Saturday, March 31, 2007

What Ifs? Patents

Patent Law

Timothy R. Holbrook, Chicago-Kent College of Law, What if patents and patent applications weren't published until they expired?

We think of patent disclosure as a fairness obligation. But getting claims of infringement out of the blue isn’t that surprising even in a publication system – so what function is really served by publication? It’s supposed to teach how to make or perform the subject of the patent. But there are a lot of incentives not to look at patents, or even applications, such as the possibility of willful infringement.

Also, the information disclosed by a patent application may not be very good – 18 months may be a very long time in a tech-intensive industry. Moreover, the content may not really help you out. Even if you do read the patent, you can’t do much with it during the patent term, given the lack of a robust experimental use exception.

The patent might not afford real public notice of its existence, but once you see it you may have an idea of the scope of the claim. But that assumes awareness of the patent.

The alternative would be more like a trade secret regime. Independent invention is a defense against trade secret, of course, so you may have more notice of what you can’t do under a trade secret regime. And in a world without publication, the assertion of a patent claim would put others on notice – you’d lose some of your trade secret-like protection by asserting infringement.

What about the available defenses? Other unexpired patents wouldn’t be available as recent prior art. But patents may not be the best source of prior art, especially with new tech or new subject matter – people complain that the PTO is too patent-dependent. But patents are a network, and you can look at how they cite each other. The use of the patent system as an index to find other patents is an oft-overlooked benefit of publication.

Insight: need to create incentives for applicants to include greater non-patent-based prior art, to improve the utility of this patent network.

It’s not as if the PTO does a great job right now. So what about hidden patents? There does seem to be some public oversight, e.g. when the peanut butter and jelly sandwich patent issued. Delayed transparency would be a harm. Thus, disclosure isn’t necessarily serving a fairness or quid pro quo objective, but actually helping to improve initial patent quality.

Michael J. Meurer, Boston University School of Law, What if economists ran the patent office?

What do economists think of patents as property? The majority of industrial organization economists today tend to think of the patent system as regulatory and closely connected to competition policy, rather than as property.

The FCC has succeeded by using economists to run sophisticated and successful auctions of spectrum. The economists pushed for an auction and thought very carefully about how to pick the right kind of auction given the circumstances (bidders’ valuations are likely to be correlated, which has complicated effects; collusion is possible). So maybe there’s room for improvement in patent.

The methodology is mechanism design: the state of the art in microeconomics. You structure rules of the game to elicit private information held by the people subject to regulation. Lemley and others have looked for mechanisms to add to the patent system, such as gold-plated and regular patents. That would elicit information from applicants about commercial appeal of a specific technology and the robustness of the overall field.

An economist would also want to experiment with – drastically increase – renewal fees.

The PTO has been hostile to academic economists, unlike other agencies, because it understands that economists are a threat to its way of life. The PTO’s customer service orientation helps applicants, but is the wrong structure overall. The examiner or the applicant needs to parse claim language. We need to divide the examiner’s time between different statutory issues, possibly differing depending on the technology. We need to consider whether to require a second pair of eyes in certain circumstances. Economists could help figure out what incentives to give examiners to spend their time most efficiently.

Joseph Scott Miller, Lewis and Clark Law School, What if Joe Meigs had written the nonobviousness statute?

Subtitle: How the perfect murdered the good in 1948. The issue is error detection – the PTO can make false positive and false negative errors. Miller’s gut feeling is that false positives are worse (in patent) than false negatives. Section 103(a) is a very thin operationalization of nonobviousness – it includes a timeframe (at the time the invention was made) and an audience (PHOSITA), and says that dumb luck is okay as a source of invention, but that’s about it.

In 1909, the Supreme Court says that, if skilled people have failed after repeated effots to discover a new and useful improvement, then the person who makes the discovery is entitled to protection as an inventor. This is a concrete and recognizable way of surpassing ordinary skill in the art. Judge Hand points out that we shouldn’t rely on novices, who are dumb in a dangerous way – a novice may find nothing but the obvious in a combination that hasn’t been discovered for a decade after the need arose. The best test of what persons of ordinary skill can do is what they have done. Know your history in the technology and you can evaluate obviousness! But the Supreme Court overturned Hand’s decision upholding a patent for a cordless automatic lighter for a car.

