IPSC Breakout Session V

IP Theory, Functionality & Design

Infringing Algorithms

Felix Wu

Means plus function claims: you get the structure disclosed in the specification and equivalents thereof. Why doesn’t this limit on functional claiming work?  Claims that are at root functional are not construed to be functional; Fed. Cir. decision in Williamson: need to know when we are going to interpret these claims as functional.  Now: magic words “means” are not required. But still need to know what will count as the corresponding structure, and there’s something special going on with software b/c we can see structure w/mechanical arts.  What would we mean by structure when software is effectively functional all the way down: overall function, broken down into modules that perform functions, broken down further etc.  Fed. Cir. says the structure in software is an algorithm, which then imposes a limit on functional claiming. But then: what does Fed. Cir. mean by an algorithm?  Necessary to perform the claimed function.  But this doesn’t help, b/c then we need to know what’s necessary.  Missing the idea that you can break up one function into a series of functions a number of times—the levels of abstraction problem that we’re used to in © but not in patent.

Fed. Cir. is stuck about where we were in Jaslow in ©. Fed. Cir. says algorithm is what’s necessary to perform the claimed function and nothing more, and that’s what the court did in separating idea/expression in Jaslow.  Lesson: we need abstraction-filtration-comparison for means plus function claims.  Avoiding overbreadth.  Alice and the like may provide tools to do that kind of filtration, better than using them for patentable subject matter.  SCt’s instinct in Alice about overbreadth is really about filtering.  Altai also filters out the non-novelty, as we want to do here. If algorithms are equivalent only at the level of what’s not novel, we shouldn’t regard one as infringing the other. 

Should we also use this instead of nonliteral copyright infringement?  Patent supremacy, as discussed by Lemley & McKenna?  That might make sense.

[Sorry, when it comes to patent I can only hum a few bars.  Lemley’s question seemed perfectly reasonable but I didn’t process it.]

Samuelson: Fed. Cir. Oracle v. Google: they want to say if there’s any other way to do something then it’s copyrightable.  But software usually has more than one way to do something; it doesn’t make those things non-abstract. If it’s too abstract for patent, it’s too abstract for ©, and it doesn’t make any difference that there’s more than one way to do it.

A: yeah, that seems reasonable.

Q: there’s an expressive element to software that differs from its functionality. 100 programmers would write 100 pieces of code that were different for the same function—it’s that aspect that © protects, and it’s the time it takes to code that © protects. I like your idea on the patent side b/c the same kind of filtering is needed as on the © system.  But not sure there should be lessons for © that only strictly literal copying should be actionable, b/c code is expressive.

A: not clear why expression is the source of value of the software. Depends on what you mean by expressive.  The expression is limited in that it produces the function, even if you feel like the code is beautiful. 

Q: what makes a novel beautiful also doesn’t matter.

A: but beauty in code is also about accomplishing purpose efficiently and laying out text efficiently, which matters to readability and reusability.  Both of those things are functions.

Q: but pro programmers can pick up code and tell you who wrote it.  [Hmm. Analysts in WWII could also listen and figure out the “fist” of the specific telegraph operator who was on duty.  That didn’t mean the telegraph operator was engaging in copyrightable expression.  We still have to figure out the question “what should copyright cover?” and the answer is not “all things that are different depending on who produces them”—if you ask me to drive a route, the details of how I press on the gas and where I signal will differ from the details of how other people do it, but that doesn’t make my driving pattern copyrightable.  Cf. Abraham Drassinower’s recent book.]

Functional Compilations

Pamela Samuelson

CONTU said utility is never a limit on ©ability, which is totally untrue; Easterbrook in ADA case says functionality is only a limit for PGS works.  Architecture and software are examples of functional works Congress decided to use © to protect. But at least w/architecture we know that functional parts of designs, e.g. plumbing and wiring, are unlikely to be within © scope. Altai is the leading case indicating functionality limits scope of © in software, though it didn’t direct filtering out of processes, methods, etc.  There is can be a merger of function and expression in software cases, usually as to particular elements of programs, not programs as a whole. But Lexmark is an example of merger resulting in invalidating ©.  Argument: When courts use selection, coordination, and arrangement, that doesn’t mean that compilation is protectable if it is a functional compilation.

