Thoughts on the Ungaro order

Hegel bemerkte irgendwo, daß alle großen weltgeschichtlichen Tatsachen und Personen sich sozusagen zweimal ereignen. Er hat vergessen, hinzuzufügen: das eine Mal als Tragödie, das andere Mal als Farce.
—Karl Marx, Der 18te Brumaire des Louis Napoleon, 1852

On March 20 Judge Ursula Ungaro dismissed a Malibu Media case in FLSD (14-cv-20213) due to concerns over the IP address geolocation tech used by the plaintiff. I don’t usually comment on copyright trolling cases here but there’s some interesting stuff in this case and the order has become pretty popular around the web.

The crux of the issue is that in order to open a federal case and file, as trolls typically do, for an ex parte subpoena to  the ISP to identify the defendant, the plaintiff needs to establish that the district court has jurisdiction over the defendant. Showing that the defendant resides in the district suffices. Showing that the IP address implicated in the infringement geolocates to that judicial district, as Judge Ungaro has decided, does not suffice.

This is an interesting take on the IP-address-is-not-a-person doctrine that we wish courts would take up more widely, mostly because it rests on highly technical arguments. Judge Ungaro, from her Order to Show Cause:

The Court is unpersuaded [of the trustworthiness of the IP geolocation data] after reviewing technical literature which suggests otherwise.

Among the many challenges presented by the Internet is that the Internet itself possesses no inherent mechanism for determining the geographic location of connected devices. While Domain Name System (DNS) entries can include a location record, there is no standard protocol to provide global location data which corresponds with an Internet Protocol (IP) address. Brian Eriksson, Paul Barford, Joel Sommers & Robert Nowak, A Learning-Based Approach for IP Geolocation, Proc. of the Eleventh Int’l Conf. on Passive & Active Measurement (Zurich, Switz.), Apr. 7–9, 2010. Furthermore, the size and complexity of the Internet coupled with its highly diffuse ownership and user-base provides no single repository or authority which could maintain such data. Brian Eriksson, Paul Barford, Bruce Maggs & Robert Nowak, Posit: A Lightweight Approach for IP Geolocation, Submitted to SIGMETRICS Performance Evaluation Review Dec. 2011. Instead, IP address geolocation technologies rely primarily upon active network measurements, or alternatively databases of IP to location mappings. Phillipa Gill, Yasgar Ganjali, Bernard Wong & David Lie, Dude, Where’s That IP? Circumventing Measurement-Based IP Geolocation, Proc. of the Nineteenth USENIX Conf. on Security (D.C.), Aug. 11–13, 2010. The former falls victim to inconsistent Internet topology and incomplete information regarding this topology due to the Internet’s rapid and continuous expansion. Anukool Lakhina, John W. Byers, Mark Crovella, Ibrahim Matta, On the Geographic Location of Internet Resources, IEEE J. on Selected Areas in Comm., Spec. Issue on Internet and WWW Measurement, Mapping, and Modeling, 2003. Similarly, databases of IP location mappings fall victim to being rough and incomplete—also due to the Internet’s rapid expansion. Yong Wang, Daniel Burgener, Marcel Flores, Aleksandar Kuzmanovic & Cheng Huang, Towards Street-Level Client-Independent IP Geolocation, Proc. of the Eighth USENIX Conf. on Networked Systems Design & Implementation (Bos., Mass.), Mar. 30–Apr. 1, 2011.

Thus, despite the existence of geolocation software designed to approximate the geographical location of Internet-connected devices, this Court cannot rely solely upon Plaintiff’s assertion that such technology was used for purposes of establishing proper jurisdiction and venue. This is particularly true in instances where 28 U.S.C. § 1400(a), the exclusive venue statute for copyright infringement, controls and permits venue to be laid in the district where the defendant resides or may be found.

Now if you’re counting, that’s 5 cites to technical papers: two from USENIX conference proceedings, one from an IEEE journal, one from SIGMETRICS and another from PAM. These are serious references, and importantly, ones which trolls simply do not have the technical know-how to refute. (This also applies to Lipscomb’s colleagues in Germany, unless Tobi Fieser secretly got a PhD in networking while we weren’t looking.)

That didn’t stop Lipscomb from trying. He has, it must be conceded, excellent points: the inaccuracy of geolocation databases is basically only academic, and it has always sufficed before. (The rest is nothing but fluff trying once again to impress the Court. [Hint: getting technical to try to impress a Court which just threw five technical conference papers at you is unlikely to work.]) Note though that he has done nothing to refute any of the five cites Judge Ungaro threw at him. (I doubt he even read the papers; even if he wanted to he probably wouldn’t be able to find them.)

The judge seemed to sense that the IP-geolocation-doesn’t-work argument was somewhat weak and dismissed the case on other, more traditional grounds:

Plaintiff has shown that the geolocation software can provide a location for an infringing IP address; however, Plaintiff has not shown how this geolocation software can establish the identity of the Defendant. There is nothing that links the IP address location to the identity of the person actually downloading and viewing Plaintiff’s videos, and establishing whether that person lives in this district.

We have heard this argument before. It is, essentially, Judge Otis Wright’s unfairly-ignored “old-fashioned stakeout” idea from Ingenuity 13 v. Doe.

The point here is that Judge Ungaro is pointing out the core of copyright trolls’ modus operandi of taking an IP address geolocated to perhaps county level, going to the corresponding federal court, and obtaining the defendant’s identity via a subpoena to the ISP. (Then, as we know, an offer is sent to the defendant to settle for some amount just under the cost of a vigorous defense, which if declined, leads to the defendant being publicly named in the suit, which most of the time leads the defendant to reconsider.)

The subtext here is not, like Lipscomb seems to think, that copyright trolls need to be more careful on the technical side (although they certainly do). It is that they have it backwards to begin with: they really ought to start with a defendant identified first, and only use the court’s subpoena power to gather further evidence (subpoenas for hard drives, for example, are perfectly legitimate). But then the question arises: how do you identify a defendant if all you have is their IP address, and no subpoena power?

Old-fashioned stakeout.

Trolls have gotten used to being able run through this log-sue-subpoena-name-settle routine quickly. As long as they can settle each case for over $350 (the federal filing fee), they’re making money. Given that it is suspected they settle each case for several thousand dollars1, they’re making a lot of money.

This answers the old question of why trolls never seem to go after original seeders or the release groups that download the movies from the actual website in the first place.2 It’s because that’s hard: such people are likely to be behind the proverbial seven proxies; no amount of IP geolocation is going to help there. Trolls sue IP addresses, not people. If judges take away the ability to blindly sue an IP address, as Judges Wright, Ungaro, and increasingly more judges around the nation are, suddenly everybody is that difficult to identify3, and copyright trolling will be impossible.

  1. From the threat of statutory damages, which themselves deserve a hard look at, especially with the extraordinary bravado coming from X-Art that a $40.5K statutory judgement was less than actual damages. []
  2. In Prenda’s case it was because they were the original seeders. This is not the case for X-Art. []
  3. Ironically, while that would completely destroy the entire copyright trolling business, it might actually help fight piracy, by making it more worthwhile for copyright holders to go after original seeders and release groups. While X-Art is correct that thousands of people are probably torrenting their movies at any particular time, a careful look through torrent sites suggests that the number of people actually responsible for originating those torrents may be as few as one. []