Why Grooveshark Probably Needs a Defense or a License

In our last post, we looked at the case against the “feeding end” of the Grooveshark, which meant we mostly discussed secondary copyright liability.  We’ll now turn our attention to Grooveshark’s “streaming end.”

Generally speaking, there is at least one, and maybe up to three, exclusive rights implicated by streaming music (or other digital content) over the internet.  Everyone agrees that it constitutes a public performance.  It might also constitute a reproduction and a distribution–but we don’t know for sure because (1) most courts don’t reach the question, having stopped at public performance*; and (2) it might depend on the specific technology being used (which might be another reason courts don’t get into it!).  But we’re not a court–we’re a blog, so we won’t stop at public performance.  We’ll look at all three!

Recall that you only need to prove infringement of one of the exclusive rights.  Proving infringement of additional exclusive rights doesn’t add anything, except perhaps insurance.  In this case, the public performance right is so clearly implicated that insurance isn’t really necessary.

Public Performance:  It might not seem intuitive at first that streaming content over the internet constitutes a “public performance.”  After all, term public performance is defined as performance to “the public” or “where a substantial number of persons outside a normal circle of family and its social acquaintances is gathered.”  Users enjoying streaming content usually do it alone or in a very small group–and, besides, how would you know how it was being enjoyed?

There is a second, less intuitive (but no less potent) definition of public performance, one designed decades ago to capture services like hotel pay-per-view television:  “to transmit or otherwise communicate a performance … to the public, by means of any device or process, whether the [audience members] receive it in the same place or in separate places and at the same time or at different times.”*  Thus, if different members of the public receive different streams of the same song in different places at different times, it’s still a public performance of the song.

You’ll hear this referred to as the “transmit clause.”

Therefore, it’s a general rule of the internet:  you’re streaming music to the public–even if it’s just to a limited number of subscribers–you’re infringing copyright, and you’d better either be licensed or have a good defense  I can’t think of too many exceptions to this rule.*  Grooveshark is not (currently) licensed, so the question is whether it has a good defense.  More on that in the next post.

One exception might be iTunes’ home sharing feature, since only those living within range of your WiFi and authorized to use your computer can have access to the stream.    (But what if your WiFi signal was unsecured?)

Reproduction:  Because streaming is nearly always a public performance, most legal decisions stop there and don’t wrestle with the harder question of whether streaming also reproduces the work.  But we won’t let that stop us from speculating!

The first obstacle we encounter in our analysis is that there are many different ways to stream content, and the technical details might mean the difference between infringement and non-infringement.  Streaming falls into two basic categories:  progressive downloading and true streaming.  From the point of view of the user’s experience, there isn’t much difference between the two (if everything is running smoothly), but from technological point of view, they’re very different.  Strictly speaking, it’s the technology, not the user experience, that usually matters.

With progressive downloading, the user’s computer begins to download the file using standard download protocols–pretty much a normal download.  However, rather than wait until the entire file is downloaded before playing, the user’s medial player begins to play the file once enough of the file has been downloaded (“buffered”).  Ideally, the media player waits long enough so that the entire file can be played before the buffer runs out.  If it doesn’t or can’t, then the play-back will stop or stutter as the buffer is built up again.  This is how a typical YouTube video works, for example.

The key question is whether the entire file is, at some point, fully downloaded and present on the user’s computer.  To be honest, I can’t tell.  It might depend on the media player, the user’s operating system and settings, and the size of the file.  It’s safe to say that sometimes–and perhaps most of the time–entire files are downloaded and stored via progressive downloading.  To be sure, they are kept in temporary storage, marked for overwriting, but they could persist for quite some time in a lonely corner of the user’s computer.

If this is the case, then a reproduction has certainly occurred, and there is a technical infringement of copyright.  A good argument for fair use could be mounted here* because the copy isn’t supposed to last very long in the user’s cache, and the value in the copy (as opposed a public performance) is having it long term or being able to do something with it repeatedly on demand.

*  Bearing in mind what I said earlier:  fair use is like unto a religious mystery.

Even if this is not the case, enough of the file is probably stored for there to be technical copyright infringement.  But in that case, the fair-use argument is significantly stronger because the user has only a portion of the file, and the lifespan of the file is likely to be less.  I can think of few things less useful than random portion of a song file.  In fact, if the header information was deleted, the file might be even be play-able.

