This is an authorized photo.

Photographer: Ethan Miller/Getty Images

The Dancing Baby Versus the Algorithms

Noah Feldman is a Bloomberg View columnist. He is a professor of constitutional and international law at Harvard University and was a clerk to U.S. Supreme Court Justice David Souter. His books include “Cool War: The Future of Global Competition” and “Divided by God: America’s Church-State Problem -- and What We Should Do About It.”
Read More.
( Updated
)
a | A

The mother who posted a video to YouTube of her baby dancing to a Prince song may not have had to take it down after all, the U.S. Court of Appeals for the 9th Circuit ruled Monday, because Universal Music, the copyright holder, hadn’t considered whether the video was fair use under copyright law before ordering it removed. But don’t go thinking the decision was a big victory for free information or fair use or even just moms who let their babies dance to Prince.

On the surface, the court struck a blow for fair use and against copyright holders. But it also suggested that computer programs that identify copyright infringement are sufficient for Universal and other copyright holders to avoid legal liability -- and these programs account for the vast majority of takedown notices issued every day. Thus, the court didn’t so much let the baby dance as Solomonically split the baby in half.

Start with the law, namely the Digital Millennium Copyright Act of 1998. The law creates an incentive for service providers like YouTube to take down material that copyright holders tell them is infringing. And it creates a duty for the copyright holder -- in this case, Universal -- not to misrepresent that an infringement exists.

In the olden days of 2007, when Stephanie Lenz uploaded a 29-second video of her baby dancing to “Let’s Go Crazy,” Universal employed actual human beings (OK, lawyers) to review content and send takedown notices. The legal department assistant who reviewed the video judged that Prince’s composition was “the focus” of the video and ordered it taken down.

The appeals court held that this wasn’t good enough. The law requires the taking down of material that violates copyright. But fair-use material, the court explained, isn’t a copyright violation at all. Universal was treating fair use as a kind of “affirmative defense,” which is legalese for saying that the user has violated copyright unless she can prove fair use. Not so, said the 9th Circuit. Universal has to show that there was a violation -- which means Universal has to determine that there was no fair use. If it didn’t, it could be held liable.

This sounds like a big deal. After all, the court was saying that Universal could actually be found liable for telling the mom to take down her video. And it is, in fact, important to clarify that fair use isn’t some exception to copyright infringement -- it simply isn’t copyright infringement in the first place. If I’m quoting you for purposes of commentary, criticism or parody -- what the law calls “transformative” uses -- then I’m perfectly within my fair-use rights.

Too frequently academic colleagues outside the law tell me they have been hindered in publishing their scholarly work because some publisher thought their quotations from other sources needed to be authorized, when in fact they are classic instances of fair use. The 9th Circuit deserves kudos for emphasizing the nature of the fair-use right.

Nevertheless, the court didn’t stop there. It went on to say that “a copyright holder’s consideration of fair use need not be searching or intensive.” All that’s required is a good faith belief that the material isn’t fair use.

The court then gave a road map for forming the good faith belief. It observed without formally holding so that “the implementation of computer algorithms appears to be a valid and good faith middle ground for processing a plethora of content while still meeting the DMCA’s requirements to somehow consider fair use.”

Yes, you read that right. A computer algorithm can act in good faith. Or more accurately, a corporation that uses an algorithm to measure infringement can be said to satisfy the good faith requirement.

Does this make sense? Well, the algorithms that the court had in mind check that the material matches the copyrighted source and that all or almost all the posted video matches the source. The overwhelming majority of pirated video and audio on the Web is just that: straight rip-offs of copyrighted material. The programs find these easily, and send an automated takedown message. And the court signaled that this was adequate.

The court said that for all other potentially infringing material, Universal and other copyright holders could employ humans. The humans could make a fair-use judgment, because a computer can’t yet identify parody or commentary.

This part is a win for Universal and its ilk. They want to protect their copyright at the lowest possible cost. And algorithms are cheaper than people.

I confess to being a little worried about the use of quantity and entirety to make fair-use judgments. I can easily imagine a parody in which the image is preserved entirely while the words are changed, or vice versa.

From the other side, you can imagine pirated copies being embedded in extraneous material to avoid the algorithms.

But it’s also true that the scale of the Web is such that there pretty much has to be an automated solution to the infringement problem. The 9th Circuit was being realistic. Its compromise is reasonable. It just isn’t a big win for fair use.

(This article, originally published Sept. 16, has been corrected to reflect that Universal Music, as the copyright holder, was the defendant in this lawsuit and is the company that must show it considered whether the video was a fair use of the Prince song before asking for its removal. YouTube does not bear that burden. The headline and first, second, third, fourth, fifth, sixth, 12th and 13th paragraphs have been corrected.)

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

To contact the author of this story:
Noah Feldman at nfeldman7@bloomberg.net

To contact the editor responsible for this story:
Stacey Shick at sshick@bloomberg.net