Doctrinal buckets for courts imposing functionality limits: but since they were struggling in these cases they used multiple buckets: most common: if functional, lacks originality.  Or functional b/c implements method, system, procedure unprotectable under §102(b). But also invoke multiple doctrines, sometimes as many as 5—doctrinal cocktail. [Sounds like TM fair use cases.]  No protection for facts, etc.

Functionality types: (1) mechanically derived (Feist, Continental Micro). (2) Dictated by function, even if there’s more than one way to do it—recipes, involving yogurt. There’s more than one way to do that. (3) Result of systematic or methodical organization: directory of CATV systems, not a protectable compilation.  (4) Necessary to accomplishing objective or task.  Bank claimed © in its wire transfer numbers, which are required to transfer money from bank to bank.  (5) Conformance to rules, logic. Southco part number case—systematic; dictated by conformance to logic of numbering system and thus unprotectable.  (6) Efficient design. (7) Incidental to carrying out tasks/processes. (8) Industry standard. (9) Affecting cost, quality, or effectiveness of product design (Traffix).

Suppose Rural had been the first compiler of telephone white pages in what’s now the standard order, using skill and judgment to organize—that would still be functional.

CO refused to register human DNA sequence—too functional even though it’s a literary work.  Continental Micro: © in compilation of data on shape of and depth of keys as inputs into key cutting machine processes. May be difficult and time consuming but not ©able.  Baker v. Selden: arrangement of columns into functional compilation.  In many cases, functionality limited scope of ©.

We need more work on what makes a compilation expressive?  100 best restaurants.  ADA case is flatly wrong; numbers are functional for billing people for dental procedures. One of the bad things is it begat other cases where P claimed to taxonomize, not systematize, but a taxonomy is a system of organization. McLean Hunter similarly begat lots of bad cases.  NYNEX’s claim to © in settlement prices on its exchange—that would mean all prices were ©able b/c someone engaged in human judgment to set them.  Leval has been willing to confess error at least about the prices themselves, but the 2d Circuit has not completely repudiated it.

Q: do you have a view on copyrightability versus scope/narrowing determinations?

A: I talk about that issue in the paper—courts have to be willing to say some things are just not copyrightable at all for functionality. Some of the narrow scope cases involved instructions about how to use a noncopyrighted product; if there was some variation possible, the court would allow nonexact copying.  Rulebooks: often different ways to explain rules. Even if MacLean Hunter was rightly decided, it should be narrow. We should spend more time asking not just what’s original but what’s expressive. If result is thin protection for some that’s cool.

McKenna: reason for excluding from © matters: if it’s b/c they aren’t original, existence of alternatives wouldn’t matter so much. If it’s b/c they go too close to other areas like patent, then alternatives might matter.

Sandeen: did you look at whether © was registered timely, or only in order to sue against something they didn’t like?

A: I could go back and look; my sense is that generally it’s when the other guy does something that’s too close that they register and sue. No case I found involved a refused registration for compilations though they would refuse registration for forms.

Sandeen: trade secret might have been an alternative for some.

Screening Functionality in Intellectual Property Law

Christopher Buccafusco & Mark Lemley

Identify and describe three distinct functionality screens in ©, TM, and design patent. Evaluate costs and benefits of each screens and assess whether the screens are properly chosen for their fields.  Simplified model of design decisionmaking: all works can be plotted as percentage of aesthetic or functional features.  We don’t care what functional means, and use aesthetic shorthand for nonfunctional (expressive, ornamental, source-signifying).  Dual-purpose features are our problem.

1: Filtering.  In ideal world of filtering, all aesthetic stuff in and all functional stuff out. As long as you can show any aesthetic value.

2: Exclusion screens: no rights, no matter what.

3: Threshold: dichotomy—things to one side get filtered, things w/o enough aesthetic content get zero. 

© has all of these three.  Filtering, you know.  Copyright excludes cuisine, recipes, pre-1990 architecture, yoga.  Threshold: PGS works.  Not useful articles (filtering), useful articles w/separable features (threshold), useful articles w/o separable features (exclusion).  Tries to capture the issue of dual-nature features, applied art—where there’s intrusion of the functional into the aesthetic.

Design patent is the opposite: no screening, you get almost everything unless it’s totally functional.

We’ll look at administrative costs, error costs of false positives (incentive costs, maybe) and false negatives (competition costs). Filtering has low admin costs on entry but lots of stuff gets in; high false negatives if there are systematic errors.  Exclusion: low admin/litigation costs after initial exclusion, high risk of false positivves. Threshold: high ex ante costs.

