still suing Google for its keyword ad sales, I thought most trademark owners had moved on to other worries.

Sadly, this ruling–giving Rosetta Stone another chance in court–will undoubtedly revive the trademark plaintiff bar’s interest in suing Google for its keyword ad practices. Just like Google got hit with over a dozen lawsuits in the wake of Google’s Second Circuit “loss” in the Rescuecom case, I imagine a bunch of low-merit suits will follow this ruling too. (I put Google’s “loss” in quotes because ultimately Rescuecom just gave up its lawsuit). Google will almost certainly win all of the incoming cases, just like it destroyed most of the cases filed post-Rescuecom, but it will have to spend a lot more money in legal fees to get there. Worse, it will take a while for these cases to work through the system, meaning that we aren’t going to reach the inevitable final equilibrium–that keyword ad sales don’t violate trademark law–for another 5-10 years. In this sense, the opinion reminded me of last week’s Viacom v. YouTube ruling–both opinions were soulless omnibus rulings that give superficial (but ultimately false) hope to plaintiffs while remanding the case to an almost certain defense win (just at a much higher cost).

Also like Viacom v. YouTube, the appeals court reversed because the district court’s opinion cut too many corners. I noted that when I blogged the district court’s ruling:

Unfortunately for Google, the opinion contains several minor doctrinal errors that could attract attention from an appellate court. That makes this ruling vulnerable on appeal.

Nevertheless, I’d hoped the appeals court would get to the right result based on the big-picture normative conclusion that keyword ad sales shouldn’t violate trademark law. The district court got the big picture right. The appellate ruling, in contrast, is so focused on correcting the doctrinal details that it completely misses the big picture. This is why the plaintiffs’ bar will be unable to resist probing the opinion’s interstices, even though those lawsuits will all eventually fail. If the court had more sensitivity to the big picture, it would have written an opinion that fixed the lower court’s doctrinal errors but better communicated to plaintiffs not to bother suing.

Specific Doctrinal Points

Direct TM Infringement. This is probably the most obvious place where the opinion mechanically marched through the doctrinal elements without acknowledging the big picture. It’s really makes no sense to hold that Google *directly* liable for trademark infringement for keyword ad sales, because Google’s products/services or marketing do not confuse consumers about the source and origin of marketplace offerings. If any consumer confusion takes place, it’s caused by the advertisers, and Google is at most secondarily liable for that. But the opinion is tone deaf about this key point.

On the plus side, the opinion correctly notes that the standard multi-factor likelihood of consumer confusion (LOCC) test doesn’t work in nominative use situations. This is interesting on two fronts. First, I have made this point numerous times, most recently in my post on AR Pillow v. Cottrell. I hope other courts similarly rethink the multi-factor LOCC test in nominative use cases. It really don’t work at all.

Second, the court implies–without saying–that Google’s activities could be nominative use. This is a point I made back in 2005 in my Deregulating Relevancy article . I hope courts take a closer look at the interplay between keyword advertising and nominative use, but Google (as opposed to its advertisers) isn’t referring to the plaintiff’s products so much as the class of those products (what I called “associative use” in my article), so a classic nominative use argument probably still fails. Courts should think more broadly about the semantic import of keyword triggering.

On the minus side, the opinion warns lower courts not to engage in a “robotic application of each and every factor in a case involving a referential, nontrademark use.” Well, that’s true, but if it’s truly a “nontrademark use,” then what the hell is everyone fighting about??? The lack of a clear concept of what constitutes a “trademark use in commerce,” and how that affects a direct trademark infringement analysis, is a key flaw in the opinion’s architecture.

On the LOCC’s intent factor, the court says that a jury could find Google had bad intent. The only evidence the appellate court discusses is that, in 2004, “internal studies performed by Google at this time suggested that there was significant source confusion among Internet searchers when trademarks were included in the title or body of the advertisements” and in 2009 Google liberalized its trademark policy to let more parties display third-party trademarks in the ad copy. I don’t see how that provides any evidence to support a *direct* trademark infringement because Google doesn’t put the trademarks in the ad copy (advertisers do).

On the LOCC’s actual confusion factor, the opinion says that a variety of evidence–Rosetta Stone’s deposition of five confused purchasers, evidence of other complaints Rosetta Stone received, Google’s 2004 study, Google’s in-house lawyers’ inability to review search results to determine if they were legit and Rosetta Stone’s expert report–all should have helped Rosetta Stone defeat summary judgment.

On the LOCC’s sophisticated purchaser factor, the opinion says that the district court’s inferences against Rosetta Stone weren’t appropriate for summary judgment. The court notes some evidence–the buyers who bought counterfeit software and Google’s study–could be counted against purchaser sophistication.

