The RIAA Strikes Back
So it turns out that the RIAA isn’t cool with ReDigi. So much for the maybe-they’re-secretly-licensed theory. You can read the RIAA’s demand letter to ReDigi here. It raises an issue I hadn’t considered before, so wrapped up was I in applying Vernor and the first-sale doctrine. RIAA questions whether the first-sale doctrine even applies in the first instance. If the RIAA is right, then my previous analyses about ReDigi’s right to the first-sale doctrine are cut off at the knees.
Warning: this is an unusually long post (even for me), but there was no way to split it up without ending with a misleading conclusion. In light of the controversy that ReDigi (and by extension your blogger) has gotten into, I thought it best to keep this as one long post rather than split it up.
The RIAA’s Legal Position
The RIAA’s logic goes like this. When you sell a music file over the internet, you aren’t selling the copy that’s actually resident on your computer (or wherever). You’re selling a copy that ReDigi made of your song. According to the RIAA, the first-sale doctrine applies only to the exact copy of a work that you lawfully acquired. Since ReDigi’s copy of your song file isn’t lawfully acquired, that copy of the song file isn’t covered by the first-sale doctrine.
By contrast, the way I had been looking at the problem was to ask whether the first-sale doctrine applies to the original copy, then wonder how you can legally make intermediate copies to transfer the song file to the buyer. The RIAA is basically saying I have it backward–those intermediate copies (and most particularly the permanent copy that ends up with the buyer) are what’s being sold, and it is to them you should apply the first-sale doctrine.
I know this sounds like a distinction without a difference. After all, we buy things, like software, online without a second thought, and it feels like we bought it at a store. Who cares if you need to make a few intermediate copies to transfer a file from point A to point B? Why should it make a difference that you sold 10 songs on a single CD by handing the CD to the buyer, or that you sold the same 10 songs from your hard-drive (erasing the songs as you did so) by transferring the files over the internet?
But, as I’ve mentioned before, just because the result is crazy, that doesn’t mean it isn’t copyright infringement. Copyright law doesn’t always recognize crazy. So, let’s get into the existential question of when a distribution is a reproduction, and when a reproduction is a distribution, and when they might exist simultaneously. Honestly, it might be easier to try to understand Kerner’s brilliant discussion quantum mechanics from Hapgood*.
* I love this play. I used to teach it in my science and literature course.
First, a confession and an assumption. I’m a little confused about exactly how ReDigi’s technology works. According to ReDigi, its technology ensures that two copies of the same song file are never in existence at the same time. But it also says that the user must upload a file for verification (to make sure it was legally purchased), and only then is the file “accepted” and the original file deleted. Both of these statements can’t be true because, while the file is being “verified,” a second copy must exist. Previously, I had assumed that the verification occurred at the level of the user’s computer. I’m not really sure which to believe. For purposes of this discussion, I’m going to give ReDigi the benefit of the doubt and assume that at no point does two copies of the same song file exist at the same time.* (I’m also going to assume that the verification engine never mistakes an illegal song file for a legal one.)
* This sounds impossible, but perhaps it’s not. There is, after all, a delay as the file transfers over the internet. All of the packets could be in transit as the original file is wiped. In theory, ReDigi could even delay slightly the recomposition of the packets into the new copy until all packets are accounted for and there is confirmation that the original was wiped. I’m not sure if ReDigi goes to all of this trouble–but it might be worth it.
Where to start? Well, the RIAA helpfully cites its legal authority for its position in a footnote from its demand letter to ReDigi. It may be summarized as follows: (1) the actual language of the first-sale statute, (2) three court decisions, and (3) a report by the U.S. Copyright Office.
The Actual Language of Section 109(a)
Section 109(a) of the Copyright Act codifies the first-sale doctrine in pertinent part as follows:
Notwithstanding the provisions of section 106(3) [the distribution right], the owner of a particular copy or phonorecord lawfully made under this title … is entitled, without the authority of the copyright owner, to sell or otherwise dispose of the possession of that copy or phonorecord.
(Emphasis added.) The way this statute has been traditionally understood is: if you sell a copy that you own of a copyrighted work, the copyright owner can’t assert the distribution right against you. In the context of digital content,Vernor tells us whether you really “own” it or not.
The RIAA appears to have two different effective arguments based on the actual language of the fair-use statute: (1) that it works only against assertions of the distribution right, and (2) that it applies only to transfers of the exact same material object (e.g., hard drive) to which the song file was originally downloaded.
Is the First-Sale Doctrine Limited to Distributions?
The RIAA’s first argument is that the first-sale right only works against assertions of the distribution right. We’ve always assumed that the first-sale doctrine is limited to the distribution right. The statute actually references the distribution right with its initial notwithstanding clause. Numerous court cases say so. Nimmer (the leading authority on copyright) says so.
