Margaret Jane Radin – Patent Notice and the Trouble with Plain Meaning
Notice is a cost and a cost-saver; hard to figure out its overall profile. Plain meaning as ever-receding ideal.
Philips v. AWH (Fed. Cir. 2005); Festo (2002). In Philips, Fed. Cir. laid out interpretive procedure agreed on by all but 2 judges to focus on entire available text, including prosecution history. Only useful from notice POV if they generate more predictable outcomes—but they still didn’t get the same answer when they applied the agreed-on interpretive approach to the facts before them. Not clear that the interpretive models affect outcomes, compared to priors. If context really does matter, then intuition is not necessarily a huge problem. Maybe plain meaning is really closet holism.
Festo: Introduced the concept of describability. Doctrine of equivalents and file wrapper estoppel. At least estopped from claiming something described and disallowed during prosecution. But broader implications: it’s a crucial issue whether an equivalent could’ve been described in language when the claim was drafted, and this can occur even w/o claims being amended. Describability can change over time; if it was describable at the time of prosecution then you can’t get it, and if it’s only describable now then maybe you can. Describable by whom? Great unanswered Q—we don’t want to think of patent prosecutors as co-inventors, but it’s a fact.
Other patent puzzles: if fully describable at the language of the time, is it then obvious?
Inventions are emerging knowledge, and thus plain meaning will be contestable, lacking the clarity its advocates hope for it. The word describable is new in Festo but the older cases use the concept—the emergent nature, esp. of true breakthroughs—language only catches up. Thus the idea of the essential nature of an invention survives, even though it’s an echo of central claiming that died long ago and even though it’s anathema to plain meaning advocates.
Suzanne Michel, Sr. Patent Counsel, Google
Wasteful litigation boom. Lack of clear thinking about claim construction is an important reason for lack of clear patent boundaries. True, fencelike boundaries can’t be achieved when describing something that’s new. But it’s still a useful analogy/talking point/goal even if we can never get to that shining beacon—can still consider whether we’re improving or not improving. Points out the shortcomings of the term “IP,” a boon for patent maximalists.
Judges make decisions for reasons they’re not fully examining in claim construction. Unpacking methods will improve decisions. One frustration is the pablum that judges just apply the law, not make policy—that isn’t and can’t be true. Plain meaning judges will often limit claims only if there’s a clear disavowal, which means that Philips didn’t change what it was supposed to.
Would focus on the context of the patent document. Often in litigation the words the case turns on is not one that has meaning in the art—“sandwich” in an electronics patent; so you have to lean on the specification. Also relates to proportionality/patent bargaining: right should be commensurate with what you have given to the public, what is described in the specification.
In some cases interpreting a patent narrowly to sustain it is a better idea than broadly then arguing over invalidity. When a patent is interpreted broadly and the only defense remaining is invalidity, people are coerced into settling. Disclaimer approach (only limiting the scope when there’s a clear disclaimer) penalizes patentees who clearly describe their inventions. People who have a patent for a fax machine then claim the internet infringes are a huge problem.
Keith Hylton – Patent Uncertainty: Toward a Framework with Applications
Uncertainty has dynamic and static elements (uncertainty in determining boundaries of property right). Sources of static uncertainty include: (1) statistical artifacts from the sample of cases you look at. (2) Inherent uncertainty, as Radin discussed. (3) Strategic uncertainty: people gaming the system to use it to their advantage. Static/dynamic balance is why increasing scope doesn’t always increase innovation: optimal scope (which applies to all property rights).
Statistical artifact uncertainty: Oversampling, as you oversample the bad drivers you encounter on the road. Explosion of litigation costs in patent—timed to software patents. Alice produced a sharp drop in patent litigation—at least 1/3 and maybe more. Recent uptick in 2015, but not enough to take it back up to previous levels. Suggests that knock-on effect of Alice was reducing this observation of uncertainty.
Another feature of software patents: asymmetric information probably less of a problem than with things like medical malpractice.
Inherent uncertainty: Courts start from maximal uncertainty about a case and try to work towards an optimal rule—this uncertainty is unavoidable.
Strategic uncertainty: deliberate use of vagueness to increase economic value of patent/expand its boundaries. Four categories of uncertainty w/in strategic uncertainty. First, inherent level of abstraction in the patent + notice. Are you truly trying to tell people about the boundaries of this patent? Abstraction: what applications, known and unknown, does this patent cover? Abstraction can coexist with notice—they are separate in that I could patent the quadratic formula and then there’d be notice but lots of abstract uses. Bilski: people know what hedging in the energy markets is and they know what a computer is so they know when they’ve implicated the Bilski patent. Abstract w/o notice is the most troubling quadrant (remaining: not abstract + notice; not abstract + no notice).
