December 24, 2010
Deep Packet Inspection (NebuAd) Litigation: Court Dismisses ECPA Claim but CFAA Claim Continues
[Post by Venkat with comments by Eric]
Mortensen v. Bresnan Comm., CV 10-13-BLG-RFC (D. Mont. Dec. 13, 2010)
A district court in Montana hearing one of the many NebuAd "deep packet inspection" lawsuits partially granted a defendant's motion to dismiss. This lawsuit arises out of NebuAd's alleged attempt to monitor and use an end user's internet activity for advertisement targeting purposes - i.e., not using cookies or other tracking, but actually routing the communications themselves through NebuAd's "appliance." There have been a slew of lawsuits out of this practice; this lawsuit involved claims against Bresnan Communications, an Internet access provider, who is accused of letting NebuAd install the appliance for its profit.
Electronic Communication Privacy Act Claims: Bresnan first argued that it did not engage in any interception itself, so it could not be held liable under the ECPA. The court rejects this argument on the basis of plaintiff's allegation that Bresnan "allowed" NebuAd to install its device on Bresnan's network, and but for the appliance, the monitoring would not have occurred.
However, the court accepts Bresnan's argument that the plaintiffs agreed to the interception based on disclosures in the terms of service and elsewhere. The court quotes from Bresnan's "Online Privacy Notice," which says:
the equipment used to provide the service collects information . . . [including] information about . . . 'electronic browsing,' and the text of email or other electronic communications the [users] send or receive using [the] services.
The notice also references that the information that is collected will be disclosed to third parties. Bresnan's "Online Subscriber Agreement" contained similar disclosures. Finally, the court notes that Bresnan alleges that it provided customers "specific notice" and a link to opt-out from information collections.
Shockingly, plaintiffs did not contest that "they agreed, by way of Bresnan's Privacy Notice and Subscriber Agreement to the interception." (??) Instead, plaintiffs quibble with the scope of the documents in question and argued that Bresnan construes plaintiffs' consent "cavalierly." The court rejects plaintiffs' argument, and grants Bresnan's motion to dismiss the ECPA claim on the basis of consent.
Invasion of Privacy Claims: Plaintiffs brought a common law invasion of privacy claim. The court finds that the notice and disclosure (discussed above) undermines any expectation of privacy plaintiffs had in their use of the service. This ends the court's discussion.
Computer Fraud and Abuse Act Claims: Although the court rejects plaintiffs ECPA claim, the court allows plaintiffs' Computer Fraud and Abuse Claim to go forward. The court concludes (based on Bresnan's disclosures to its customers) that Bresnan's access of plaintiffs' computers had some authorization. Nevertheless, the court finds that Bresnan may have exceeded the authorization that was initially granted. The court bases this conclusion on the fact that the notices provided by Bresnan did not clearly apprise plaintiffs that "their computer settings were to be actively altered or tampered with by Bresnan." The court concludes that for purposes of surviving a motion to dismiss, plaintiffs have sufficiently alleged that:
Bresnan's act of tampering with the security and privacy protocols exceeded any authorization that Plaintiffs may have given.
The court also addresses the jurisdictional damage requirement, under which a CFAA plaintiff must show that the unauthorized access caused $5,000+ in damages. The court notes that plaintiffs' allegations of emotional distress are not compensable, since only economic losses are recoverable under the CFAA. However, the court finds that plaintiffs satisfy the jurisdictional damage threshold since they allege they were "forced to mitigate Bresnan's invasive actions by expending time, money and resources to investigate and repair their personal computer's diminished performance."
Trespass to Chattels: Finally, the court allows plaintiffs' trespass to chattel claims to go forward. With respect to the trespass claim, the court says that the plaintiffs sufficiently alleged an interference with their chattel (their computers).
This is one of many privacy lawsuits that are percolating through the courts right now. I think this one differs qualitatively from many of the others in that here, there is an allegation of improper monitoring of the contents of the plaintiffs' communications. It's one thing to surreptitiously find out what websites someone has been visiting or leak someone's unique user ID. It's another thing entirely to read their email and the contents of what they access while browsing. This is an important distinction to keep in mind. I don't think you can necessarily extrapolate a tentative result in the other cases based on this result. Apart from the damages issue (discussed below) a key unknown in the pending cases is to what extent the information that is captured or disclosed are covered by the statutes in question.
I was somewhat surprised to see little or no discussion from the court on whether the policies were presented in a "leak proof" manner, or whether the disclosure satisfied FTC standards. Was there evidence that plaintiffs could not access the service without encountering the policy? (See Prof. Goldman's post on that topic: "Clickthrough Agreement With Acknowledgement Checkbox Enforced.")
The court's conclusion on the consent issue is also somewhat perplexing, in light of the exact same judge's earlier order denying Bresnan's request to compel arbitration, which you can access here. BNA recaps the decision denying Bresnan's request to subject the claims to arbitration as follows: "A mandatory arbitration clause in an internet service provider's terms of service—which was presented in capitalized text in the ninth paragraph of the unsigned document—was an inconspicuous part of a contract of adhesion and unenforceable under Montana law."
On the other hand, if plaintiffs conceded the consent/disclosure issue, then the court did not need to get into it. [What were the plaintiffs thinking, conceding this? If you are bringing this type of a lawsuit, you have to be able to put together enough allegations of no-consent to get past the motion to dismiss stage.]
At the end of the day, if consent is going to be the basis to defend against these types of privacy claims, defendants would be well advised to really be thorough in procuring this consent. In fact, I'm surprised that Bresnan - given that it is an IAP allegedly engaging in gray area practices - didn't just secure written consent at the time it first provided the service.
I'm also surprised at the court's conclusion on the Computer Fraud and Abuse Act damage issue, given its conclusion on the ECPA issue. If it was going to split hairs on the notice and consent (as it did with respect to the CFAA claims), it could have probably done so on the ECPA claims as well. Courts often keep in claims they may otherwise dismiss if they decided that some claims are going to survive. Also, some cases construe the CFAA narrowly as requiring damage to the protected computer (or an interruption in data). It's conceivable that plaintiffs could have suffered the requisite loss (which can be aggregated in the class action context), but the court's discussion of plaintiffs' allegations made the damage allegations seem awfully light. (Two posts from Nick Akerman look at some recent CFAA dismissals and discuss the restrictive approach taken by some courts with respect to the CFAA's jurisdictional damage requirement: "Dismissal of CFAA Claim for Lack of Jurisdiction" and "Why Two District Courts Dismissed Valid Computer Fraud and Abuse Claims for Lack of Jurisdiction.")
The dismissal of the ECPA claim as opposed to the CFAA claim could have some ramifications on the damages front. Statutory damages are available under the ECPA, but not under the CFAA. For what it's worth, there's conflicting authority on the issue of whether non-economic damages are recoverable under the CFAA. (See Garland-Sash v. Lewis, 348 Fed. Appx. 639 (2d Cir. 2009) (construing the phrase "compensatory damages" - which was added to a provision of the CFAA after the DoubleClick case came down - to include damages for pain, suffering, and other emotional harms").) Even if for some reason the court decides that plaintiffs are entitled to non-economic damages, it will be interesting to see how plaintiffs prove up these damages.
The trespass claim is a bonus claim, but again, the court doesn't dig in to the damage issue with respect to common law trespass. Although the court cites to California law, the court does not discuss damage or slowdown to the machine in question as articulated by the California Supreme Court in Intel v. Hamidi (an email bombardment case) or as interpreted by the Fourth Circuit in the Omega v. Mummagraphics case.
I'm not sure how much light this ruling will shed on the many pending privacy lawsuits that involve things like surreptitious tracking, sniffing, and leakage of personal information. Damages issues aside, the ruling may highlight the importance of choice, consent, and the requirement that any disclosures or disclaimers be conspicuous, all issues the FTC seems to frequently opine on and issue reports about.
(h/t Wendy Davis)
As Venkat notes, this ruling is an inconsistent mix of formalism and realism. In light of the judge's ruling last month that Bresnan made inadequate disclosures to uphold an arbitration clause, it's odd for the judge to now find that Bresnan made adequate disclosures to wipe away the ECPA and privacy invasion claims via dense/buried EULA language plus an opt-out notice; while that same consent wasn't good enough to wipe away the CFAA and Trespass to Chattels claim. The CFAA ruling on damages was also oddly formalist given the consent ruling. I respect formalist judges for being careful and methodical, but it would have been nice if this judge had been a little more aggressive about calling a spade a spade.
I am not a fan of deep packet inspection (DPI) by IAPs done on anything but an opt-in basis. We're basically back to the old battles about unwanted adware/spyware getting onto users' hard drives as part of some bundle. Sure, the adware vendors could claim user consent through a formalist reading of the contracts, but there wasn't true consumer consent, and we all knew it. I'm reminded a little of the FTC's bust of Sears for its trackware installations--Sears paid people for the installation, but the software did things far beyond anything users might have expected, even though these attributes were putatively explained deep in the EULA. If you're an IAP trying to implement DPI on an opt-out basis, bonne chance, and don't expect a lot of friends to rally around your cause.
At the same time, I'll be interested to see if the plaintiffs can marshal any true evidence of harm. If the plaintiffs are advancing a recycled version of the old, tired and completely laughable arguments that installing cookies on a user's computer creates cognizable harm, I hope this judge will quickly give them the boot they deserve. In that respect, I'm disappointed the judge didn't more aggressively police the trespass to chattels claim on the harm requirement per Hamidi. Personally, I think these plaintiffs should have been forced to put-up-or-shut-up on the harm issue early. Then again, this case came out the day before the Ninth Circuit's recent Starbucks case, but perhaps it's consistent with it.
Overall, this ruling is just another small data point in a much larger struggle over targetable consumer data. My Coasean Analysis of Marketing article doesn't directly address DPI by IAPs, but the article tells the story of how different intermediaries are fighting with each other to capture better datasets of targetable consumer behavior. After the flameout of the early 2000s model of adware, IAPs are trying to squeeze into the middle by using their more favorable position (compared to websites) to see more complete consumer data. Similarly, Facebook is trying to use tools like Beacon nee Instant Personalization to sweep up targetable consumer data from throughout the web, not just the smaller dataset it can capture at facebook.com. Meanwhile, Google is trying to move onto the desktop (the toolbar, Desktop, Chrome and its various OSes) to let it get closer to the honeypot of consumer data residing there, rather than just rely on the data it can get at google.com properties. Adware circa 2005 may be dead, but battles between different intermediaries fighting to get the good stuff is a perennial. For more, see my posts Adware is Dead and Relevancy Trumps Creepiness.