Rejected heresies: the requirement of a flash of genius. The synergism requirement – illustrated by a case involving the modern checkout counter – space for one customer to unload while the previous customer was being checked out. The 6th Circuit said that a long-existing problem born in 1917 was never solved until 1937, thus justifying patent protection, but the Supreme Court reversed and struck down the patent because “two and two have been added together and still they make only four.” The patent bar demanded that the standard be fixed to eliminate both the flash of genius and the synergism.

Giles Rich felt that a positive statement about patentability was dangerous as possibly restriction; though he didn’t want judges to indulge personal standards, he wanted the ability to make patentability arguments to help his clients. Likewise, another drafter said that § 103 had no standard; it was a requirement with no implementing guidelines.

Joseph Meigs was a chemist who died in 1959, the named inventor on 22 patents, 18 as sole inventor. He was also a patent lawyer. Meigs proposed a patent statute that made nonobviousness a question of fact, and that said that sufficient proof of nonobviousness could come from (a) filling a long-felt want, and (b) prior to the invention, the skill of the hard had not supplied, and could not supply, that want. It’s a thick description, but it’s not exclusive.

Meigs was not politically connected; he testified on his own behalf. The biggest question: at what level of granularity do you need to state the “want” at issue? Giles Rich testified against the bill as too rigid, and might incline courts away from sustaining patents where there was no long-felt want. This is absolutely true: it would make more false negatives than §103. But it would make fewer false positives.

In practice Meigs’ standard is useful for litigators – if they can show trial and failure, they are likely to succeed in showing nonobviousness. Putting it in the statute would signal better to inventors where they should work to get protection. It would also encourage them to defy conventional wisdom in the art. The analysis wouldn’t be hindsight-prone. And it would put today’s evidence into stronger narratives – the art taught away from the invention, the invention was greeted with surprise and skepticism. It would also reanimate multiple invention as a contraindicator to patentability, a doctrine that has now died on the vine.

Katherine J. Strandburg, DePaul University College of Law, What if we relied on user innovation to produce business methods and did away with business method patents?

Some people invent things to use (not sell) them. User innovators are less likely to need patents.

Business methods are often user innovations – administrative or “back-office” methods, such as keeping accounts, inventory tracing, payment processing. They’re often tailored to specific business contexts and embodied in software. They’re sometimes purchased through consulting services or individualized software. They tend to be non-self-disclosing in use, which means they can be kept as trade secrets by user innovators.

A different category is products sold to consumers: hub-and-spoke mutual funds, online gambling, long-distance packages, Netflix, iTunes, services like massages, surgery, piano lessons – not user innovation; often self-disclosing as sold.

A third category, customer service methods, somewhere in the middle: one-click purchasing, method for internet payment, method for advertising. Very often these are user innovation, but self-disclosing in use unlike the back-office innovations.

Greater intrinsic rewards: efficient business practices are their own reward, and create a first mover advantage including customer switching costs. Business methods also meet heterogeneous needs and provide customized solutions. Some business methods are skill-based and inventors do them better. User innovators may have lower invention costs, since innovation is a side effect of use rather than the product of a research program.

User innovators are also less likely to use patents to disseminate their inventions. When use gives a competitive advantage, the user wants to avoid others having embodiments of the invention like one-click purchasing. But seller innovators want to disseminate their inventions to customers. Thus user innovators tend to use patents to restrict others’ use.

User innovators therefore make systematically different inventions and engage in systematically different patent behaviors. If there are these differences in responses to patent’s incentives, can we tailor patent doctrine to recognize that user innovation is different? One possibility: a broad exception for use as a business method, like a research use exception. There would then be no need to determine ahead of time whether something is a business method patent. Linedrawing would be easier in context, and there would be no defense for competing sales of the patented product.

For back office methods, sellers would be protected from competitors by patenting and from users by copyright, know-how, and difficult-to-implement methods. Users would gain the benefit of either earlier disclosure or earlier nonexclusivity.

For products, sellers would be protected from competitors by patenting, and from users by make v. buy tradeoffs.

Finally, customer service methods: if it’s non-self-disclosing, a use exemption would promote trade secrecy. If it’s self-disclosing, competitors will copy the method. The desirability of this effect depends on the extent to which inventive costs (often low) are balanced by first mover advantages, switching costs, etc. (often high).

Bottom line: an exception would address concerns about exclusive control of basic tools of eocnomic competition, similar to concerns about patents on research methods. It would tend to neutralize patents on “trivial” business methods, which are least likely to need patent incentives. Costs: more reliance on trade secrets.

Greg Vetter, University of Houston Law Center, What if the GPL had been patented?

