Saturday, May 29, 2010

Open and User Innovation and Intellectual Property Enforcement

Thanks to the Engelberg Center at NYU for making this meeting possible, and Pam Samuelson for hosting!

Pam Samuelson: Aim: policy levers to make user innovation more lawful. What would a user-friendly IP system look like?

Eric von Hippel: Couldn’t believe it when he found out what the law said; thus began a conversation with lawyers. A firm can be both user and producer, depending on the innovation. Users tend to innovate before manufacturer comes in; producers look like innovators because they come in with resources and can invest more resources per unit. Missing: at the beginning of a market, the size of the market is small and uncertain. Producers may not have a large-scale advantage. Not that manufacturers identify user needs, develop products at private expense, and profit by protecting and selling what they developed, but rather that lead users innovate to solve their own needs at private expense, then freely reveal their innovations, some of which get adopted by other users, and begin to grow/separate out from the froth of experimentation. People get tired of reproducing the innovation each time, small firms appear to sell it, then large firms come in (and take credit).

Lead users have strong needs. The market isn’t there for them yet. Firms are not acting irrationally in refusing to provision them. But creates invisible front end where users act as feedstock for manufacturers. Because it’s a profitable feedstock, manufacturers trying to stop users might be made to see the purpose of encouraging innovation (he is very optimistic!). Survey of major new products: e.g., scientific instruments. Very few real game-changers over four decades. They come from users, and they don’t look like “products.” Manufacturers use engineering to turn the things into products with operator’s manual, etc. Manufacturers say they invented it, because the user design was a piece of crap and they engineered it to be nice. Typically it’s the manufacturers who then apply for patents.

Users developed GPS orientation for tractors to put fertilizer in right places, center pivot irrigation, and other agricultural developments. Prototype of latter used old bike and other materials to hand. Commercial version is very nice, but the same thing as the prototype, except from the manufacturer’s point of view—the quality of the welds on the prototype is terrible!

Similar results in other fields: health products (Gatorade), personal care (protein-base shampoo), sports equipment (mountain bike, mountain-climbing piton), apparel (sports bra), food (chocolate milk, graham cracker crust), etc.

Users aren’t always innovators: 77% in scientific instruments, 10% in engineering plastics. Users tend to develop novel functional capability (first sports nutrition bar, first scientific information of new type, first mountain bike)—82% of novel functional capabilities were user-developed. Manufacturers then develop dimension of merit: better-tasting nutrition bar, improvements to speed or size of laptop, smoother ball bearings for bike, improvements on power supply—87% of dimension of merit improvements were manufacturer-developed.

8% of UK consumers in last three years developed/modified a product. Scale of dark matter here is huge, 100-1 relative to producer activities. Because it’s not visible, it’s not in policy, and that’s a problem. Note that publications and patents are better evidence than use in fighting off a patent.

Dan Burk: what are patents supposed to do? Reward innovation or its commercialization?

Kathy Strandburg: Also depends how easy it is for the user to make it, or make it for someone else—that affects commercialization.

Von Hippel: also notes that users are predominant innovators of technique, which is often beyond the scope of patent. In rodeo kayaking, users developed 100% of technique innovations, sometimes modifying rivers to create moves. Product innovation is the area most favorable to producer activity. Consumers collectively spend much more money and person-hours on innovation than do producers in at least one field—data from whitewater kayaking, $100 million sport (though we can’t tell whether/how much it’s wasted effort).

Users are also developers of many important services: study of retail and commercial banking services; hotels (remember trying to use internet in your hotel room in early days, when you’d unscrew the phone so you could use AOL?—hotels responded with tamper-proof screws, then users responded with screwdrivers, then hotels eventually monetized it and said they’d invented in-room internet—a typical story of active resistance to user innovation). Typical time to commercialization: 7 years, waiting for demand to build up.

Bottom line: determining authorship of community developed innovations, or allocating shares of authorship, is very difficult. Lego has 200 internal R&D people and ~50 model (piece) developers, and 20,000 self-organizing people outside, many of whom innovate. Lego has now in effect outsourced design—users can post a design for an object and Lego allows you to create a box with the necessary pieces; Lego then monitors how many boxes sell, and can commercialize ones that do well.