Copyright: high risk of competition costs may undermine filtering here. Perhaps saved by merger doctrine; that’s what CONTU seemed to think, that we needed some kinds of rights for incentives to exist. PGS works: thresholds make some sense, at least to the extent that Congress decided to lump car parts w/paintings.  Exclusion makes sense where we think there are no substantial incentive costs.

Trade dress: serious competition risks if decisionmakers wrong; exclusion might be better.

Design patent: failing to screen out dual nature features of design means risk to competition. We’d have to think there are no serious competition risks and that there are serious risks to incentives from underprotection to make this sensible.

Linford: given your focus on competition, why doesn’t the existence of alternatives always lead to nonfunctionality finding.

Lemley: Trade dress does better than design patent in having abandoned the idea that alternatives = nonfunctionality. Affect on cost/quality = subject to functionality screen.  Design patent has a miniscule definition of functionality: only if there’s no other way of doing that, at best—there must be some design protection for every element, if you read cases broadly.  We don’t want to accept alternatives as precluding functionality.

[unscheduled break for me]

Claiming Design

Jeanne Fromer & Mark McKenna

Claiming looks very different across regimes. One important aspect of design patent is that it happens early on, often ex ante, before commercialization.  Another important aspect: mode is visual; law discourages use of words to describe design.  You draw more than what you’re claiming, with broken lines to indicate what’s not claimed.  Trickery aspect enhanced by visual format.  Infringement test is more of central claiming by exemplar b/c infringement test asks whether ordinary observer would be deceived into purchasing one design supposing it to be the other.

Trademark has 2D claiming; you have to figure out the scope of someone’s rights along both dimensions at the same time. PTO has elaborate rules for mark depiction and description of goods and services. But you don’t need a registration for TM rights and none of the rules apply if you claim the design as an unregistered mark, which is most litigated cases. Those elaborate rules also don’t matter once you start litigating your registered mark, and courts evaluate the nature of your use.  Allows you to make ex post judgments about what D is doing and shape your claim w/defendant in mind.  If you’d been forced to delineate Two Pesos trade dress up front, you almost certainly would have identified colors, but that’s not what D copied. Central claiming by exemplar, but the exemplars aren’t very good.  PTO registration of TM in iPhone screen: long and specific, detailed about color.  In litigation, description is broader and shorter. What courts have noticed, rightly, is that TM claiming in litigation creates significant risk of moving target.  Sometimes tried to impose a requirement that you state upfront for purpose of litigation what your trade dress is; many courts make you plead this. But there’s no constistency across courts about level of generality required.

Internal consequences of claiming rules: broken line phenomenon leads to overclaiming.  External consequences of claiming rules: overlapping regimes; compounds strategic aspect.  Illusion of precision and notice, but trade dress can pop up later on as in Taco Cabana: don’t claim color in litigation.  You can say TM is about protecting source identifiers so colors shouldn’t matter if consumers are confused, but it undermines notice.

Lemley: I always understood that dashed lines had to be there but weren’t sufficient. 

McKenna: not clear. You could make the preamble have operative effect: design of a clothing hanger, even if the entire outline is dashed and the only claim is to the color of a clothing hanger.  It’s not true that the dashed features have to be present.  There’s no all elements rule.

Fromer: In UK, broken lines have meaning; US law is less strict.

McKenna: maybe people would file more design patents to make sure they’d have rights down the road; that’s a cost. TM allows you to identify only what’s proved valuable.

Rosenblatt: utility patents—the moving target will depend on what the competitor is doing, but without prior art to keep the patentee honest. You want a very broad claim in design patent that you might not want in utility patent. On the other side: secondary meaning may develop for things you don’t really know are going to be your mark. The opportunity for a mark to emerge for the shape of an awning rather than the color is very different.

McKenna: true, incentives differ. Fromer: we might want to give people incentives to think about that up front.  A lot of doctrines in utility patent are about forcing people to develop inventions far enough to think through the ramifications.  McKenna: functionality is the constraint on broad claiming in TM; but the more broadly you claim in design patent, the less likely it is to be functional.  [Not clear if McKenna is defining breadth in the same way in both contexts.]

Q: consider tacking in relation to scope.

McKenna: right now you can use design patent to claim priority/secondary meaning in TM b/c of 14 years of exclusive use, which you could not do for utility patent.

Copyright as Tortious Interference

Shyam Balganesh

Focus on looking at common law has been principally structural: analytical basis of IP, not necessarily the normative basis. Normative reasons for common law doctrines are not a great fit for IP—deterrence, corrective justice, cost avoidance.  We should start thinking about common law not just as analogy but as homology: distinction by Abraham Drassinower.  Analogy identifies a resemblance and makes a claim on the basis of that identification.  Homology identifies a similarity but in addition makes an assertion about the basis of the similarity.  The common origin can be anything: single source, evolution, motivating normative ideal. In law, homology makes assertion about shared normative goals to explain or justify the similarity.

Tortious interference w/contract: origins in Lumley v. Gye, expanding and absorbed into US common law by 1900.  Breaks off into tortious interference w/prospective advantage—a standalone action. Instead of looking at existence of a valid contract, a prospective economic advantage or lost chance is sufficient to allow liability to kick in. Better known in the US than in other countries.  Exists in most jurisdictions in the US; NY and Cal. have best-developed; different names in different states, but all the same. Identifies a market prospect that is protectable b/c reasonably likely to be realized and creates an exclusionary liability regime. Only certain actions by a D trigger liability.  Standard: knowledge of prospect and intended interference; prospect reasonably likely and not merely speculative; improper/wrongful interference; ensuing economic harm. Intent = volitional act, rather than other mens rea.

Most litigation focuses on identifying reasonably likely prospect and improper/wrongful interference.  The prospect and the wrong.  Prospect: not a mere hope or subjective expectancy. Courts want reasonable likelihood of realization, based on market patterns and nature of activity in particular domain. There’s an inverse correlation b/t the level of probability the court requires and the public policy goal underlying the need for protection. Even probabalistic recovery can justify liability if there’s a perceived need to create incentive for P to invest in this activity.  Not purely an epistemic probability based determination; supplemented by public policy/social welfare goals where incentive to invest in economic activity justifies protection.

The wrong: must be wrongful means or motive; unethical behavior based on commercial norms; anticompetitive conduct.

The defense: privilege of interference.  (See similarity: not all copying is wrong.)

Very few normative rationales offered for this tort; three most commonly offered are (1) free riding and the ex ante incentive, BeVier 1990. There are many domains where info investment is needed; allowing third party to free ride when it can obtain the info on its own can justify allowing the action, in order to preserve incentive to invest in info gathering. (2) Commercial morality, as w/trade secret; completely nebulous. (3) Ownership of prospect as a property right, Epstein.

If one adopts tortious interference as homology w/copyright, helps us recast analytical structure of ©. What’s the object of protection? © theory oscillates b/t the work and the action (Drassinower).  Tortious interference says it’s a prospective economic advantage that should be allocated to P.  Not every form of interference is actionable; copying is a normative identification of elements that make the action improper.  Intentionality: need volitional conduct.  Forms of exempted copying = privileged interference.

Looking at homology shows (1) ©’s logic of incentive creation is fundamentally flawed. The incentive doesn’t come from ©; that doesn’t underlie tortious interference either. Incentives come from the market. The rights are circumscribing the existence of that incentive when the market has independently created it. © doesn’t supply the incentive.  It protects an incentive that exists, when it exists.  (2) Idea of liability—when is copying wrongful—should be our focus.  Substantial similarity is a black box; we should get a handle on normative question of what should be allocated to individual claimant.  (3) Copyright recovery should be understood as probabalistic.  Causal indeterminacy is central.  Can be factored into damages computation. But isn’t right now.

McKenna: Tortious interference is maddening b/c law is completely unclear about what kinds of conduct count as wrongful interferences. I get the homology, but are you importing more uncertainty than you’re gaining?

A: black box = jury question. I am trying to give a framework for judicial decisions. I also don’t want to transpose the uncertainty. But there are patterns in the case law—3 distinctive categories of improper conduct: (1) independently actionable unlawful conduct; (2) acts in the shadow of unlawful conduct; (3) lawful but unethical conduct.  But I love living w/uncertainty.

Madison: remedies: Under Cal. law, it’s an intentional tort allowing punitive damages. W/full mapping, you’re borrowing a big problem at the remedial level.

A: I’m not seeking to apply all aspects.

Q: if market is where the advantage comes from, would we need to investigate whether the market existed case by case?

A: Absolutely. It’s not done now.