The Functionality Defense. One of the district court’s goofiest rulings was that keyword triggering qualified for trademark law’s functionality defense. Keywords do act as inputs into Google’s software, so in the colloquial sense they are “functional.” However, trademark law’s functionality defense is narrower than that, and in practice it really only should act as a defense to trade dress claims The appellate court unsurprisingly reverses the district court’s approval of a functionality defense emphatically. Among other things, the opinion says “it is irrelevant whether Google’s computer program functions better by use of Rosetta Stone’s nonfunctional mark.” Many trademark geeks and commentators have already rejoiced that the appellate court fixed this issue.

Contributory Infringement. The opinion defines the applicable legal standard:

It is not enough to have general knowledge that some percentage of the purchasers of a product or service is using it to engage in infringing activities; rather, the defendant must supply its product or service to “identified individuals” that it knows or has reason to know are engaging in trademark infringement.

This is an OK standard but it is slightly different from other recent articulations of contributory trademark infringement, and I’m not sure why this panel felt the need to re-express the test in its own words. The opinion says that the district court’s reliance on Tiffany v. eBay was misplaced because that ruling came after a bench trial and therefore couldn’t help on summary judgment. Unfortunately, the opinion gives short shrift to what evidence from Rosetta Stone could support contributory infringement. The only evidence cited by this opinion is the fact that Rosetta Stone identified some ads from rogue sellers and Google didn’t block other unidentified ads from those sellers. If that’s Rosetta Stone’s best argument for contributory infringement, I think it will lose the contributory claim handily.

Dilution. Characterizing keyword ad sales as dilution is unprecedented and virtually unsupportable doctrinally. This opinion doesn’t address that. It simply points out minor doctrinal errors in the lower court’s opinion (not surprising, as I said the lower court’s “rejection of the dilution claim was bizarre”), but misses the big picture that there’s absolutely nothing Google does that could dilute or tarnish Rosetta Stone’s marks. (The court also doesn’t discuss contributory dilution, but almost all courts have said it doesn’t exist, so if the advertisers are diluting Rosetta Stone’s marks but Google itself isn’t, Rosetta Stone still loses this point).

The opinion says the lower court misallocated the burden about any fair use exclusion to dilution; the burden is on the defendant to show the defense, and the lower court should have held the issue past summary judgment. The appellate court’s implication that keyword ad sales might be nominative use looms large here, but nominative use would be a complete defense to the dilution claim. The opinion also says the lower court truncated its analysis on the likelihood of dilution. Finally, the opinion tells the lower court to reexamine the dates on which Google commenced use of Rosetta Stone’s trademark and when Rosetta Stone’s trademark became famous. I don’t see any way for Rosetta Stone to win the dilution claim, and nothing in the appellate court’s doctrinal nitpicking changes my assessment.

Other. The court dismisses the vicarious trademark claim because Rosetta Stone didn’t provide evidence that “Google acts jointly with any of the advertisers to control the counterfeit ROSETTA STONE products.” The court also dismisses the unjust enrichment claim because “Rosetta Stone failed to allege facts showing that it “conferred a benefit” on Google for which Google “should reasonably have expected” to repay.” The district court partially relied on 47 USC 230 to dismiss the unjust enrichment claim; the opinion expressly leaves the 230 point unresolved.

Case Library

Bonus: with the help of an RA, we scanned and posted the material from the Joint Appendix in this case. Some day I hope to blog on some of the more interesting findings (but nothing earth-shattering).

The case library (see also the Joint Appendix material):

* The Fourth Circuit’s opinion. Blog post.

* Public Citizen’s motion (with Marty Schwimmer and me) to intervene and request to unseal the joint appendix.

* Rosetta Stone reply brief.

* Public Citizen amicus brief in support of Google.

* Public Knowledge/EFF amicus brief in support of Google.

* eBay/Yahoo amicus brief in support of Google.

* Google’s opening response brief: redacted and unredacted (warning: 60MB file).

* UK Intellectual Property Law Society amicus brief in support of neither party.

* Rosetta Stone’s opening appellate brief: redacted and unredacted.

* INTA’s amicus brief in support of Rosetta Stone.

* Carfax et al amicus brief in support of Rosetta Stone.

* Association for Competitive Technology et al amicus brief in support of Rosetta Stone.

* ConvaTec et al amicus brief in support of Rosetta Stone.

* Volunteers of America amicus brief in support of Rosetta Stone.

* District court’s main opinion granting SJ. My blog post.

* District court’s opinion granting a motion to dismiss on the unjust enrichment claim.

* Rosetta Stone’s initial complaint. My blog post.