Saturday, September 26, 2015

Notice and Notice Failure at BU, panel 5

Lydia Loren – Fixation as Notice in Copyright Law
3 different roles: (1) protectability; (2) infringement—reproduction etc. require fixation; (3) preemption—no preemption for unfixed works.  What kind of a notice function might fixation play in each area?
Statutory definition: fixed when its embodiment is sufficiently permanent/stable to be perceived, reproduced etc. for a period of more than transitory duration. Copies are material objects (other than phonorecords) in which a work is fixed by any method now known/later developed. 
Protectability: Early cases: AV works can be fixed in software.  Recently: Kelly v. Chicago Park District, wildflower garden wasn’t fixed/copyrightable.  Garcia v. Google: performer didn’t fix performance under her authority.  Kelly says: “the law must have some limits”—authorship is a concern as well as fixation.  The real barrier to fixation is essential, not temporal—the garden may continue season to season but its nature is dynamic change.  Garcia: the creator of the film fixed the performance; Garcia played no role in the fixation.
Infringement cases are about digital copies.  MAI gives us RAM copies as physical embodiments, but Cablevision says there’s a duration requirement and partial, constantly overwritten (after seconds) buffer copies don’t qualify.
What is this really about?  Decisions reflect underlying instinct about a type of notice: crossing boundaries into something eligible for copyright protection/remaining in state-law eligible area.  Entering into something that qualifies for a type of infringement liability. Tangibility and its relationship to markets.  Fixation demonstrates intent to exploit the work in the market: either the market for a single copy or the market for copies.  Laura Heymann made this argument a while back: what kinds of works are worth protecting, can be engaged w/as commodities rather than experiences—by a person who is more consumer than audience. A work that is not a thing has difficulty becoming an article of commerce.  Infringement: sufficiently stable to interfere w/markets.
If fixation is notice, the audience for the notice that’s relevant in copyrightability is the author: you’ve now created a thing.  When it’s an encounter w/a fixed work (infringement), the authorship in that fixed work may be protected, the audience is others.  Says nothing about the scope of copyright, just about the potential/something in the contents of the fixed work is or may be copyrightable.  Copyright owner also gets notice that someone might have created something that interferes w/their market.  Both copyrightability and infringement have limits.
One should not take infringement fixation cases to talk about copyrightability fixation cases.  Even in the legis. history, Congress discusses White-Smith v. Apollo, an infringement case rather than copyrightability in its discussion of mistakes.  In copyrightability, we are ill served by a focus on durational permanence of minutes v. seconds.  Sand castles built below the high tide line; ice sculptures in hot places.  Authorial intent of creation of an exploitable copy should matter, which affects our analysis of bridal bouquets and tweets.  We should use our instincts to say that digital ephemera isn’t fixed enough—it’s a RAM copy, but still not eligible.
Fred Von Lohmann, Google: Notice to whom? In many of the papers there is an issue of audience. Loren suggests that fixation is notice to creators, notice to users, and perhaps notice to lawmakers/courts.  Garcia makes real the feature of fixation as focusing creators on a single moment in which we can determine who is the author.  At least that gives us a time to pay attention to—not the 3 months in the studio before that, or the idea a year before.
Unfixed infringements? Why do we think a derivative work can be infringing without fixation? Why do we allow unfixed public performances to be infringing? [Tony Reese has eloquently explained why you shouldn’t be able to combine derivative work + public performance right in this way, though unfixed public performances can be infringements of an original work.] 
Sufficiently stable to be exploited: does that work?  Has that ever worked?  Vine, Periscope, other new live streaming tech.  The organizers of a boxing match didn’t think they were organizing an event too evanescent to be exploited.  If fixation was intended to be a marker of that which had market value and that which didn’t, how do we deal with tech eroding the line.  Performances as things for which the door was adequate protection: you charged audiences to attend.  That’s not new—the rest of the fixation definition has the NFL clause, allowing © in broadcast of live events where there is a simultaneous fixation.
Other attack: nonpersistent copies: what do we do about copies that are brief, but create value—machine learning; Shazam app that can ID music—lots of reproductions, briefly, to extract metadata.  Have already seen legal friction along that frontier. What’s too evanescent to be exploited?  European approach in InfoSoc directive: introduce a concept of “no independent economic significance.”  Moved focus away from formal Q of fixation towards a more market-oriented definition. 
If we think fixation is so important, is there some limit on when it can be cast aside? This goes to the preemption question.  There is an exit opportunity for lawmakers: they can go to the Commerce Clause to evade the Copyright Clause; states can legislate on unfixed works.  What does it mean if lawmakers can treat fixation as optional?  Does that mean that fixation has to be part of a bargain that lawmakers make?
