Anarchy Has Ensued In Courts’ Handling of Online Contract Formation (Round Up Post)
[Eric’s introduction: ever since the Ninth Circuit mishandled the already-muddled definitions of “clickwrap” and “browsewrap” in the Nguyen case, we’ve seen a steady decline in the coherence of the law of online contract formation. Judges can’t figure out whether or not the contract at issue is a “clickwrap” even when it’s obviously a mandatory clickthrough agreement, so they are characterizing the contract as somewhere in between prior -wrap definitions, proliferating new -wrap terms, and imposing increasingly exacting standards before finding a properly formed online contract. The result is more contract formations are failing, less predictability and a surplus of new -wrap terminology to contend with.
The online contract formation cases are coming out at an accelerating pace, so we can’t keep up with them all. In this post, Venkat will start us off with the Second Circuit’s latest ruling, which possibly undermines Amazon’s user agreement for hundreds of millions of users, and then he and I will recap other recent cases. If you are hoping this post will leave you with some sense of where the law is, I anticipate you’ll be disappointed. A hazy fog of anarchy has descended upon the online contract formation jurisprudence.]
In July, Eric and I “celebrated” TOS Arbitration day by blogging a couple decisions addressing arbitration in online contracts. Perhaps we should make it a monthly or a quarterly thing, because the decisions keep coming….
Nicosia v. Amazon, 2016 WL 4473225 (2d Cir. Aug. 25, 2016): This is a lawsuit over dietary supplements purchased via Amazon, which comes to us from the Second Circuit. Plaintiff asserted claims under the Consumer Product Safety Act and state law. He also sought an injunction preventing Amazon from making continued sales.
Amazon filed a motion to dismiss based on the arbitration clause in its terms of service, but it did not ask the court to compel arbitration. The district court, applying the motion to dismiss (rather than the summary judgment) standard, dismissed the claims in favor of the arbitration clause. (My blog post on that ruling here.)
There was one catch. Nicosia started his Amazon account in 2008, and the 2008 terms did not contain an arbitration provision. Amazon added mandatory arbitration to a later version of its terms, but it was not apparent that Nicosia had agreed to the later terms. Amazon’s order page in effect when Nicosia ordered the weight loss supplement product (in 2013) says “by placing your order, you agree to Amazon’s privacy notice and conditions of use”. But there was nothing on the order page that required a user to say he or she agreed to the terms. Nor was there anything that indicated Amazon had provided notice to its customers of changes to the terms.
The court affirms the denial of injunctive relief. But Amazon has less success on the arbitration issue. Drawing on the Second Circuit’s painful “apple cart” analogy from its 2004 Register.com v. Verio decision, the court asks whether Amazon, as the apple stand owner, had provided Nicosia sufficient notice:
it is as if an apple stand visitor walks up to the shop and sees, above the basket of apples, a wall filled with signs. Some of those signs contain information necessary for her purchase, such as price, method of payment, and delivery details, and are displayed prominently in the center of the wall. Others she may quickly disregard, including advertisements for other fruit stands. Among them is a sign binding her to additional terms as a condition of her purchase. Has the apple stand owner provided reasonably conspicuous notice?
We think reasonable minds could disagree.
In a seeming effort to streamline customer purchases, Amazon chose not to employ a clickwrap mechanism. While clickwrap agreements that display terms in a scrollbox and require users to click an icon are not necessarily required, see Register.com, 356 F.3d at 403 (an offeree need not specifically assent to certain terms by clicking an ʺI agreeʺ icon so long as the offeree ʺmakes a decision to take the benefit with knowledge of the terms of the offerʺ), they are certainly the easiest method of ensuring that terms are agreed to, see Starkey, 796 F.3d at 197 n.3 (noting that it would have been ʺsimpler to resolveʺ this question had a clickwrap mechanism been used).
To be clear, we do not hold that there was no objective manifestation of mutual assent here as a matter of law. Rather, we conclude simply that reasonable minds could disagree on the reasonableness of notice. See Cascade Auto Glass, Inc. v. Progressive Cas. Ins. Co., 135 Wash. App. 760, 767 (2006) (ʺWhether particular notice was reasonable is ordinarily a question of fact for the jury.ʺ). We therefore hold that Amazon has failed to show that Nicosia was on notice and agreed to mandatory arbitration as a matter of law. The district court thus erred in concluding that Nicosia had failed to state a claim under Rule 12(b)(6).
