Friday, February 08, 2008

Columbia Fair Use conference: Panel 2

Panel 1 was mine; notes on it later, when I've written them up.

Panel 2: The Statutory Factors Reconsidered

Moderator: June Besek

Barton Beebe: Beebe summarized his empirical study on the role of the factors. Reminder to us: transformativeness is probably overrated academically, given its less prominent position in the decided cases. He found no real relation between judicial ideology and fair use outcomes.

Robert Kasunic: Not giving us the official position of the Copyright Office; he took the day off to come here. He wants to find a normative purpose for copyright, which will help us with fair use, which isn’t an ancillary doctrine but an integral part of copyright. The Court has told us that copyright’s purpose is to incentivize authors to create and disseminate works to the public.

How can the second factor assist in this goal? Court cases suggest it’s not that important. Kasunic wants us to reevaluate its scope and purpose. It shouldn’t just be published/unpublished and factual/fictional. Harper & Row led to rigid, deterministic analysis for unpublished works; in published work cases, courts just looked at factual/fictional. We might ask what types of incentives this type of work needs to get created (that could also be a factor four inquiry). The Sega case involves close attention to the type of work – a computer program, which required a different kind of copyright analysis.

Next, after looking at the general, we could ask where a work falls within its category. We can learn more about customary and traditional markets, and the reasonable expectations of authors prior to their creation of a work – what was their incentive to create? Different classes of works are represented by different industries with varying practices and licensing mechanisms. We should ask about a fair return, a return that was enough to provide the incentive to create, not all possible returns. And we also can look at whether there’s a divergence of interests between the initial author and the current owner; if the incentive function is designed to get authors to create, then incentives for the transferee are only indirectly relevant.

Joseph Liu: Seeking modest improvements in predictability within the existing statutory structure. What if we just asked about the purpose and character of the use and the effect on the market, turning it into a two-dimensional balancing test? Kasunic wants to rehabilitate the second factor, and Liu to kill it off. Many cases already deemphasize the second and third factors; Beebe’s work confirms these factors rarely drive results. But expressly abandoning 2 and 3, which often serve to frame the analysis, might allow courts to consider user and copyright owner interests more directly and calibrate their intensity.

Perhaps this would also help intrafactor doctrine. To some extent, courts have looked at types of market harm, but factor one analysis could also be deepened, looking at the types of artforms at issue (documentary, rap) and the benefits of various uses.

Liu largely agrees with Kasunic about the diagnosis, but the remedy (get rid of factor two, removing information versus enhance it, adding information) is where they diverge.

June Besek: How should we be weighing predictability v. nuanced analysis? If you want clarity, you need guidelines. Would removing factors 2 and 3 be any help? People rarely get bogged down in 2 and 3.

Liu: You’ll always have to pick a point on the spectrum between flexibility and security. He admits that his proposal is a thought experiment. Indeterminacy, like the poor, will always be with us. His idea is that stripping out extraneous factors makes it less difficult to think about and predict fair use situations.

Besek: If you get rid of the unpublished nature of the work, you’ve reduced everything to an economic decision about when to publish.

Liu: Yes, and that’s my normative bias.

Beebe: There are a lot of close cases, but if you look at the reversal/appeal/dissent rates, fair use case law is exactly in line with every other area of law. So Beebe tentatively suggests that the case law is not the morass that it appears to be, at least the non-leading cases.

Tim Wu: A friendly question in an aggressive manner: Aren’t you all missing the point? If you believe that judges make up their minds first and give opinions later, then factor analysis is not that important. Then isn’t the real question the treatment of precedent, if we’re interested in predictability? That is, whether judges regard existing fair use cases as creating rules they have to follow in the specific, or the general – e.g., Campbell means that parody is fair use; Harper & Row means that scooping a soon-to-be-published book is unfair. This comes up in the Google cases: if a market emerges for something, can it change the result in Kelly?

Kasunic: Bias is a big problem, even if the judge has no intentions – people view copyright either as a property interest to be preserved or a more limited right that’s a means to an end. Predictability would require agreement on the purpose of copyright, which will always haunt the fair use analysis. Patterns and customs do matter – best practices in different categories of authorship may be the way to go, because you’re differentiating between types of works and uses.

Liu: Goldstein’s view is common-law lawmaking by the federal courts. If that’s right, precedent is key: how tight is the link between past and current cases?

Wu: It’s not clear that someone would be willing to bet the company on a two-factor test either. Predictability comes from precedent.

Ginsburg, for Kasunic: How do we look at markets and time? If the market didn’t exist at the time the work was created, then can the author ever claim a right to it? The market that didn’t exist in Kelly developed by Perfect 10. (Comment: I beg to differ; Perfect 10 didn’t show a general licensing market had developed, just that it had granted a license overseas.) Likewise, if an author writes a story because she feels a compulsion to write it, how do we take that into account if someone then wants to copy it?

Kasunic: It’s important to recognize if a work was created without reference to a market. If the market develops, that’s a windfall, which doesn’t cut in any necessary way but ought to be recognized. His analysis is a way of separating all possible markets, which is an unlimited category, from actual incentivizing markets. That may mean that a newly created work has a more expansive market expectation!

Jessica Litman: Kasunic’s reformulation lets us account for some things that we occasionally pull into factors one and four but shouldn’t. Barbie is an icon, 50 years old, and symbolizes a great deal about how we think about femininity and growing up in this country. She’s a much more appropriate target for commentary than one of the Bratz. Using Gone with the Wind or Barbie as your target is an important indicator for fair use (compare Koons’ use of an unknown photo).

Judge Leval: By the time judges come to discuss 2, they’ve already discussed 1, and they are already implicitly discussing both factors – transformativeness requires consideration of the nature, purpose and character of the first work in order to compare it to the accused work. So 1 and 2 together lead you to comparison.

Kasunic: That’s a reasonable way to approach it. The legislative history of 2 is scant, but some of the things Congress pointed to were the purpose and nature of the original work.

Beebe: His empirical numbers support this idea of merger between 1 and 2 taking place.

Matthew Sag: There’s fundamentally a limit to what we can deduce about fair use factors from litigated cases – the fact that a case didn’t settle makes it anonymous. Most of the impact of factor 2 may be felt somewhere else than the decided cases.

Q: In lots of publishing industries, people rely on factors 2 and 3 more than 1 and 4, so eliminating them would cause a lot of uncertainty!

Liu: good point, and he wants to consider that as part of his thought experiment. (Comment: transitions always cause problems! Just like a making available right would be a tough transition in the US where we’re used to dividing rights up.)

2 and 3 are more objective and less dependent on characterization by plaintiff and defendant. But do they help point us in the direction we want to go? Taking the entire work can be okay, if it’s a reverse engineering situation or Sony or Kelly. So you can get precision, but not accuracy, from using them.

Q: Richard Prince’s photos of Marlboro ads: How would they fit into this?

Liu: He believes that 1 and 4 provide almost all we need to know – market impact (which markets count?) and nature of the use (does the art-world reaction matter? Do we want to encourage appropriation art?). 2 and 3 don’t tell us very much – that the original work is an ad, but also a creative work, points in no particular direction, especially since appropriation art is generally going to appropriate creative works; likewise the fact that Prince copied the whole work doesn’t add much to 1 and 4.

Kasunic: He’d want to focus on the fact that the ad was an ad, to start with.

No comments: