Ninth Circuit’s Mixed Opinion in Glider/WoW Bot Case — MDY Industries v. Blizzard

[Post by Venkat, with comments from Eric]

MDY Industries, LLC v. Blizzard Entertainment, Nos. 09-15932 & 16044 (9th Cir. Dec. 14, 2010)

The Ninth Circuit issued its opinion in the Blizzard Glider “bot” case, which is one of three cases in the Ninth Circuit that tackle the license vs. sale issue and what conditions the licensor can put in place. Vernor v. Autodesk, was the first, and Augusto v. BMG is still pending. The Blizzard ruling also covers some other ground in addition to the license vs. sale issue, including: (1) what parts of a licensing agreement support infringement claims (as opposed to breach of contract claims) and (2) the scope of a DMCA claim under section 1201. It will be interesting to see the effect of the case on two big cases from the Federal Circuit: Jacobsen v. Katzer (an open source case involving model train sets) and The Chamberlain Group v. Skylink Technologies (a garage door opener DMCA case). It took me a couple of times to read this case – it made my head spin!

The crux of the Blizzard dispute is that MDY made available a “bot” which World of Warcraft players could use to advance through levels without actually playing the game. Blizzard alleged that players’ use of the Glider “bot” infringed on Blizzard’s copyright (because the end user agreement contained a restriction on playing the game using a bot or other automated means), and Blizzard asserted claims for secondary infringement against MDY based on this. Blizzard also implemented measures (a software program ominously called the “Warden”) to detect Glider’s operation, which MDY allegedly circumvented. Blizzard asserted a DMCA claim based on this.

Copyright Infringement: Blizzard argued that MDY was secondarily liable for infringement by the players who used the bot. Blizzard’s claims of infringement focused on the RAM copy made in the course of playing the game. If the transaction between Blizzard and the players was a sale, the players could take advantage of the “essential step defense,” under which there is no infringement for copies that are created and used “as an essential step in the [use] of the program.” (section 117(a)) The threshold question was whether WoW players were licensees or owners of the particular copies of the software that they had purchased. Applying Vernor v. Autodesk, the court concludes that the players are licensees and therefore, the RAM copies may be infringing unless they are licensed.

WoW License Agreement: Whether or not the bot-using players infringed turns on the terms of the WoW EULA. The court runs through the EULA process (which looks leak-proof) and then turns to the question of whether a violation of the no-bot provision of the license constitutes infringement. The court distinguishes between “covenants” and “conditions” — violations of covenants are actionable as breaches of contract and violations of conditions are actionable as infringement. (A copyright infringement lawsuit has obvious advantages, such as statutory damages and injunctive relief, and in this case, infringement would have been a hook for Blizzard to hold MDY derivatively liable.) Ultimately, the court concludes that infringement by breach of a condition occurs only where the licensee:

exceeds the scope of the license in a manner that implicates one of the licensor’s exclusive statutory rights.

With respect to Glider, the court holds: “Glider does not infringe any of Blizzard’s exclusive rights. For instance, the use does not alter or copy WoW software.” End result, no infringement by end users and no liability for infringement to MDY. But that wasn’t the end of the story.

Blizzard’s DMCA Claims: Although the court took a restrictive view of what violations of a license constitute copyright infringement, the court took an opposite approach with respect to Blizzard’s DMCA claim. The key question here was whether the cause of action created under section 1201(a) was “linked” to copyright infringement, or whether it created a totally separate cause of action. Blizzard implemented measures which prevented access to the game while the bot was in operation, but these measures did not prevent copyright infringement by the players (per the court’s conclusion on the covenant/condition issue discussed above). The court concludes that regardless of whether the underlying use is infringing, anti-circumvention of a measure that restricts mere access to a work can support DMCA liability under section 1201. The court repudiates the Federal Circuit’s approach in Chamberlain, the infamous garage door case, and concludes that:

a fair reading of the statute . . . indicates that Congress created a distinct anti-circumvention right under section 1201(a) without an infringement nexus requirement.

According to the court, Glider violated 1201(a) with respect to WoW’s “dynamic non-literal elements” — i.e. “the real-time experience of traveling through different world, hearing their sounds, viewing their structures, encountering their inhabitants and monsters, and encountering other players.” The court found no violation with respect to the source code or actual audiovisual elements of the game because the anti-circumvention measures implemented by Blizzard did not effectively restrict access to these.