Plot twist: In passing the court notes that even if the account was created in Nicosia’s name, “Nicosia could have used a shared account created by a member of his family to make his purchases.”
Tompkins v. 23andMe, Inc., No. 14-16405 (9th Cir. Aug. 23, 2016); This is a lawsuit against 23andMe, which has had its share of legal troubles. The FDA first told 23andMe to discontinue its practice of making health claims, and plaintiffs piled on. Numerous plaintiffs’ suits were heard as an MDL in the Northern District of California, and 23andMe moved to compel arbitration. The district court (Judge Koh) found that while the plaintiff’s initial purchase did not require him to agree to the terms, his later registration on 23andMe’s website (to view the results) sufficiently bound him to the terms. Plaintiff also argued that 23andMe’s terms, which contained an arbitration clause, were unconscionable. Judge Koh rejected this argument as well. (Here is Eric’s blog post on that ruling: “23andMe’s Browsewrap Fails, But Its Post-Purchase Clickthrough Works Anyway–Tompkins v. 23andMe“.)
On appeal, the 9th Circuit skips over the contract formation issues entirely (!), and focuses on the question of unconscionability. It finds that the following terms do not render the clause unconscionable:
- a (bilateral) prevailing party clause
- requirement that the loser pays the costs of filing
- a forum selection clause requiring disputes to be resolved in San Francisco
- a carveout exempting intellectual property claims from arbitration
- a one year limitations clause
- a term that 23andMe can freely revise the terms of the agreement
Judge Watford concurs in the affirmance, but on narrower grounds. 23andMe waived its right to enforce the fee shifting clause, so he would not consider it. And he did not view the venue clause as unconscionable as to these plaintiffs. Similarly, as to the carveout argument, there was no showing that their application would be unconscionable in this case. Finally, and this is a big one, he notes that the dispute would proceed to arbitration on a class basis.
Bekele v. Lyft, 2016 U.S. Dist. LEXIS 104921 (D. Mass. Aug. 9, 2016): This is a district court case where drivers assert misclassification claims against Lyft. In resolving a motion to compel arbitration, the court focused on three issues: (1) whether the drivers agreed to be bound to the arbitration clause; (2) whether the agreement was unconscionable; and (3) whether the class action waiver was invalid because it violated the NLRA’s rules restricting collective bargaining limitations. (I’ll focus on the first two questions, but the third could end up being the focus of an appeal down the road, given the varying appellate court conclusions on this point.)
On the first question, the court says that the touchstone is reasonable notice, and here, Lyft provided ample notice. The court notes that Lyft’s sign-up process was leakproof, and plaintiff’s manifestation of assent was unequivocal:
the implications of clicking the ‘I accept’ button were clear.
Having concluded this, the court has no trouble rejecting plaintiff’s arguments around the sign-up process. In fact, the court compares Lyft’s sign-up process to Uber’s and says the expression of assent at the point of signing up is even clearer in Lyft’s case than Uber’s.
Second, plaintiff argued that the arbitration clause and the way it was presented was procedurally unconscionable because the clause was “buried in a lengthy document that was presented on a tiny smartphone screen.” This does not sway the court. Plaintiff also argued that he lacked the ability to freely read the document, but the court says this is not credible. He could take his time and read the document at his leisure before deciding to become a Lyft contractor (or as he argues, employee). It’s unclear what Lyft could have done differently. Even emailing plaintiff a copy of the agreement and requesting a “wet” signature would not have added any benefits to the process.
Because the court finds that the clause was not procedurally unconscionable, it declines to address plaintiff’s arguments on substantive unconscionability. In passing, the court notes that the unilateral-modification clause does not undermine the arbitration clause because it’s a part of the agreement overall and does not enter into the court’s assessment on unconscionability, which is focused on the arbitration clause. (For what it’s worth, the court notes that Massachusetts law differs from California law on unconscionability. California employs a “sliding scale” approach under which the court typically considers both substantive and procedural unconscionability.)
