When “Unique” Doesn’t Mean “Special” (and Why that Actually Makes Sense)
Last time, we looked at how venerable Pandora fits into the legal ecosystem of online music services. It’s a webcaster, which means that, if it plays its cards exactly right, it can avail itself of a statutory license for the right to stream sound recordings. It turned out this is an arduous task, but Pandora (and many other webcasters) have been up to the task. Except for one thing, which could have blown the whole project out of the water.
The first requirement that a webcaster must meet to be eligible for the statutory license is not to be an “interactive service.” The Copyright Act provides a lengthy definition of interactive service, but it boils down to (1) playing requests, or (2) creating a program “specially” for the user. It’s easy to avoid definition (1). But what does it mean for a program to be “specially created for the recipient”? The most successful webcasters are successful because they can create programs that are unique to the user. Does that make them also “special” to the user?
As it happens, in 2001, the music industry sued one such webcasting service called LAUNCHcast. Like Pandora, it provided a personalized music channel, this one based on a short questionnaire, on-the-fly ratings, and so forth. Songs you rate highly do get played more frequently (something that’s not a Pandora feature, I don’t think), but you don’t know when they’ll be played. In addition, LAUNCHcast included lots of songs from your selected genres but which you’ve never expressed an opinion on. In all events, you have no idea what song is coming next, or really what songs are coming up, except that there is a better than random chance that one of your favorite songs will be coming up–but again, you don’t know the order or the exact time.
The music industry argued this was an interactive service because, well, the user had to interact with the service to get it to work. They pointed to one of the definitions of interactive service: “one that enables a member of the public to receive a transmission of a program specially created for the recipient.” The music industry’s point was that LAUNCHcast created a program of songs just for the user.
The Second Circuit Court of Appeal disagreed, in Arista Records, LLC v. Launch Media, Inc., 578 F.3d 148 (2d Cir. 2009). It identified the term specially created as ambiguous. The court delved (I mean really delved) into the well-documented legislative history and determined that specially created meant a set of “pre-packaged” songs that the user either picks or whose contents the user knows about before he or she receives the stream. The key, the court wrote, was the user’s ability to know what’s coming, which in turn is meant to discourage copying the streamed music. Thus, although the playlist is unique to the user, and it’s meant to be more meaningful to the user than to others, it’s not “specially created” for the user because the user doesn’t exercise enough control (or perhaps the right kind of control) over the playlist to predict upcoming songs.
The LAUNCHcast case took until 2009 to finally resolve. During that time, Pandora’s founders were working on the Music Genome Project (work on which started in 2000, I think); and the Pandora service launched (in 2005). I don’t think I realized it at the time, but it was pretty gutty of Pandora to launch when it did, with the LAUNCHcast case still pending. Had the LAUNCHcast case gone the other way–and it really could have*–all of Pandora’s work would have been for nothing.
* In my opinion, the LAUNCHcast decision is quite persuasive. The court’s research into the legislative history is exhaustive, and its understanding of the relevant technology is impressive. But, if you read just the text of the definition of interactive service, it’s a pretty counter-intuitive result. I suspect it took a very talented team of lawyers on LAUNCHcast’s part to dig up all that legislative history and to explain it cogently to the court, along with the technology, and it took a very engaged court to take it all in. The court could easily have taken the easy way out, say that “specially created” means what it superficially appears to mean, and be done with it.
Next time in this series, we’ll look at a much more complex and fascinating “internet radio” case: Turntable.fm. We’ll see if the LAUNCHcast decision has something to say about it, too.
And finally, if you want to keep up with the law of internet radio, subscribe to David Oxenford’s Broadcast Law Blog (Internet Radio).
Update: One more thing. Here is a proposal for a separate licensing scheme to cover internet radio services, like Pandora and Turntable.fm, that are partially interactive. The author would resolve the tension between the plain language of the statutory definition of interactive and the legislative history by creating a new category of webcasters that fall in between what Congress had in mind (internet rebroadcasting) and what Congress was afraid of (piracy).
Thanks for reading!
Thanks Tara for covering this crucial point.
You could go even further, the concept of channel, basis of the DMCA for interactive radios, can hardly be applied to interactive radios (Pandora would provide several billions of channels; at 500$/channel/year, it would be pretty expensive…).
DMCA format fails on 2 grounds:
– it is an enormous constraint limiting the emergence of new formats: if a service is not complying with the various rules (5 skips/hour, not more than 4 songs from the same artist/3 hours,…), it needs to pass exclusive agreements with right holders.
– It does not provide defferenciated tariffs for different levels of interactivity: for instance different tariffs for sound quality (32, 128, 320kbs), nb of skips per hour,…
music service could provide maximum value to their customer and structure freemium models optimising revenue generation.
Players (righs holders and streamers) as a result are not abiding by the rules strictly.
The consequence is today a sub-optimally structured market: Youtube, which provides “on demand” format and video on top of audio and paying pays a ridiculous payout on exclusive agreements: it should pay per stream say 10×2=20 times more than semi interactive radios like Pandora, but it is actually paying 60 time less. It is actually paying to the tune of 1000 times less relatively: I do not know any industrial sector with distorsions of that magnitude.
The question should not to spend hours arguing whether such service comply with the exisitng regime but make it more flexible and tariffs structure differentiated.
Thanks for the thoughtful comments. If you don’t mind, I’ll respond by looking at things from the point of view of a (rational) rights holder. I’ll be the first to admit that the DMCA’s statutory-royalty scheme for internet radio is overly complex. The rules governing the “sound recording complement” appear finely detailed to a silly extreme, but they do have the advantage of being a “bright line” set of rules. Are you suggesting that Congress just scrap the requirement, or replace it with something more flexible? If I’m a rights holder, I’m going to resist, for example, allowing channels dedicated to specific artists on grounds that that really would affect the market for the artist’s songs. On the sound-quality issue, do you know if anyone raised this issue with the Copyright Royalty Judges? Again, if I’m a rights holder, I’m not sure how I’d benefit from a lower royalty on lower quality streams. They all would sound better than AM radio! If the purpose of the statutory-royalty scheme is to approximate the deal that would be struck by the average rights holder and the average streamer, then the rules are going to be, necessarily, milktoast and one-size-fits all. It seems to me if you want flexibility, you’ll need some way of clearing masses of copyrights, but copyright clearinghouses just haven’t caught on.
Let me take a comparison: it is as if in a plane all sits had the same price: no first, no business, no economy, no low cost,…flying companies would forgo a substantial part of their revenues (without differentiated tariffs and yield management pricing) and have no incentive to propose value added service. In the music industry rights holders likewise are foregoing a substantial part of potential revenues.
The freemium model for instance generates little revenues because the free element of most freemium services offers already a very high level of experience and provides no incentive for listeners to get Premium subscriptions.
Tariff should be more differentiated, not flexible, and DMCA’s complexity almost a secundary issue. Little has been done to experiment empirically pricing optimisation.