Peter Menell – Economic Analysis of Copyright Notice
In land, notice is a magic bullet: inform people in advance of rights and boundaries—a really good focal point for notice generally. Land is geographically unique, can be physically inspected and marked.  Recordation—reinforced by tax rules etc.  Controlled boundaries: you know what acts will interfere w/it.  You typically can preclear whatever problems you might encounter—real estate developers don’t put much at risk of expropriation b/c they get the zoning board to back this up.  Wasn’t always so perfect, but nearly so.  [Except for mortgage securitization and Ibanez problems which most of us have agreed to ignore, because banks must own something.]
Copyright and real estate are different b/c of building and borrowing as key feature of creativity. Notice deals w/discrete issues: (1) tracing—what governs a work you might interact w/; (2) scope—where are the things in a protectable work that you are free to use, such as ideas/tropes/facts?
Tracing: consider “Happy Birthday to You.” There isn’t a geographically unique resource but multiple copies; marking is optional and would be done at earlier time, with possibly unknown changes; recording isn’t required; even if it was still valid, there are unprotected regions within boundaries; no effective preclearance institutions such as opinion letter from Copyright Office.
Tracing: there is a magic bullet solution.  Today we have tech that is capable of doing this on a massive scale.  You can scan images that aren’t even that close and find them.  We now have a perfect magic bullet system in Content ID.  (!)  If we rewrote the Copyright Act today, we could use this system.  What you’re marking w/these technologies is the work itself, and there’s nothing preventing everyone from digitizing & enabling everyone else to find the work. Voids all problems of geographic/int’l disparity.  Unique global identifier: a concept that exists in computer science.  Failure to mark could limit remedies.
Digital ID safe harbor act: propose that © owners should be encouraged to make a digital deposit, in which works are archived into a searchable public archive.  Cumulative creators can search via contentID.  If no match, then no statutory damages available.

Scope would remain a problem; there is no magic bullet: the New Yorker New Yorker’s eye view of the world versus the Moscow on the Hudson poster.  Grateful Dead case in which reproducting posters is fair use when used for historical value and not expressive value.  When judge learned that Bill Graham Archives was being abusive in its demands, concluded it was fair use.  Why not have algorithms decide this too?
We should have up-front insurance.  Registry/quiet title, with preclearance institutions. Promote bargaining through licensing. One problem is that no one has to respond when a license is requested. A good citizen should have to respond.  If someone makes you a reasonable offer and you don’t respond—you should pay atty fees if you later sue.  PROs like ASCAP and ContentID are helping.  The solution doesn’t have to be perfect.  Compulsory licensing would also be good.
Substantive law: fair use—judges are better at saying “play nice” than trying to figure out the four factors. (RT: Who decides niceness? When it comes to whether someone who is not in a position of power has been “playing nicely” with a powerful person’s work or even just negotiating in good faith, I am reminded of Dale Spender’s observation: “The talkativeness of women has been gauged in comparison not with men but with silence. Women have not been judged on the grounds of whether they talk more than men, but of whether they talk more than silent women.”  Look at what happened to 2 Live Crew and Alice Randall in the lower courts, for example.)  
Remedies: Injunctive relief should be limited to exact copies—law is moving in that direction anyway.  Don’t shut down a billion dollar brand because it may have started w/two misappropriated names, as Kozinski said in Mattel v. MGA.  No disgorgement if someone has altered the work, v. piracy—just fair monetary. Statutory damages should also be limited to piracy.  (Is Bill Graham Archives an example of piracy? They were selling the posters, along with a package of other information?)
Wendy Gordon: much is plausible and convincing, esp. on tracing v. scope.  Intuitive reactions aren’t necessarily wrong. When you’re dealing w/ a question that can’t be solved with notice, like scope, then trying to make notice externalities less destructive by changing the remedies makes sense.  But digital watermarking doesn’t convince, b/c she suggests that would disqualify the unsearchable.  Orphan works problem: a lot of the time it’s the kind of ephemera that would never be watermarked now.  Buying into a larger question if you imagine making watermarking mandatory would solve notice problems.  Need for a new device to protect certain kinds of privacy (I was thinking about revenge porn as he was talking but Gordon has identified many other elements of the problem).
Transaction costs are meaningless in the abstract; must always be judged by the value of the transaction.  5 minutes to ask for permission may be significant if it gives you only a moment’s pleasure.  When you use real estate as an analogy, you’re dealing with something that has high value in many senses, including long term economic value. A dangerous model for ©, which serves many goals for people often producing things of little commercial value.  Recording requirements may be excellent for a land parcel, but not for the ephemera that make up orphan works or diaries.
Oren Bracha – Copyright Accidents (coauthored with Patrick Goold)
Judge Young said: It’s your duty and right as academics to pursue the ideas you think are right, even if they aren’t yet supported by judges and legislatures.