Algorithmic patents should just be unavailable/ineligible. But other software patents are worth worrying about. Rowe patent: detects driver impairment. We want that innovation incentivized in that space, so Alice shouldn’t be quick to kick out those kinds of innovations.
Benjamin Roin: Policy levers—blunt tools are hard to think about b/c there’s so much going on at the theoretical and empirical levels. Some qualms about use of terms static/dynamic—in some sense static uncertainty about what a claim covers is always dynamic b/c you don’t know what tech will arise. It’s clear that abstraction and notice aren’t the same thing. Roin suggests: also considering patents as necessary incentives for underlying invention—that’s actually distinct from notice and abstraction. Probably relates to R&D costs, market, uncertainty, cost of capital, cost of imitation. If we want to get rid of chunks of patents, incentives should be part of that analysis. Driver safety software: significant premakret safety testing, long time to market, need for licensing to big auto, need to create the market—they may or may not need a patent.
Benefits of uncertainty: Incentive benefits of patents are highest when blocking copycat imitation and “fast followers.” Practitioner literature emphasizes difficulty foreseeing & drafting claims to block all design arounds. “Ease of designing around” is most common reason patents are seen as ineffective in industry surveys. Rules v. standards: vague(r) claim language helps or could help block mere design-arounds. In tax rules, precision makes evasion easier. Dissidents in USSR wanted clear standards so they could avoid censorship; censors refused so they could punish them after the fact.
Notice to whom? Potential infringers w/advance notice are more likely to avoid liability by designing around the claims. Copycats and fast follwers are probably most likely to have notice of the patent. Are we getting the worst of both worlds? Maybe we give sufficient notice & certainty to allow them to design around, undercutting main beneficial incentive effects; while giving insufficient notice and certainty to more substantive follow-on innovators (particularly ones in complex downstream product markets and ones in more distant tech/market spaces), exacerbating the system’s main adverse incentive effects.
Stan Liebowitz – The Benefits of Unbundling Copyright from Patent in Economic Analyses
© and patent are often lumped together in discussions, including discussions about notice, but they differ in important ways, leading to analytical errors when they are lumped together. [Liebowitz thinks that it is easy to tell when you have infringed copyright because, except in unconscious copying cases, you know when you’ve copied and thus it’s easy to avoid, unlike patent infringement which does not require copying. This entirely ignores idea/expression difficulties and his own discussion of genre and how no copyright owner can own an entire genre even if she creates it. By definition, creators who participate in the genre know of the existence of the founder, and their question is how close they can go.]
Notice costs are resources used to impart ownership information. The idea that IP/intangibles have higher notice costs than tangible goods. This may be true for patented goods; less likely to be true for copyrighted goods.
Simplifications: creators of new works try to avoid infringing previously created works or wish to get permissions. This is the forward-looking view. I will assume © lasts forever with no limitations on ownership rights, which eliminates the need to examine the costs of dealing with restricted ownership, such as fair use or compulsory licensing. [So, assuming a can opener, © would be different from patent? And yet it’s still not true b/c of idea/expression.] If you don’t copy from someone else, it’s ok. If you never saw the first work, it doesn’t matter if the second looks like it. [The other important thing he has assumed away is any difference between “never saw” and “never copied,” which is idea/expression. And yet successful works are seen and heard by millions, because that’s how they have value.] It’s perfectly fine for the Beatles to write “Back in the USSR” to josh about the Beach Boys’ “California Girls” b/c all they took is the idea, so they don’t have to worry about notice costs. There are no notice costs as long as people are only creating new stuff. This is possible in © but not in patent or real property. [It’s distressing to me that he doesn’t notice the contradiction inherent in his story of the Beatles creating as a but-for consequence of the Beach Boys.]
Of course, this isn’t the real world. There are copiers out there, and that produces notice costs for permission, evasion, or application of exceptions. Costs are higher b/c copiers want to avoid payment by checking on whether works are in © (or limits apply)—duration imposes extra notice costs. Fair use imposes extra costs, but these are not notice costs b/c neither the owner nor the copier can tell another whether a use is fair use. That’s a cost, but he doesn’t want to call it a notice cost b/c it comes from the legal system and not from a claimant thereunder. We have decided to bear the costs of having fair use.
Bottom line: copyright has lower notice costs than other properties for works created de novo, at least when people know when they’re copying. Guesses that majority of works that have market value are works of this type. Thus, patent notice costs are likely to be higher.
Wendy Gordon: Key point: he reminds us that if we’re not copying, we don’t need to worry about copyright, in the abstract. In the concrete, given the ubiquity of works, you can be forced to prove you didn’t copy b/c access will be possible. The practical issues arise, though perhaps not to the extent as in patent. Favorite example: Spider Robinson’s sf story, Melancholy Elephants—if we knew our borrowings all the time we’d always infringe; we need the ability to forget.
Notion, adopted as a matter of mere convenience, of inherent copyright (that lasts forever with no holes)—it’s inherently contradictory. If a copyright were completely owned, then its ideas would also be owned. A classic ambiguity—and you’re trying to rule out ambiguity—is idea/expression. We also recognize indirect liability: if I copy Stacey & she copied Fred, Stacey’s grant of permission to me doesn’t help my violation of Fred’s rights.
If your hypothesis is that everything the creator did was new, you have the paradox of human creation—language is not new. You are always building on the foundations built by others. There is no such thing as inherent copyright.
Patent too often takes for granted that independent inventors, if they come/file later, are guilty of infringement. This gives rise to trolling/submarine problems. Arguing that patents shouldn’t have a defense for independent inventions because that would be inefficient requires some heroic assumptions. She’s never been persuaded that simultaneous invention justifies the absence of an independent inventor defense; also seems inconsistent w/ human rights entitlement to the fruits of one’s own labor. Winner-take-all seems unjustified.
Lemley: for Radin: Agrees w/approach and suggestion that plain meaning won’t bring clarity. Take your argument to its logical conclusion, and it’s not obvious that adding more words on top of the claim gets us any further. Claim construction = lawyers fight about meaning of claim construction. 19th century model: would we be better off w/central claiming and what the patentee actually built, not what the lawyer wrote?
Radin: sympathetic to that—we do that sub rosa. Causes judges sometimes to construe “plain meaning” broadly. Thinks we always have both.
Menell: pragmatic issues around Markman hearings are so strong—dealing with judges not trained in these areas—you need assistance finding the essence of an invention. © also needs a Markman process trying to delineate what the uncopyrightable elements of the work are for the jury. That’s an important part of the translation.
For Roin: suggests benefits of uncertainty, but you can’t have a public system of property rights w/o a serious public disclosure of boundaries. Left w/trade secret if you don’t want that.
Roin: It’s a rule v. standard thing—ex ante v. ex post. (So I take it that central claiming could address some of his issues?)
Q: when you have a patent, you have multiple claims, not just one claim. The life of a patent isn’t over during the span it’s in effect. You can cross-license with an improver. You can use continuations to extend its life.
Radin for Michel: Michel suggested that boundaries were a good analogy/aspiration, but Radin thinks it can be a misleading one. We still have lots of agreement about what judges are doing.
Chiang: You can’t capture the policy goal of complete definition with language. But if you want a rule w/ a result that’s clear one way or another in a particular case, then you can provide notice/plain meaning—everyone can predict the result ahead of time. Radin claims that the fact that judges purport to apply plain meaning doesn’t show such a thing exists, and he agrees. But the fact that they disagree on what the plain meaning is also doesn’t show there is no such thing as plain meaning. You have to add in dishonesty [Radin clarifies: not in the sense of lying, but in the sense of having intuitions that are not fully explained].
Radin: we can get many easy cases, but when there’s a lot of money that matters we won’t get help from plain meaning.
Chiang: but as a philosophical claim that plain meaning doesn’t exist, it’s a problem that your claim only covers hard cases, not easy cases.
Radin: sure there is plain meaning when we all agree, but that’s not b/c of inherent meaning but rather b/c we all agree.
Henry Smith: can agree that plain meaning doesn’t exist in practice/hard cases but still not think it necessarily matters. You could think language is indeterminate, and nonetheless formalism/contextualism depends on the nature of the system. We have different varieties of English, some more and less formal. There’s a dividing line about what we should care about. The plain meaning folks are saying “in this context, we need only so much information, not more, to do what we want done”—that’s a pragmatic judgment. Philosophy of ideal points doesn’t tell us much about that.