Doe v. Xytex Corp., 2016 U.S. Dist. LEXIS 94213 (July 19, 2016): in a venue dispute in a case against a website that sold semen for artificial insemination, Judge Alsup orders a mini-hearing on the appearance of the website and whether it gave plaintiffs sufficient notice:
Relying on the assumption that the website looked as represented herein in 2004, this order would hold that the Xytex failed to give plaintiffs constructive notice that the “Site Usage” button included a hyperlink to the terms of a contract that supplied the terms for use of Xytex’s website and services. Nevertheless, the record is inadequate to establish whether the site, in fact, appeared as set forth above. At oral argument, counsel for plaintiffs specifically asked for leave to take a Rule 30(b)(6) deposition regarding Xytex’s position on the appearance of the website in 2004. Plaintiffs shall be permitted to take two four-hour depositions and to propound up to eight requests for the production of documents directed solely at that question. As discussed below, Xytex shall also be permitted to depose both plaintiffs and request the production of documents. That discovery may address this issue as well as the issue of actual notice. (The burden is on Xytex to prove up the existence of the agreement to select a forum.)
Kearney v. Okemo LLC, 2016 WL 4257459 (D. Vt. Aug. 11, 2016) [this case recap written by Eric]: This is a personal injury lawsuit. The plaintiff claims that a ski resort negligently installed safety netting, which caused him injuries during a ski race competition. The defendants invoke a well-drafted release that the plaintiff purportedly assented to while signing up for a US Skiing & Snowboarding Association membership online. The court says:
Plaintiff admits that he applied for a USSA membership online, but states that he has no recollection of seeing or acknowledging the release. He attempts to create a factual dispute by asserting that Defendants have yet to produce a release signed or initialed by him. However, unlike when a person physically signs a paper contract, such documentation does not necessarily exist in the click-wrap context. Nevertheless, courts frequently enforce such agreements.
To provide the proper evidence of a clickthrough agreement, the USSA brought in an IT representative:
Ms. Alexandrescu testified that she has been familiar with USSA’s website and the online membership process since its inception in 2008. She produced demonstrative exhibits of the release currently in use by USSA and testified that the same agreement has been used in USSA’s online membership process since 2008. All online registrants since 2008 have been required to read and acknowledge the release by checking a box that states, “I HAVE CAREFULLY READ THE FOREGOING AND UNDERSTAND IT TO BE A LEGALLY BINDING RELEASE AND INDEMNITY AGREEMENT.” Ms. Alexandrescu testified that the USSA website has never permitted a registrant to become a USSA member without checking this box. If a registrant were to not agree to the release and leave the box unchecked, the page with the release would continuously reload and prompt the registrant to check the box. Only after checking the box would the registrant be permitted to complete the membership process. Upon completion of the process, the registrant receives a confirmation receipt, welcome letter, and USSA membership number.
The court says it’s heard enough:
It is undisputed that Plaintiff applied for and received his USSA membership online in December 2014. Plaintiff admits that he received a confirmation email from USSA and that his credit card statement reflects a payment for his USSA membership. Though he does not remember whether he saw or acknowledged the release, Plaintiff has offered no evidence and asserted no specific facts to rebut Defendants’ evidence that one cannot receive a USSA membership, as Plaintiff did, without first accepting the release….The court finds that no triable issue exists concerning Plaintiff’s acceptance of the release.
But then…the plot twist. The court says that it will disregard the contract’s choice of law provision and apply instead the law of Vermont, where the accident occurred. Vermont law declares releases in ski resort agreements void, so the court voids the contract’s release. So the defendants’ joy over having a binding contract was short-lived. Without the release, the odds of the plaintiff getting cash go up substantially.
None of these rulings are particularly novel, but they still show that companies–even sophisticated ones–can struggle with the online contracting process. The cases all don’t deviate from the key rule that if you force users to agree to the terms using an expression of assent that is unequivocal, plaintiffs will have little success arguing against contract formation. Once the court finds a valid contract, unconscionability is a pretty tough argument to make. The language surrounding the check-the-box matters, and courts are perfectly willing to nitpick on this point. Lyft passes the test, but Uber’s recent experience in front of Judge Rakoff went the other way. (See “Judge Declines to Enforce Uber’s Terms of Service–Meyer v. Kalanick“. For what’s its worth, there are a slew of rulings across the country involving Uber’s and Lyft’s terms, with slightly varying results.)