Working on this assumption, the RIAA is saying: Go ahead and sell your digital downloads. We won’t try to stop you from distributing the copy you bought.* But if you try to make a copy of that copy, we’ll come after you, because the first-sale doctrine covers distributions, not reproductions. The argument is a lot like Portia’s (winning) pound-of-flesh-but-not-a-drop-of-blood argument in The Merchant of Venice: go ahead, distribute you want, but just don’t make even a tiny copy along the way.
* Well, maybe they will anyway, if they think the sales were really licenses under Vernor.
If the RIAA were responding to my last post on this subject, in which I suggest a fair-use defense for the necessary copying of the original song file, they’d respond: you’re confusing the right to (re-)sell with the right to distribute. You assume that once you come under Vernor, you have the right to take steps necessary to effect the re-sale, but the first-sale doctrine is strictly limited to distribution.
The main response to this argument is: Regardless of how we’ve always looked at the first-sale doctrine, it is, at heart, a right to re-sell (or give away or lend or whatever) a legally-acquired copy of a work. Until now, we’ve never needed to look beyond the distribution right. But today, a very common type of commerce requires making temporary copies of things in order to sell them, and copyright law can accommodate new ways of doing old things.
The obvious objection to this line of thinking is the notwithstanding clause at the beginning of the statute. It actually specifically references the distribution right and no other right. Doesn’t that mean that the first-sale doctrine works only against the distribution right?
What’s more, the notwithstanding clause is relatively new. In the 1909 version of the Copyright Act, the statute describing the first-sale doctrine made no reference to the distribution right (or any exclusive right) but merely protected the right to “transfer … any copy of a copyrighted work, the possession of which has been lawfully obtained.” In the current version of the first-sale doctrine (in 1976), Congress added the reference to the distribution right. Surely that means Congress intended to limit the doctrine to the distribution right.
These objections are themselves not air-tight. First of all, although Congress added the notwithstanding clause, it also announced that, in enacting the current version of the first-sale doctrine, it intended to “restate[] and confirm[] the principle … established by” the old version of the first-sale doctrine. H. Rep. at 79. The reference to the distribution right may not be as meaningful as it first appears.
Perhaps more important, notwithstanding clauses are often just emphatic in nature, not substantive. They usually mean, “despite what you may have read.”* In this case, it might mean nothing more than, “We’re focusing on the distribution right, but we’re not necessarily limiting to the distribution right; it’s just that the distribution right is the right most clearly implicated.”
* Not always. Sometimes you have a situation in which two competing rights appear to apply, and you’re not sure if Congress meant one to trump the other. In that case, a notwithstanding clause can be substantive (or at least clarifying) in nature. But even then, the clause acts to empower its statute, not to limit its application.
Further, the use of entitled strongly implies the grant of a right, in this case, the right to “sell or otherwise dispose of” the copy that the putative seller owns. When it’s already clear what the scope of this right is, do we really need to read into it another restriction?
In the end, the RIAA’s argument boils down to: the first-sale doctrine has always been limited to the distribution right; everyone says so. It’s up to Congress to fix that. ReDigi’s argument is: yes, but, that’s only because the courts and the commentators didn’t know better because they’ve never before been confronted with this situation.
Is the First-Sale Doctrine Limited to the Original Phonorecord?
The RIAA has a more powerful and fundamental argument. The first-sale doctrine is limited to whatever medium is embodied the song file at the moment it was purchased. Thus, if you downloaded the song file onto the hard-drive of your 1.7 GHz P4 Dell in 2001, your right to re-sell the song is limited to selling the entire hard-drive. (In theory, if your song library migrated to a better computer, probably legal under fair use, then maybe you could sell the newer hard drive.)
The RIAA’s argument focuses on the term phonorecord*, which is defined in the Copyright Act as a “material object[] … in which sounds, other than those accompanying … [an] audiovisual work, [are] fixed … and from which the work may be perceived, reproduced or otherwise communicated, either directly or with the aid of a machine or device.” The first-sale doctrine is limited to re-sale (or whatever) of “that … phonorecord” that you lawfully own. In your case, that means your hard drive.
* Actually, copy or phonorecord, but I don’t think copy operates any differently from phonorecord in this context.
Can it really be that simple? Outside the context of digital downloads, the distinction between a “sound recording” (the song in an idealized form) and a “phonorecord” (the actual thing you put into your CD player) makes sense. But in the context of digital downloads, the distinction gets weird.