The GPL is a business method of licensing software. Vetter claims it’s useful, novel, and nonobvious. There would be very interesting issues of equivalents (e.g. to the viral elements of the license) and other infringement questions.

New issue: dual licensing, which allows a choice of GPL or proprietary terms. That wouldn’t be possible if the GPL were patented. People have also adjusted the GPL to make exceptions, for example in the Linux kernel – application programs that run on the kernel don’t have to be GPL. Netscape released Mozilla as open source, under a “weak” copyleft – you could make changes to files but would have to apply GPL terms within those files. The exclusion of these possibilities could have had dramatic effects on FOSS.

Still, the hypothetical GPL patent enforcement possibility wouldn’t seriously diminish FOSS’s potential. There are other licensing modes, such as BSD-style licenses. There are synergistic effects of having code out there and available even if it’s not pure GPL. Also, there’s been notable pragmatism in updating the GPL v.3. There are issues about license proliferation – patent control could have limited such proliferation, assuming (as is likely) that the owner, Richard Stallman, would have licensed the licensing method to everyone.

Some software patents now address open source – Microsoft has a patent application for isolating the licensing layer of software, and another company has a patent application for looking for FOSS license terms in software. So this isn’t an impossible hypothetical, despite the incongruity of imagining Stallman with a patent.

Elizabeth I. Winston, Columbus School of Law, Catholic University of America, What if you couldn't patent seeds?

Like Strandburg, Winston is talking about the relationship between patents and innovation. There’s no empirical evidence of a connection between patent protection and innovation in the seed industry, especially given the non-patent alternatives in the industry.

The Plant Patent Act and the PPVA allowed plant patents, and the Supreme Court also decided that they’re eligible for regular utility patents. What impact do they have? Patenting has increased 3400% over the past ten years, but it’s not correlated with rises in productivity.

Technology, the shrinking number of industry players, the shrinking varieties of seed planted, and the shrinking alternatives to genetically modified seeds have all had more impact than patents. Trade secret is also an important part of making the seeds differ from year to year. Contracts, meanwhile, govern farmers’ purchase of seeds, and circumvent research and saved seed exceptions found in the patent laws. Courts have upheld these contracts as implementing patent exclusivity.

Q: Is the seed level important any more? If the changes are occurring at the molecular level, then maybe we wouldn’t notice innovation at the seed level.

Winston: She’s using “seed” in a general sense, to include these features. The courts have acknowledged that the contracts/patents extend to nonpatentable features, like the germ plasm.

Adam Mossof, followup: A lot of these techniques are only allowed under antitrust because of the patent protection. So without patent protection, Monsanto might not be able to use these other methods.

Winston: There are 140 varieties of croppable potatoes – but 7 varieties accounted for 70% of production. Seed companies argue that there are no antitrust problems because there are other varieties of seed, but in practice there aren’t, yet courts accept that argument.

Lemley, for Strandburg: Netflix, an intermediate system – what do you do about that, where there are back-office as well as customer service innovations?

Strandburg: Still working on an answer. The distinction between users and sellers is not well enough defined. Particularly with Netflix, a lot of what you might worry about competitors copying from the back office can be protected by trade secrets. It might be that competitors would need to know those things to copy successfully, which preserves a first-mover advantage. If all a competitor is doing is copying mail-order DVD rental, maybe that’s a “sales” use instead of a “use” use and should be protected by patent.

Lemley, for Vetter: wouldn’t patent licensing have prevented license fragmentation and thus been good for FOSS?

Vetter: The question is whether that would have helped or hurt the movement. Centralizing control over licensing would mean less forking, but is that good? Another point is that patent might have avoided some of the debates over whether the GPL was using copyright or contract, and whether contract was legitimate.

Q for Strandburg: The English approach would be to say that people who are good at inventing business methods should be able to do so and license to people who are good at using them.

Strandburg: But the empirical study of user expertise is that you can’t separate out user inventions that way – certain kinds of inventions tend to come from users because of their experiences.

Q for Strandburg: What is the product I buy from Netflix? The DVD or the service? Does the user/competitor distinction collapse?

Strandburg: She defines it according to what the patent claims. Better statutory language to figure out what “use” is could help.

Q for Strandburg: Why not focus on the inventor instead of the invention? If it’s developed by a business user, maybe there will be a better correlation between the features and the rules.

Strandburg: That might be easier in the research tool exemption area. It’s hard here because figuring out whether an inventor was a user is difficult, though maybe no harder than other lines. (Wouldn’t that threaten to restate the flash of genius test?)

No comments: