Rethink Music: A Compulsory Sampling License

Note: This is a post I originally wrote on 26 April 2011, on a blog that I’ve since deleted. I’m reposting it here, as I’ve discovered a few people who liked the old post and wondered what happened to it. I’m not sure if I fully agree with what I wrote here; terms like “transformativeness” have taken on a new life in copyright law since this was written. But I think I still agree with the heart of what I argued here.—AFS

I’m spending the next couple of days back in Boston at the Rethink Music conference, with a humbling collection of music industry minds. I wanted to mark the occasion by bringing back a thesis about which I wrote extensively last year, following several discussions at the Future of Music Coalition Policy Summit in October 2009. In response to perceived market failures in the licensing market for sampling music, I proposed that Congress should develop a statutory license for sampling. The details of this proposal are below. The paper I drafted received a warm response from my peers (thank you), but needs substantial revision in light of Peter DiCola and Kembrew McLeod’s excellent new book, and my own Sisyphean push for excellence. I wanted to share my thoughts here and take this time to solicit some feedback. So here they are. Conference attendees, non-conference attendees, friends, strangers: rip this apart.

I’ll start with some background. Scholarship around music sampling is as diverse as it is endless, but I think there a few key points on which we all agree. First, sampling remains an important ingredient in modern music. Much of the literature focuses on the De La Soul and Public Enemy era of sampling, but music today still uses the technique extensively. According to Ultimate Chart, the top three artists right now are Katy Perry, Rihanna, and the Black Eyed Peas. All three are known to sample. Second, the jurisprudence around sampling — with circuit courts dividing on their interpretations of copyright — has created an atmosphere of legal uncertainty. It is virtually impossible to state with any confidence whether a given unauthorized use is permitted or infringing. Third, the industry’s response to the increased desire to license samples is incoherent and burdensome, creating overwhelming administrative costs and often resulting in situations where the would-be sampler is forced to surrender more than 100% of her income, especially when creating “mashup” music.

Two other arguments are necessary to accept as a predicate to agreeing to a compulsory sampling license. The first is that we as a society want people to sample music. As a cultural matter, sampling allows us to democratize our existing art, using fragments of our shared culture as building blocks to create new and powerful forms of music. DiCola and McLeod have provided many great arguments to back this up, so I will not repeat them here, other than to add that the benefits of sampling relate back to the sampled artist, too. As anecdotal evidence, consider the Imogen Heap album Speak for Yourself. After its initial fame in 2005, it was selling a few hundred copies a week in mid-2009. Then Jason Derülo sampled Heap’s “Hide & Seek” in his “Whatcha Say.” Derülo’s song peaked the Billboard charts in November 2009, and at the same time sales for Speak for Yourself spiked up as well, selling roughly 1000 copies per week from August 2009 through January 2010. (This is all based on Soundscan data I just pulled.) It’s safe to assume people liked Derülio’s (extremely prominent) sample, sought out the original accordingly, and Heap made money not only on the sampling license, but on sales of the original based on the subsequent fame.

Second, I would submit that most artists would like their work to be sampled, provided they are paid some reasonable form of compensation for the use. There are obvious exceptions to this — Prince and the Beatles chief among them. But artists tend to enjoy seeing and hearing their music take on new forms of life, provided respect is paid to the original. Again, DiCola and McLeod provide great evidence in their book. One omnibus exception to this is when a sample is used to suggest that the sampled artist is endorsing something that they would not want to be associated with. (Imagine a Tom Waits sample in a song licensed in a potato chip commercial, to analogize to a famous incident.) But there are ways to avoid that problem, as detailed below.

With these assumptions the argument for a compulsory license becomes clear. People like to sample because sampling sells, and we want to encourage people to sample because sampling adds value to our culture. The current law leaves ambiguity as to unauthorized sampling’s legality, so rational artists will try and obtain licenses. But the current licensing system is rife with market failure. Many artists have no idea who to ask for a license, and thus are deterred from the start. Still others will send in a request, but lack the prominence to become a priority for a licensor, and will never hear a yes. (This is especially dangerous for the longevity of our culture, as it means sampling will become a technique of the rich. Young artists will hear their heroes sample but will lack the means to do so themselves. We will lose the natural churn of artists that keeps music healthy and vibrant.) And even if an artist is lucky enough to get a negotiation started, the prevailing rates for sampling often create situations where the artist is forced to give up more than 100% of projected income. Rational artists will never sample. A compulsory license would remove the administrative hurdles and allow artists of all sizes to sample, while generating revenue for sampled artists they would not otherwise receive. Sampling songs can become like covering songs — everyone can do it, and everyone benefits. And, just like covers, a private-side body like the HFA could step in and become the de facto clearing house for sampling, allowing the government to set the terms of the license while not dealing with the everyday transactions.

All of this has been argued before, but this is the point at which the scholarship largely stops. I want to see if we can’t push the ball forward a little bit here. Here are my eight guiding points for a compulsory license:

1. A compulsory license need not be all things for all artists. A statutory license can sit atop of whatever else the industry adopts to address sampling. If people want to work around the license, fine. If they think what they are doing is “fair use” and proceed without one, fine. If the industry decides to grant a gratis license for noncommercial works, fine. Nothing here would replace those efforts. This is tailored to artists that want to sample and are willing to pay, but lack the means to negotiate a license directly or would be significantly disadvantaged by doing so. Keeping it narrow also helps keep it simple. Lawyers have a nasty habit of overthinking things like this. We want to take lawyers out of the loop here, and give artists something they can handle themselves.