For example, you “own” your hard drive. You “own” the song file (under Vernor). You can sell your whole hard-drive*, but you can’t sell the song you “own” by itself because, to do so, you must make a new phonorecord, and that phonorecord isn’t “lawfully made” because it violates the reproduction right.
* Assuming you remove all of the copyrighted material thereon that you don’t “own” under Vernor.
Or weirder still, what if you “own” a song file (again, under Vernor) but you’ve always stored it on someone else’s hard drive (e.g., in the cloud)? You don’t even have the option of re-selling the phonorecord since you don’t own it, even though you own the song file.
But wait, is really the case that the new phonorecord wasn’t “lawfully made”? A while back, we discussed how it might be a fair use for consumers to store copies of their songs “in the cloud” under a theory of “space shifting.” (Admittedly, I threw a bit of cold water on that theory, but it’s still a viable one.) If you can store an additional copy of a song file in a remote location via the internet for personal use, could you not also temporarily store a copy of a song file so you can complete a sales transaction involving that song file, especially if you wipe the original song file immediately upon uploading.
The situation isn’t precisely analogous. In the case of the Amazon Cloud Drive, you are storing the music for personal use, but you are doing so more or less permanently. In ReDigi’s case, you’re storing it temporarily, but so you can sell it.
Does a Star Trek Transporter Violate Copyright?
Actually, do we even need fair use? Is this even a “reproduction”? Going back to the big assumption I made earlier, if ReDigi wipes the original file at the same time as it reconstructs the copy of that file, is it really reproducing the original file?
We have no pre-digital analogue to fall back on in this case. This isn’t like making a sculpture out of a copyright-protected mold because that would positively increase the number of copies in the marketplace. It isn’t quite like photocopying a book, selling the photocopy while destroying the original book, because (1) why would anyone do that? and (2) the photocopy is still a separate physical item. It’s a bit more like unwinding someone’s sweater, then knitting the exact same sweater from the same wool, only the “wool” isn’t exactly the same.
If anything, it’s most like a Star Trek transporter: does it physically move the atoms that make up its travelers, or does it destroy the travelers then makes a perfect copy somewhere else*, or does it matter?
* I suppose it converts the travelers into “energy” then reconverts that “energy” back to matter, but isn’t that exactly the same thing as “destroying” then “reproducing.”
As a starting point, we all have to agree that if you arrange the 0’s and 1’s that comprise a song file on a different medium, while keeping the original set of 0’s and 1’s on the original medium, you’ve reproduced the song file. If you don’t believe that, then copyright law doesn’t apply to digital content at all.
OK, let’s say you copy the contents of your hard-drive, including the song file in question, onto a new hard drive (maybe as part of a migration), using a USB cable and a protocol that overwrites data blocks on the old hard drive as soon as the new data blocks are written and confirmed. Is that a reproduction of the song file?
OK, same thing, but you’ve moved across country at the same time you’ve upgraded to a new computer. Because you’re just that sort of person, instead of physically hauling your old computer across the country to where your shiny new computer is waiting, you do the migration over the internet, using the same protocol. Is that a reproduction of the song file?
The question is whether reproduce, as used in the Copyright Act, has as a requirement the simultaneous (even if temporary) existence of two copies of the work. In other words, does it mean “duplication” or just “production.” The Copyright Act doesn’t define reproduce, and even if it did, it’s unlikely that it would have had the foresight to address this problem. Even Professor Nimmer isn’t sure of the answer.
Next time, we’ll take a look at the three court decisions cited by the RIAA. But be warned: one of them is completely bonkers.
Thanks for reading!
A very thoughtful analysis of a challenging issue. But the issue is not at all novel. The whole concept of when reproduction becomes distribution and vice versa was endlessly debated in Congress and at the WIPO (under US urging) from 1989 to 1995; consulting the debates there will give you a better sense of the full range of arguments debated – and partially resolved in the WIPO Internet Treaties (1996) and subsequent implementing legislation. (And yes RIAA participated vigorously in all of those debates and largely prevailed).
Two points need to be added, however. First, you overlook the fact that the First Sale Doctrine only applies internally, in the US. The Importation Right is an exception to the First Sale Doctrine and will always defeat any attempt to bring into the US a copy made or purchased abroad unless the rightsowner authorizes the importation. 17 USC sec. 601. It will be the rare digital file processed through ReDigi that bears no international, ex-US footprint. And so it can be barred from qualifying for the First Sale Doctrine privilege for that reason alone.
