Sampling and innovation

  1. Excludable music
  2. Creativity and excludability
  3. Is intellectual property law necessary?

Excludable music

If you think of music as a recipe for sound, then the non-rival interpretation is straightforward. As you probably are well aware, the same song can be played on multiple radio stations at the same time, and people can be streaming it all over the place. Copies of music are essentially zero cost, making music non-rival.

The theory of growth we developed suggests that if we want people to make new music - to innovate - then we are going to need to make that music excludable and allow the artists to reap some economic profits. Spotify pays each artist some small amount for every stream, and radio stations do something similar. This is covered by intellectual property laws involving copyrights. And it seems hard to argue that at least some of the motivation for people making music is to get rich.

So, economics for the win? Kind of. Because this gets to the intriguing point that “excludable” is not a binary concept, and thus the question of what is excludable in music is a gray area. And once we consider that there is a gray area regarding excludability, we can consider that perhaps there is a gray area in the definition of rivalry, with implications for how we think about growth.

I think the best way to consider this is with the process of sampling. In the middle of the 1970s Grandmaster Flash comes up with quick-mixing, which involves building continuous music out of two copies of the same album, flipping back and forth using two turntables and a fader.

The break-beats that Grandmaster Flash was using to build the continuous sounds were samples out of other songs. He didn’t create the original music, he cut snippets of it together into something different. And here’s the question. Is the sample excludable? Does the original artist have standing to stop Grandmaster Flash and every DJ and producer who came afterward from using their sample? Do the DJs and producers have to pay up every time they sample a song?

For decades after Flash created quick-mixing no one cared. Hip-hop was small potatoes in the music business, so it wasn’t very relevant to record companies or artists if some DJ sampled their song. That started to change in the late 1980s, when hip-hop started to get huge, the money got real, and artists wanted to assert their right to exclude others. In practice, they wanted to get paid to allow the use of their original music.

This all came to a head in 1989 with De La Soul. On their album 3 feet high and rising they had a song called “Transmitting Live from Mars”. It’s short, take a listen to the first 10-15 seconds or so.

This was sampled from the Turtles song “You Showed Me”. Take a listen here to the first 10-15 seconds as well.

The Turtles sued. And won. One member of the Turtles is quoted in this article: “Sampling is just a longer term for theft. Anybody who can honestly say sampling is some sort of creativity has never done anything creative.” Harsh. And while the Turtles didn’t say it this way, they were asserting that any use of their music was excludable, and hence they retained any and all rights to it. Just as an aside, the Turtles didn’t write “You Showed Me”, they just recorded a version of it.

This changed the terms of hip hop. Producers had to be more careful with getting “clearance” from original artists, which could be simply permission to use the song, or might involve a payment. The point was that the original artist had the excludable right to make that decision. This didn’t stop sampling without permission, in a lot of cases because the original artist may not have realized that the sample was made. Here’s just one example involving Jay-Z and Timbaland.

What does our theory say? It says that this enforcement of copyright laws should have increased the incentives to create original music, and reduced the amount of sampling going on. There is no formal econometric test of this I know of (and I don’t think one is possible) but it seems hard to claim that hip-hop involving samples declined in relevance compared to original music after 1989.

This will ruin your productivity for the day, but check out WhoSampled and go down the rabbit hole of the sampling. If you don’t know where to start, try the most-sampled tracks list. The top one, the “Amen break” by the Winstons, appears on over four thousand different songs.

Creativity and excludability

The Turtles lawsuit changed the terms of hip-hop nevertheless. If you go listen to the whole De La Soul album or to the Beastie Boys Paul’s Boutique and compare them to hip-hop produced in the 1990s, you’ll hear a difference. The 1990 samples are shorter and less identifiable. Producers started slowing down beats or hooks, or slicing them up, so that they weren’t noticeably similar to the original any more. They created legal space between their work and what they sampled from.

At some point this becomes a discussion of what constitutes an “original”. Watch this about J Dilla, who has been described as “your favorite producers favorite producer”. There is no question (in my mind) that Dilla created new music, even though almost everything he did involved a sample of an old song.