The Amazon ruling is interesting because it touches on the question of notice and how a company can update its agreement and have assurances that all users have effectively agreed to the new terms. This is a perennial question that (as far as I know) has not resulted in an answer that has received judicial blessing. Amazon recently won an arbitration ruling in the Southern District of California where the court concluded that the plaintiff was bound by the terms despite no check-the-box. (See “Court Enforces Arbitration Clause in Amazon’s Terms of Service–Fagerstrom v. Amazon“.) That decision is on appeal to the Ninth Circuit. (This ruling also dealt with various consumer-friendly terms and whether the arbitration clause was unconscionable under California law.) For whatever reason (probably user experience-driven) Amazon still does not force users to check the box at the point of ordering. Eric blogged about how Zappos’ contracting failures left it exposed, this could be the same result for Amazon. I wonder if Amazon is pushing the envelope on this question and hoping to end up with favorable precedent?
The 23andMe ruling was surprising in its failure to address the online contracting rules. As Eric explained in his blog post, there was a lot to talk about there, and Judge Koh made a few leaps that the appeals court could have taken a close look at. Plus it may have been an opportunity to clarify the law. Nevertheless, the unconcsionability analysis is fairly useful. Companies vary in how consumer-friendly they make contracts, and its useful to have decisions addressing these issues.
Some other open questions: (1) what type of evidence a court looks to in determining whether a party agreed to on-line terms (as Eric notes, this is becoming increasingly important); (2) whether contracts deployed via smartphones are somehow different; and (3) the status of class action waivers.
I have to agree with Eric’s take on the apple stand analogy. It must go.
* I was stunned that the Tompkins v. 23andme ruling did not address the contact formation piece. The case had the perfect fact pattern for testing when a physical goods manufacturer formed its contract with its users. Was it at the time of purchase from a retailer? When the consumer opened the physical package? When the consumer logged into the website to get the test results? Never? The Ninth Circuit’s opinion assumes proper contract formation–which was the only legal question really in doubt–and then dismisses the unconscionability challenge. The unconscionability discussion wasn’t ground-breaking but it had plenty of favorable language for vendors–especially the implicit endorsement of the 1 year statute of limitations clause, which is a powerful and controversial risk management technique.
* It’s no longer good enough to deploy a properly implemented online clickthrough agreement. You also need to prove it in court with admissible evidence. Sophisticated practitioners are now thinking ahead to gather and organize the evidence they need to convince the court that their clickthrough worked properly and which terms applied at the relevant timeframe.
* If courts routinely cannot determine if a contract is a “clickwrap,” “browsewrap,” both or neither, IT’S NOT A HELPFUL DOCTRINAL TOOL.
* Because no one knows what constitutes a “clickwrap” or “clickthrough” agreement any more, online contracts will be getting more and more clicky. #GettinClickyWitIt So if you’re finding it harder to get from point A to point Z online, you have the degradation of the -wrap nomenclature to thank. #2ClicksIsTheNewBlack #ClickInflation
* Courts are also paying closer attention to the specific words of the call-to-action that elicits the consumer’s click. Make your “if” and “then” statements very, very explicit and foolproof.
* I can’t quite figure out the implications of the Nicosia court’s discussion about the possibility that the plaintiff piggybacked on someone else’s account. The court says: “The Amazon declarations assert that: (1) to make a purchase on Amazon.com, a registered account had to be used; and (2) Nicosia’s purchases were made using an account created in 2008. Even assuming these statements to be true, they do not exclude the possibility that Nicosia used an account that he did not create. Nicosia could have used a shared account created by a member of his family to make his purchases.” The whole point of doing clickthrough agreements is that it moots the need to link any specific person to the legally significant click.