Second, your claim that ‘In theory, if your song library migrated to a better computer, [it would be] probably legal under fair use’ seems very far-fetched. Assuming you mean ‘migrated’ to be a euphemism for ‘copied without authorization’, then the pure substantiality of the copying would surely mean that no court would see this as a true instance of ‘fair use’, not would this kind of copying qualify for any of the uses that are cited in the introductory clause of the fair use provision: it is not scholarship, criticism, research, etc., nor any other qualifying use. Besides, and more fundamentally, there is no ‘fair use’ of a sound recording in any instance. That was the bargain that underlies the 1974 amendments to section 114 of the Copyright Act: sound recording owners lost the right to authorize or prohibit derivative works that every other work enjoys; in exchange they got the right to stop any reproduction, no matter how small, of the master sound file, including a single note. (See Grand Upright case). Otherwise, the entire market in sampling of sound recordings would not exist. Those all qualify under even the strictest fair use analyses. But, alas, there is no fair use in sound recordings. Reproducing a single note from a sound recording is infringing activity that cannot be saved by a fair use defense; much less can reproducing your entire music collection be seen as fair use.
Still, your conclusion is correct; but for a different reason. Realizing that this state of the law skewered things a bit strongly in favor of the record labels, Congress took away from them the right to sue for copyright infringement if the unauthorized copying is done solely for ‘noncommercial use by a consumer’ of any digital audio recording device. 17 USC sec. 1008. This home use exception applies across the board to all instances of unauthorized home copying of sound recordings, including your entire music collection, as in your hypothesis. You still have infringed copyright. But the rightsowner has no remedy against you. It is taken away by sec. 1008. (This was the compromise under the Audio Home Recording Act of 1992 whereby the record industry foolishly agreed to give up their infringement remedy at law in exchange for an illusory tax on blank digital audio media).
Thanks for reading, and thanks for the analytical comment. Sorry the Thanksgiving holiday delayed my response.
First, I’ve been assuming that ReDigi is limited to U.S. customers (or at least U.S. sellers). (I can’t tell from ReDigi’s site–does anyone know?) If that’s the case, wouldn’t most, if not all, transfers be entirely within the United States? And if a few happen to go through, say, Canada, would they really be that easily traceable? Even if there were an importation, don’t we need to know where the phonorecord was “manufactured,” so we know whether Quality King or Omega applies, and even if Omega applies, are we really sure Omega is going to be adopted outside of the Ninth?
Second, by “migration”, I only mean a copying from one device to another of all of the first device’s contents, followed by the wiping of the first device’s storage. It’s something everybody does any time he or she upgrades a computer. My hypothetical simply imagined a migration over the internet (which, I’ve been told, actually happens). Whether that reproduction is “unauthorized” is precisely the question. I find it hard to believe that a traditional data migration isn’t a fair use (either computer-to-computer or over a LAN), and I don’t see why it would make any difference that it occurs over the internet. (Obviously, ReDigi is a different situation, but I was leading up to that.) I simply cannot agree that there is no such thing as fair use in the context of sound recordings in any instance. I don’t see where Section 114 provides so, Grand Upright doesn’t hold so, and I’ve never heard of any other decision holding so. If this were a blanket rule, a lot of celebrated cases in the last 10 years would have a lot shorter! Even in the Bridgeport/Dimension Films case, 410 F.3d 792, which held there was no “de minimus” defense for sampling sound recordings, the 6th Circuit invited the lower court to consider a fair use defense.
Third, using 17 U.S.C. 1008 is a really intriguing idea. If we applied the spirit of 1008, it might work, if we could convince a court that the transaction ReDigi is brokering is “noncommercial,” for the reasons you state. Read more technically, however, section 1008 appears limited in a way that would exclude almost any device to which the song files might copied. Under the Diamond Rio holding, 180 F.3d 1072, MP3 players (and a fortiori general-purpose computers) are not digital audio recording devices for purpose of the Audio Home Recording Act. (The Act, alas, seems to cut both ways in this regard.)
I’m interested in your thoughts! Thanks for reading!
An interesting analysis. In my view, if ReDigi’s technology works as you presume for this piece (deleting as it transfers), in Australia they would face no such issues in terms of the temporary copy. Essentially, the act of transfering does not amount to a reproduction given that “the data [is] not in a material or corporeal form, but in a non-material, incorporeal form, comprising essentially electronic impulses” Stevens v Sony [2005] HCA 58 at 68 (http://www.austlii.edu.au/cgi-bin/sinodisp/au/cases/cth/HCA/2005/58.html). This is a decision of our High Court, equivilent to the US Supreme Court. Interesting in that decision is the discussion of the treatment under US law at p 77-78.
Thanks for the comment and the insight into Australian law! From a quick reading, it appears that Australia doesn’t define “fixed” to include purely mechanical copies (i.e., that no human being could possibly ever read), and so would reject MAI. But what about things that can be perceived only through a machine? And what does this do to the Australian software industry? Without MAI, the U.S. software industry would be VERY different…