2. The uses of samples should be transformative. Sampling serves its cultural purpose when it uses existing material as “building blocks” to create new and different forms of music. The compulsory sampling license should be used for this purpose. This should not become a shortcut for people to “sample” an entire song and thus become empowered to sell and license of the original music. Congress contemplated and rejected a statutory license for wholesale copying of sound recordings when it first extended copyright to sound recordings. I see no reason to rethink that decision. The biggest issue here, of course, is how transformative is “transformative.” I propose adopting a substitute goods analysis for this: if a typical consumer would buy the sampling song in lieu of buying the original, then it is not transformative. If not, it is. Note this is in lieu of buying the original, not just because they like the Jason Derülo over the Imogen Heap. The thought of the purchaser should be “I would buy the Imogen Heap song, but this is close enough.” (This approach bears some resemblance to Judge Posner’s method of analyzing fair use.) A transformativeness test replaces the need for a length-limit approach, but sampling length still can be significant, for reasons stated below.

3. Use percentages, not penny rates. A fatal flaw of many of the existing compulsory sampling license proposals is that they try too hard to mold sampling into the framework the law has adopted for covering music. This is inappropriate and anachronistic. Sampling and covering are very different animals, and what’s good for one may not be good for the other. Furthermore, musicians today are forced to diversify their income related to their sound recordings — selling recordings sometimes, giving recordings away at other times, and using licensing, touring, and special merchandising as supplemental sources of income. So instead of making artists surrender pennies they may not be receiving from the buyers of their music, the law should set a license based on a percentage royalty from income attributable to the sampling song, and broaden the percentage base to cover sync licenses and other alternative sources of income. (Relatedly, a compulsory license should allow for derivative works based on the sampling song, provided compensation is paid to the sampled song’s rightsholders based on how much of the sample is used in the derivative.)

4. Calculate the percentage based on the length of the sample used. The only way to make an effective and usable license for sampling is to calculate the royalty rates on objective characteristics. A logical objective characteristic would be the length of the work sampled. I know that some will suggest that some consideration should be made based on how the sample is used in the secondary work. I’m not opposed to this per se, but trying to classify uses of samples (“looping” versus “standalone,” “altered” versus “unaltered”) falls apart in practice, and, again, the license should be as simple as possible. I also know that many artists and scholars strain to find ways to quantify the qualitative “heart” of different samples. (My friend and sensei Dave Herlihy often speaks of the James Brown “heh!” as an example of a sample that is quantitatively short but qualitatively significant.) I recognize that not all samples are created equal, but relying on objective characteristics will allow subjective excellence to win out in the market. Subjectively great samples will be used more often. After all, artists generally sample things because they are qualitatively significant or interesting. The James Brown “heh!,” the Amen break, Atomic Dog, and Funky Drummer will see increased instances of sampling, which improves the sampled artist’s overall income.

5. Leave enough for the sampling artist, but match royalty levels to the current market rates as closely as possible. The point of this license is to shift sampling from a property regime to a liability regime, not to undercut the compensation levels existing for publishers and sound recording companies. As Terry Fisher notes, compulsory licenses are blunt and destructive instruments, and should be used in the least intrusive way possible. This fear of undue intrusion often circulates discussions of a compulsory sampling license: a big fear for rightsholders is that sampling revenues will drop down to the revenues expected for covering music. We should avoid this. The percentages currently used for sampling should be used, to the extent one can extrapolate those and put them on a duration-of-sample gradient. But no matter what, at no time should the total cost of sampling exceed 100% of income, or even 90 or 80%. The license should preserve the economic incentive to create; something must be left for the artist.

6. Distribute royalties SoundExchange-style, with shares for featured artists, non-featured artists, the sound recording copyright owner, and the composition copyright owner. The performance rights organization SoundExchange is responsible for collecting and distributing royalties earned by sound recordings transmitted digitally. They’ve developed a metric of royalty distribution where a percentage goes to the sound recording copyright owner (usually the band’s record company), a percentage goes to the featured artists (the “band”), and a percentage is set aside for non-featured artists (session musicians and the like) that had a part to play in the recording. All of these people, as well as the composer of the underlying song, are instrumental in making a great sample. All should be compensated. The percentage shares should mirror market practice, but I want to make sure the Clyde Stubblefields of the world get paid too. An outstanding question here is whether all artists on the song or just those that are represented in the sample should be compensated. I’m not sure how to handle this quite yet.

7. Give credit. A sampling license should contain a waiveable right of credit for use of the sample. This increases the “findability” of the original recordings, allowing for those records to be purchased and thus boost the sampled artist’s sales, and in a small way acknowledges the moral right of attribution, which American law doesn’t formally recognize but artists still seem to care a great deal about. Good samples should encourage music discovery, and the system should do all it can to foster that discovery. (And I know some artists take sample obscurity as a point of pride and deliberately hide samples. That’s fine. Just don’t use this license.)

8. Preserve a cause of action for right of publicity to guard against false endorsement. Every discussion of a compulsory sampling system always raises the hypothetical sampling use in a Republican campaign ad, or crass commercial use, or use in the context of Panzerfaust Records release. That is, of course, a legitimate concern for artists, but not an insoluble issue. The law already recognizes a remedy for false endorsement in the form of the right of publicity cause of action. Bette Midler and Tom Waits have both successfully defended against “sound-alike” recordings being used to falsely suggest endorsement of products. Admittedly, the law used by Waits and Midler is presently confined to the context of advertising, is based on state common laws, and is only actionable with a large degree of fame in the original artist. But the right could be expanded to cover the concerns above, provided certain First Amendment safeguards are in place (the sorts of safeguards already recognized in the doctrine of fair use). And what’s more, people already recognize that covers of songs often are done without the consent of the original songwriter and don’t hold the cover’s faults against the original. There’s no reason to believe that people couldn’t figure that out for sampling as well.

Those are my thoughts. Please tell me how and where I am wrong!

Director of the BU/MIT Technology Law Clinic. Technology law, intellectual property, and freedom of expression.