Copyright accidents: e.g., using an old photo in a documentary; using publicly licensed code “tainted” by proprietary code; including an “orphan” book in a digitization project. Should they be treated differently than any other infringement? 
All of them ex ante are risks, which only materialize ex post.  In other areas, we tend to treat such cases differently than cases involving more than ex ante risk, such as certainty of transgressing on some protected interest.  [Aren’t examples one and three examples of certainty of transgressing, with uncertainty about whether the owner of the interest cares in the slightest or even knows they are the owner?  As Henry Smith has said, you don’t need to know who owns the car you encounter on the street to know it’s not your car; but here the object you encounter is unlikely to be as valuable as a car.] Standard answer in ©: strict liability (though this hasn’t always been so). 
In tort cases, we often consider the ability to bargain/ability to solve the problem through markets instead of the law when parties impose risks on each other.  Market won’t take care of © problems because the boundaries of the rights are elusive; the legal status of the work is uncertain; the owner’s work is unknown and preferences uncertain; the owner’s identity may itself be unknown. These are issues of notice failure.  Bracketing in this paper the issue of legal uncertainty (is this fair use or not?).
The problem is one of reciprocal nature: Coase, The Problem of Social Cost.  If it’s a bilateral conduct issue, then look for cost effective prevention.  As would-be infringer: People can invest in search to find the owner, the copyright status, etc.  Or you could not create what you’re making, avoiding the risk entirely. But there are also many things © owners can do to decrease risks of accidents: registering ©, recording transfers, affixing notice.  Registering w/Content ID.  If user offers well-publicized opt-out, then use it.  Going after a third party if the third party is enabling lots of such uses. Strict liability gives optimal incentives to injurer to invest in prevention; but it doesn’t give any incentives to the victim/owner to invest in prevention.  Negligence gives them both optimal incentives.
What to do? If there is perfect Content ID, then we have perfect external means to solve the problem and don’t need to worry.  But he doubts this will happen any time soon.  A negligence element in the prima facie case?  Negligence proxy rules? Incorporate into fair use?
Fred Yen: Not all of these are really accidents. One can come across an orphan work, such as a digital photo (clearly taken after 1978)—hard to call going ahead and using it even after reasonable attempts to find the owner “accidental” infringement.  You don’t know if the owner will discover it/care, but not an accident. 
Legal accidents and negligence is a complicated topic, which Bracha doesn’t purport to address, but the line between fact and law is itself unclear: the D may think she’s only borrowing the facts but find out she’s taking protectable expression, or vice versa.  Sometimes these are treated as factual questions.
But it’s not always the case in tort cases that a defendant is allowed to rely on efforts of employee to avoid accidents—often true w/publishers dealing w/authors; publishers have staff and expertise.  Negligence might not let them evade liability.
Fault: there’s a reason we often start teaching torts w/intentional torts—there is an intuitive understanding.  It’s the later torts that get hard v. walking up and punching a person in the nose.  When we discuss the object of intent that can give rise to a battery, we ask what does the plaintiff reasonably find offensive: a punch in the nose, which everyone understand. What would be the analog in ©—full copying?  What about a not-well-known acquaintance giving me a hug—is that reasonably offensive?  Might be equivalent of copying a few pages for use in class.  Incidental bump on the subway?  Might be similar to ephemeral use.
There are various forms of intent courts sometimes accept as dividing line between faulty and not faulty: intentional touch (intent to copy p’s work); intent to touch in a way a reasonable person would find harmful or offensive, regardless of subjective intent (Nation’s publication of quotes from Ford’s memoir before the memoir was publication); intent to touch in a way D subjectively understands P will find offensive (piracy).  Maybe a publisher who genuinely believes there’s no unlicensed material used in a book simply lacks intent to touch/copy.  Time-shifting: maybe P thinks it’s offensive but we understand why individual people wouldn’t agree/understand.  Posting an entire short poem on a website to express your own thoughts—is that subjectively ok even if we understand why a reasonable person would be offended?
Privilege: in a torts model, you’d encounter public necessity and private necessity. These also come into our intuitions.  Transformative value/fair use.  Public necessity, we don’t ask you to pay; private necessity: you do have to pay for the damage you cause—Texaco or maybe even orphan works.
Gordon: what you’re describing wouldn’t be treated as negligence in the common law.  If I do a good faith search and make a factual determination about the boundaries of my land, if I’m wrong I’m still a trespasser.  I will have to disgorge profit I make.  There’s lots of historical questions about why a good faith belief in erroneous facts should trigger trespass liability, but that’s the pattern.
You assume strict liability means there will be no internalization of precautions, b/c P can always sue, but everything’s reciprocal. If a class of people isn’t liable, they might nonetheless be bribed by the other side (consumers willing to pay more for a safer product)—even after strict liability consumers may be willing to pay more for safety they won’t have to sue over, despite their legal entitlements.

Orly Lobel: Agency letter—safe haven if you tried but made missteps?  Computerized ways of dealing w/some issues—could we have tech of crowdsourcing to do the analysis? Ratings and reviews?
Menell: Perceived by many as property rights; I think of them more as regulatory and I’d like © to be seen as capable of admin adjustment and fine-tuning. DMCA anti-circumvention safe harbors is done in regulatory setting. B/c we never have perfect scope knowledge, we should have these other types of adjustments.  Crowdsourcing—trying to estimate what a jury would do after litigation; no more acceptable to © owners.  Agency might work better, but can’t be perfectly solved.
RT: Q for Loren: tell me why this distinction between fixation analyses depending on whether copyrightability or infringement is at issue is ok w/the statutory language.  And are either of them fit for use in preemption? 
Loren: other examples where terms turn out to mean different things in different context—willfulness for statutory damages/criminal liability; derivative works for protectability/infringement.  Doesn’t like that but might be something worth living with.
RT: Q for Menell: What is the basis for the claim that Content ID is perfect?  Matching problems in part b/c as we heard yesterday from Viacom’s GC they fear deliberate attempts to evade (not open and notorious use) so the algorithm is not written in the way you want what you are calling Content ID to be written.  This is evident already in the example you show: two different recordings being identified as the same despite visually apparent differences; they were clearly broadcast on different channels with different additional materials and it’s quite possible that the claimant only has a claim in the chyron despite using the rest of the image to create the digital fingerprint to identify what it is that they have a claim to; separately, there are ownership claiming problems, where the existence of “orphans” creates a massive squatter problem—like fake liens filed by right-wing citizens.
Menell: [the initial burden would be on the would-be user:] anyone who wants to be part of system would deposit stuff into the database. People could run scans against this. That would tell you whether any matches occur.  Doesn’t solve tracing problems.  At least if you’re Ken Burns you get some data.  Maybe we need separate protection for private materials. Anyone who wants to put materials out into the world has to register those works.  If you didn’t also go into a very low cost registration option, others would be able to use.  Clear out a lot of tracing problems.  (And create them when trolls register zillions of photos they find online that they think won’t be registered by anyone else … as Gordon says, this isn’t land.)
Von Lohmann: When I hear tort, I think insurance.  Many of these solutions come back to “fix statutory damages.”  If this is really a problem of statutory damages, aren’t these solutions only solutions in the margins?  The first-order problem is really remedies.
Bracha: still the Q of who should get insurance, the plaintiff or the defendant?  But yes, remedies are a huge part of the problem. 
Q: why not apply Bracha’s analysis to patent as well?
Bracha: there is some writing in this vein.  Trespass in patent might be different.
Gordon: though trespass does operate where the trespasser has done everything he can but was wrong.
Henry Smith: the temptation is to load more things into the recording system.  Liens, easements, etc. What about people who want to add in limited permissions.  Raises a contractual boilerplate problem—touch and concern is the policing device in land. But once you’ve got something for tracing people will want to load on all this additional info.  (Carol Rose's fabulous Crystals and Mud in Property Law talks about this overloading and the effects on notice/predictability.)
Menell: content ID is a bit like that—allows people to specify CC, etc.  (Hunh?  Allows you to specify whether YouTube will run ads, whether claimant will allow others to monetize their channels if they use Content ID’d chunks, but I am not aware it supports CC.) We’ve reached a tech stage where we’re capable of making the work itself the way of solving notice. No longer geographically/tied to a specific copy. Then you layer responsibilities onto it—e.g., you have to maintain contact info.  Would we want maintenance fees? 
Gordon: why do you think the magic bullet will work?  How do I determine whether a painting infringes another painting—it’s only the digital versions that can be traced and many infringements aren’t digital.
Menell: can do it for music; digitization is getting better.  Architectural works: architect will have to submit photo. But mostly people are suing over AV works and music.  But if we were sitting down w/technologists, we could say it’s not going to be that hard a problem.  (See xkcd on this issue.)  Easy to add photos, text, film—99% of what’s commercially significant.
Jane Ginsburg: Fixation performs the notice function of letting everyone know what the work is.  But you may have conflated the work with the fixation.  The fixation tells us what the work is.  But for the sand castle, ice sculpture: those are works. They happen to be incorporeal.  The tide washing away/sun melting doesn’t take them out of copyright b/c they’ve been fixed in a tangible medium of expression through photographs, which prove they’re perceptible for more than transitory duration. Once we know that the work exists, that’s enough even if the original instantiation disappears.
Loren: sure, photo can fix. But what’s a transitory duration?  Courts are saying that perception isn’t the only requirement. There’s a durational component.

No comments:

Post a Comment