This case made my head spin. Eric discusses more of the opinion’s messiness below, but I wanted to offer a few quick comments.

1. Application of Autodesk to determine sale/license: the Ninth Circuit recently decided Vernor v. Autodesk where it ruled that Autodesk CAD software was licensed and not sold to its clients. One question the Autodesk decision raised is how the analysis would play out in other contexts — i.e., where the software is less expensive, off-the-shelf, and looks more like a license. Surprisingly, the court here didn’t spend much time on this, giving just one page to the treatment of the license/sale issue.

2. RAM copies and the essential step defense: The court’s opinion contains this zinger:

Since WoW players, including Glider users, do not own their copies of the software, Glider users may not claim the essential step defense. 17 U.S.C. § 117(a)(1). Thus, when their computers copy WoW software into RAM, the players may infringe unless their usage is within the scope of Blizzard’s limited license.

Whether it’s a matter of clunky legislative drafting or otherwise, the conclusion that people who purchase copies of software can be held liable for RAM copies that are necessary to utilize the program (i.e., “an essential step in the utilization of the computer program”) is remarkable. If you were to accept the court’s conclusion, I don’t see much utility for section 117(a)(1) at this point. If section 117(a)(1) is not available to licensees, not many people are going to be able to take advantage of it.

3. The covenant/condition distinction: The court’s conclusion that a breach of a license agreement term can constitute copyright infringement only where the licensee exceeds the scope of the license “in a manner that implicates . . . the licensor’s exclusive statutory rights” drastically narrows the scope of available claims for copyright infringement in the licensing context. One big question this raises is how this squares with Jacobsen v. Katzer [pdf] where the federal circuit held that a failure to comply with certain open source attribution and “modification transparency” requirements constitute copyright infringement. The Ninth Circuit’s opinion in Blizzard contains some pretty restrictive language with respect to the covenant/condition issue (the breach of license must be “in a manner that implicates one of the licensor’s statutory rights”). Katzer, in contrast, referenced the fact that the requirements would inform downstream users and generate awareness for the incubation page which would further the “economic goal” of the copyright holder. How would Jacobsen have fared under the Ninth Circuit’s standard in Blizzard? A 2008 EFF blog post by Michael Kwun flags the tension between the two decisions: “Condition or Covenant, and Why Should You Care?

4. The DMCA Claim: What the Ninth Circuit taketh away with the copyright analysis, it giveth back with the DMCA analysis. After concluding that use of Glider while playing WoW does not cause the players to infringe, the court concludes that MDY violated the DMCA when it circumvented Warden. Here is the crux of Chamberlain [pdf]:

the anticircumvention provisions convey no additional property rights in and of themselves; they simply provide property owners with new ways to secure their property.

The obvious concern here is that the parade of horribles (and absurd claims) that Chamberlain warns of will come to pass if courts allow anti-circumvention claims that are divorced from copyright claims, as the court did here.

Related posts:

Vernor v. Autodesk–Does the Right to Possession Distinguish Between Sales and Licenses?

Software Vendor Trumps First Sale Doctrine via License–Vernor v. Autodesk

Other coverage:

paidContent: “New Ruling Will Impact Software Licensing Agreements

EFF: “A Mixed Ninth Circuit Ruling in MDY v. Blizzard: WoW Buyers Are Not Owners – But Glider Users Are Not Copyright Infringers

Wired: “Court Upholds Ban on World of Warcraft Bot


Eric’s comments:

Before I go off the rails here, let me start with the only piece of good news: I think the court reached the right result in finding for Blizzard and against MDY. I’ll probably get some flack for that position, but I thought it was reasonably clear that MDY could not do what it did. As a result, if the panel had simply affirmed the lower court per curiam and vacated the lower court’s opinion, I would have viewed this outcome as appropriate.

(This assessment differs from my normative view. I understand that Glider posed some potential problems for WoW gameplay, but I think Blizzard overreacted. At minimum, Blizzard should thank MDY for showing the existence of an underserved market segment worth millions of dollars. The fact that Blizzard doesn’t directly capitalize on that market demand, such as by setting up a “fast play” WoW on servers separate from the normal WoW, puzzles me).

Unfortunately, the Ninth Circuit panel attempted to explain *why* it was ruling for Blizzard, and everything went to hell.

Nevertheless, I’d be more upset about this opinion if I actually believed it. The Ninth Circuit clearly makes up Internet law with each new case and each new panel, so I have no reason to think this opinion will stick any more than the dozens of other implicitly reversed Internet law opinions from the Ninth Circuit over the past 15 years. Because I think this opinion’s predictive power for the next opinion is so low, I don’t really believe this panel means what it says. At least, I choose not to believe it. This would be a great opinion to take en banc, although that would probably only further solidify the doctrinal disasters lurking in this opinion.

The opinion gets itself into trouble for a number of reasons, but the biggest reason IMO is that the opinion dances around the seminal cyberlaw issue of whether web browsing is infringement. We all know what copyright law says, and should say, about browsing. So long as copies into RAM are putatively infringing unless there’s a defense, current copyright law says browsing can be infringement. Copyright law *should* say that browsing isn’t infringement, which would eliminate some pressure that caused the court to engage in doctrinal gymnastics here.

This was not exactly a browsing-as-infringement case because Blizzard also licensed client-side software that facilitated player access to the copyrighted materials on WoW servers. However, as more websites load client-side software as part of web browsing, that distinction may be diminishing. Plus, Glider didn’t modify Blizzard’s client-side software; it only interacted with Blizzard’s servers. As a result, Glider was part of the way players browsed Blizzard’s web service.

Perhaps in response to the problems of treating browsing as copyright infringement, the court makes it clear that violating a website’s EULA by browsing would most likely just be a breach of contract, not a copyright infringement. Eliminating browsing as infringement would clean up the problems applying 1201(a) because it would not treat browsing as “accessing” copyrighted works on web pages. To take this issue off judges’ minds, we would benefit from a statutory amendment saying that web browsing isn’t infringement.

Treating a software license breach as sometimes creating copyright infringement and other times creating solely a breach of contract isn’t new legal doctrine, but it’s problematic. It does ameliorate concerns about oppressive but marginally relevant license terms buried in EULAs; in many cases, this opinion says those terms can’t be enforced via the coercive power of copyright remedies. However, trying to cleave between infringement-related and other contract restrictions sounds like a litigation morass; and I’m unclear if fancy drafting (e.g., stating that all EULA covenants are conditions on the software license) is an all-too-easy workaround. See, e.g., the provision where the court says “Although ToU § 4 is titled, ‘Limitations on Your Use of the Service,’ nothing in that section conditions Blizzard’s grant of a limited license on players’ compliance with ToU § 4’s restrictions.” By inference, simple changes in EULA drafting could change this calculus.

I also wonder how this ruling affects open source licenses. The Jacobsen v. Katzer case held that an open source license was a condition on the enjoyment of the copyrighted software. The Jacobsen case didn’t directly address the fact that the open source license failed to meet basic contract formation requirements. To the extent this case indicates that some software license agreement breaches are only contract breaches, not conditions on the enjoyment of copyright, this case leads to one of two outcomes, both of which are weird. Either (a) the “conditions” in open source licenses are not subject to the rules in this case, in which case provisions not included in a binding contract will be more enforceable than provisions in properly formed software license agreements (a result that can’t be right), or (b) conditions in open source licenses are equally enforceable as software license agreements, in which case some provisions of open source licenses are enforceable only as contract breaches–but since open source licenses usually aren’t properly formed as contracts, those provisions will be effectively unenforceable. Either result seems like a catalyst for anarchy.

In addition to these issues, I’ll reiterate two other obvious areas of mischief noted by Venkat:

1) By reiterating Vernor’s standards for ownership vs. license of material, this case does not reduce the risks that copyright owners will contractually eliminate the first sale doctrine for copyright-containing chattels (such as books) by characterizing the copyrighted material transfer as a license. I’m still hoping the UMG v. Augusto case gives us some better news on this front.

2) The court’s standards for a 1201(a) circumvention raise all of the anti-competitive concerns we feared with the Chamberlain case where two independently manufactured chattel (which contain software) need to make an electronic handshake with each other. This happens all the time for interoperability purposes, including the sale of replacement or complementary goods. The court punts on the antitrust issue:

Concerning antitrust law, we note that there is no clear issue of anti-competitive behavior in this case because Blizzard does not seek to put a direct competitor who offers a competing role-playing game out of business and the parties have not argued this issue. If a § 1201(a)(2) defendant in a future case claims that a plaintiff is attempting to enforce its DMCA anti-circumvention right in a manner that violates antitrust law, we will then consider the interplay between this new anti-circumvention right and antitrust law.

Only time, and future Ninth Circuit panel flip-flops, will tell us if these fears will come to pass.