Here, if the plaintiff used someone else’s account, does he even have privity with Amazon, or would the accountholder be the proper plaintiff? Furthermore, the appellate court is responding to Amazon’s motion to dismiss, so the court doesn’t have to resolve factual ambiguities. However, this factual issue is definitively resolvable. Ultimately, Nicosia will have to prove that he made a purchase from Amazon, so eventually we should know exactly what account he used to make the purchase. And if Nicosia had his own account but nevertheless made the purchase under a different person’s account, he might still be bound by the terms applicable to the account he formed. Finally, note that if Nicosia really did piggyback on a third party’s account but most class members are in a different position, Nicosia may not be an appropriate class representative and there could be serious commonality problems with class formation.
Still, the court’s casualness here opens up a difficult-to-resolve hole in clickthrough agreement. Plaintiffs can always claim that they used the online service piggybacking on a third party’s account and thus never agreed to the service’s online contract. (Note that if the online service’s user agreement restricted such piggybacking, both the plaintiff and the accountholder may have breached the contract and possibly committed federal crimes and may not want to admit this in their pleadings). If these piggybacking allegations are sufficient to survive a motion to dismiss, then online contract formation disputes inevitably will routinely drag on into summary judgment.
* The Nicosia court says “While the 2008 Conditions of Use did reserve Amazon’s right to change those terms at any time, this did not necessarily bind Nicosia to any change of terms without notice. Under Washington contract law, such unilateral modifications are only binding if there is notice and assent to the changed terms.” (Cf. the Blockbuster ruling and the many other cases in this genre).
The court provides a screenshot of Amazon’s order page that purportedly provided the requisite notice of the new terms:
The court explains it thinks this disclosure was insufficient (cites omitted):
Near the top of the page, below the “Review your order” heading, the critical sentence appears in smaller font: “By placing your order, you agree to Amazon.com’s privacy notice and conditions of use.” Add. B. The phrases “privacy notice” and “conditions of use” appear in blue font, indicating that they are clickable links to separate webpages….
Notably, unlike typical “clickwrap” agreements, clicking “Place your order” does not specifically manifest assent to the additional terms, for the purchaser is not specifically asked whether she agrees or to say “I agree.” Nothing about the “Place your order” button alone suggests that additional terms apply, and the presentation of terms is not directly adjacent to the “Place your order” button so as to indicate that a user should construe clicking as acceptance….
The message itself — “By placing your order, you agree to Amazon.com’s . . . conditions of use” — is not bold, capitalized, or conspicuous in light of the whole webpage. Proximity to the top of a webpage does not necessarily make something more likely to be read in the context of an elaborate webpage design. There are numerous other links on the webpage, in several different colors, fonts, and locations, which generally obscure the message. Although it is impossible to say with certainty based on the record, there appear to be between fifteen and twenty-five links on the Order Page, and various text is displayed in at least four font sizes and six colors (blue, yellow, green, red, orange, and black), alongside multiple buttons and promotional advertisements. Further, the presence of customers’s personal address, credit card information, shipping options, and purchase summary are sufficiently distracting so as to temper whatever effect the notification has.
So it sounds like the court wants the cross-reference to the new “conditions of use” to be more prominent, associated with a better call-to-action, and perhaps with fewer competing links or colors? Uh, OK. Could Amazon have improved this implementation? Sure. But did Amazon totally miss its mark? I think most of us would have assumed that Amazon’s current implementation worked (and it still could on summary judgment). Nevertheless, lesson learned: MAKE YOUR ONLINE CONTRACT FORMATION CLUELESS-PROOF, AND THEN DO IT ALL OVER AND MAKE IT EVEN MORE CLUELESS-PROOF.
Of course, unlike many other websites, Amazon has enough repeat usage from most customers–and all buyers–that it could easily force all users through a mandatory clickthrough of all new amended terms. It chose to cut that corner–for good reasons, but still–which has made its life harder in court. Unfortunately, all of us suffer another murky appellate court ruling on online contract formation as collateral damage.
* Because the court questions Amazon’s amendment process, the Nicosia ruling could mean that all of Amazon’s pre-2013 user agreements do not include the mandatory arbitration clause. Oof.
* If any student ever uses an apple stand analogy for online contract formation without mocking it, I pledge to give that student an automatic F. (Same deal applies to highway billboards like the Ninth Circuit’s Brookfield ruling). #NeverAgain