If, however, the player overwrites the file as it is being played, then perhaps not enough of the file is copied at any one time to constitute a reproduction.*  I’m honestly not sure if any media player actually does this.  There seems to be no technological benefit to doing this.  The only benefit would seem to be legal in nature–to avoid copyright infringement.

*  One wrinkle:  the Sixth Circuit, at least, has held that there is no de minimus defense for sound recordings.  I think that holding, though, is limited to “sampling”–i.e., copying small but identifiable bits of songs for use in your own song–and not to tiny bits of computer code that, if played, would play a small but identifiable bit of a song.

With true streaming, information (you can’t really call it a “file”) is sent to the user’s computer, and software on the user’s computer is able to convert the information into audio and video.  This happens nearly on the fly, with only a very small amount of buffering.  It requires an entirely different set of protocols, and special software on both the transmitting and receiving ends of the stream.  The technology was developed to broadcast events live over a network.  Even though its use is fairly common, for such things as webinars and concerts, and it’s been around since the early 1990’s, I think everyone agrees that the technology is in its infancy.

As I understand it, the information is never kept in non-volatile storage (hard drive, etc.), but is cached in volatile RAM for a very brief period of time, played, then lost.  When there’s an interruption of the stream, the system responds either by degrading the quality of the output or hanging then picking up where the interruption left off.

If this is the case, then true streaming shouldn’t constitute an infringement of the duplication right.  Too little of the information is kept in a volatile environment for too long, in my opinion.  Not everyone agrees*, but there is at least one major decision (the Cablevision decision) holding that 1.2 seconds of buffered data in RAM isn’t really a copy (because it isn’t really “fixed” to the underlying media).**

Conceptually, this issue is a huge headache.  We have one decision (MAI) holding that copying software from non-volatile media (hard-drive, CD-ROM, etc.) into RAM constitutes an infringing reproduction of the software.  From the point of view of copyright law, there really isn’t any difference between software code and streaming data–it’s all protectable expression–yet, buffering data in RAM feels qualitatively different.  Cablevision resolved the tension by focusing on the short period of time the data was kept in RAM, but not everyone finds that a satisfactory solution.

** No one knows how long would be too long.  I’m assuming that the data cached in true streaming isn’t held for an appreciably longer time.

To sum up on the reproduction right:  (1) progressive downloads probably infringe the reproduction right (but consumers might have a good fair-use defense, depending on the nature of the progressive download); and (2) true streaming probably does not infringe the reproduction right, but not everyone is sold on that conclusion.

Distribution:  If streaming does constitute a reproduction, then it follows that it constitutes a distribution, just as surely as letting members of the public access and download your song files.  There is an evidentiary wrinkle, however:  when confronted by many accessible files, and evidence of data transfers from the site, how do you know which ones have actually been distributed (i.e., streamed or downloaded)?  If no one ever made a copy of a song file, it was never distributed, right?

Compare this situation to books in a bookstore.  If a bookstore has in inventory 1000 copies of, say, What Color Is Your Parachute?, and it later has 900 copies, it’s a pretty safe bet that at least some of the 100 copies were distributed to the public.  With digital content, however, you always have the same number of copies in your “inventory,” just one.  When members of the public access and download a song file, few fingerprints are left at the scene–unless the operator chooses to keep an activity log (and why would it?).  In short, there’s no easy way to tell that a song file has been distributed simply by looking at the song file.  You need evidence that a member of the public actually downloaded the file–which is almost the equivalent of obtaining evidence that an actual customer went into your store and bought a copy of What Color Is Your Parachute?, only without any kind of paper trail.

As I mentioned in my post about Jammie Thomas-Rassett, some courts have resolved this evidentiary conundrum by equating access with distribution.  I.e., if you made your file available to the public, you are deemed to have distributed it, regardless of whether members of the public actually downloaded it.  Most courts, however, insist upon proof of actual access and downloading.  This can be difficult, though not impossible.  Even the RIAA could establish actual access of only eleven of Thomas-Rassett’s many song files.  Less well-heeled plaintiffs might simply be unable to obtain the required proof.

To sum up about the distribution right:  to the extent streaming constitutes a reproduction of the work, it probably also constitutes a distribution of the work.  The problem is proving actual access and downloading of the work.

In the next post in this series, we’ll examine Grooveshark’s defenses.  Surely, you didn’t think Grooveshark went to all the trouble of starting and funding a legitimate business without some legal strategy?  I think that legal strategy is called “the DMCA safe-harbor provision.”

Thanks for reading!