Here’s Mark Ronson (producer for Amy Winehouse, among others) on the whole concept of sampling and whether it constitutes “copying”. His argument is that sampling is how others interact and engage with music, not really different than someone dancing to that music.

One point to draw from this is that the lines deciding what is excludable and what is not are arbitrary and fuzzy. The Turtles case could have been decided the other way; it wasn’t obvious that De La Soul had done something wrong in creating what was arguably an entirely new song.

A second point is that hip-hop samplers were not arguing for an intellectual property rights free-for-all. Dilla’s music is all copyrighted, as is Mark Ronson’s. They are just as protective of their rights over their creations as the Turtles were. The argument was not about whether innovation should be protected by intellectual property rights, but where the lines of that law should be drawn. Ronson and Dilla cashed their checks.

Is intellectual property law necessary?

There is a different gray area here we can explore, which involves the rivalry of music. Once a song is recorded (a sampled song, a more traditional song, whatever) this seems to be pretty clearly a non-rival good, or something very close to “perfectly” non-rival. Copying music is straightforward, and making a copy of a song (to an MP3 file, to a cassette tape) means I can listen to a song at the same time that you do.

But not every innovation is so clearly non-rival. Go back to Grandmaster Flash. He not only invented the quick-mix, but had to invent the technology to do the quick-mix in the first place. Watch about three minutes of this interview/demonstration with Flash, where he describes some of his motivation for coming up with the technique. More important, watch until his buddy Mike (far left) talks about how long Flash was at this in the mid-70s.

Three years to figure out how to do the quick-mix. And Grandmaster Flash, as far as I can tell by searching through the US Patent Office website, holds no patents on any technique or equipment. So he holds no intellectual property rights over his innovation.

And yet, he innovated. Not only innovated, but turned those innovations into a multi-decade career. The producers that came after him essentially stole his technique, indicating that those techniques are non-rival. And to some extent they are not excludable. Flash couldn’t (or didn’t) stop other producers from using his technique, and now even Taylor Swift songs contain samples.

Does this represent a failure in innovation policy? Should Grandmaster Flash be entitled to a little piece of the royalties on essentially every song produced since about 1987? Or put this a different way. How does Flash pay his bills without those royalties?

He pays the bills by being Grandmaster Flash. And while his techniques are non-rival, they are excludable without intellectual property rights by virtue of being hard to replicate. You can give me two turntables, a mixer, and two stacks of records and in thirty years I’d never be able to do what he does, let alone figuring it out in three. You can put me in a state-of-the-art sound studio and I still wouldn’t be able to do what Flash or any other competent producer in the world does.

The issue is, of course, that I have like zero musical talent. I have no ear for this stuff, nor do I have the patience to sit there are dig through old 70s funk albums looking for a three-second hook that will turn into your favorite song next summer. The excludability involved in music production comes in large part from talent, not laws.

Michele Boldrin and David Levine are economists who have taken on this type of situation, and argue forcefully that intellectual property laws are in fact stifling innovation. Part of their argument is that excludability is more of an inherent property - as with talent - than intellectual property laws allow for. They would suggest that the situation with Grandmaster Flash and subsequent producers is the situation that maximizes innovation and hence the creation of new music. If everyone had to pay royalties to Flash, it would shut down innovation as marginal producers decided it wasn’t worth it.

Biasi and Moser have a paper showing that relaxing copyright laws led to more innovation by allowing a larger group of people to get access to existing knowledge.

The basic tools of the model we developed so far are still the same, for them - production functions, capital accumulation, productivity growth - but the assumptions about rivalry and excludability are different. Because intellectual property is in some sense embodied in people (e.g. as talent) then there is no true non-rival input in production. And hence in their conception it is quite possible for a constant-returns economy (in capital and labor) to pay inputs their marginal products and support innovation. Whether the basic model I started with or the extreme version of Boldrin and Levine is more appropriate depends on your conception of how intellectual property works. If most IP is like music itself - easily copyable and non-rival - then then basic model makes sense. If most IP is like the skills involved in producing - inherently excludable and rival to some extent - then Boldrin and Levine is the better setting to use.