Users tend to be open than producers: process innovations more open, product innovations more closed. Dutch companies as both producers and users: Many more product innovations (made in the role of producer) are protected with IP (45% v.s. 11%); 85% of process innovations (made in the role of user) voluntarily help with copying/transfer of knowledge, because what they want is the product, compared to 17% of producer/product innovations. Greenhouse owner wants a robot to tend flowers to save space; shows the manufacturer how to make it.

Comment: hedge funds are user-innovators, but aren’t open—competitive advantage is relevant.

Von Hippel: yes, but at least some users typically are unlikely to regard an innovation as a competitive advantage.

Dan Burk/Michael Carroll: process and product are not the same as user and producer—there may be a terminology issue, especially since process has a specific meaning in patent, whereas von Hippel is using it to define whether the innovation is an input to the thing they want to sell.

Pam Samuelson: Data show firms are much less likely to patent process innovations, in part because they can be kept as trade secrets. More generally: you just want to make your thing better/faster/cheaper; you’re still focused on the product.

Von Hippel: as competition goes up, you might try harder for secrecy; but secrecy is expensive.

Kathy Strandburg: The voluntary transfer data suggests that the trade secret story isn’t particularly explanatory.

Burk: process patents are murder to enforce; why waste the money on getting one/maintaining secrecy?

Strandburg: in von Hippel’s terms, process innovations (perhaps better term would be internal innovations) could be transferred to competitors, but really they’re more usefully transferred to the people who will make/engineer the machine for the user—the guy who will make the robot for the greenhouse operator. Whereas what von Hippel calls product innovations, if transferred, would be transferred to competitors; the reasons for doing this would be different. And you might be able to maintain some control over competitors’ access to an internal innovation even if you outsource production.

Von Hippel: data on user innovations transferred to producers. Many innovations aren’t interesting to producers (asset specificity: not useful), of course, whereas others are held for competitive reasons, but the percentage of innovations transferred was in the 22-26% range in international comparisons. 48%-60% were transferred at no cost.

Strandburg: compare this to %age of patents that are successfully commercialized, which is a lot lower.

Carliss Baldwin (presenting on work she did with von Hippel): what kind of system will give rise to what innovation and by whom? Since Schumpeter, economists, policymakers and business managers have assumed that the dominant mode of innovation is a producer’s model. Producer initiates change and consumers are taught to want new things. Teece 1996: “[T]he business firm is clearly the leading player in the development and commercialization of new products and processes.”

But that’s only one mode of innovation. Others: innovations by single user firms or individuals, and open collaborative innovation projects. Not new, but newly important. Requirements for successful open collaborative innovation projects: modularity in design, chunked in ways that are largely independent but will operate in an overall system. Task divisibility within the modules. (Related to the expense of information about the tasks.) Another requirement: Very cheap all-to-all communication technology, so you don’t need a producer coordinating.

Modeling innovation: assume people are rational actors, who compare value of innovation to costs. Value is use value for users (willingness to pay) whether they collaborate or not. Revenue to producers is less than the sum of users’ willingness to pay. Costs: design, communication, production, transactions. Assume the last two are constant across type of innovation for now (though they won’t be). Communication is the cost of working together. Design cost is the effort of creating the design—a set of instructions on how to make something—can take experimentation and testing—right number of minutes to bake a cake, for example: design cost is the cost of starting with an idea/need and ending up with a set of instructions that will allow you to make the artifact/fulfill the need. Production cost is the cost of making the artifact using the instructions. (In reply to question from Burk about fixed/marginal: Fixed costs of setting up channel of communication may be much larger than marginal cost of using—that can be important to communication costs. It’s not costly for me to use my phone; moreover, buying the phone didn’t cost as much as it would have 50 years ago. Cost of information transmission over a distance has dropped a lot.)

Samuelson: design v. production cost?

Baldwin: the interface between design and production is the set of instructions. The process of coming up with a new artifact is part creation of instructions and part carrying them out. Production cost is the latter. In real production facilities, people tweak design all the time, but this is a clean theoretical interface because the design is pure information; the good itself may also be an information good, in which case the difference between the two is the cost of copying. As long as you’re tweaking the instructions, that’s design. So for software, coding that is nonalgorithmic but choice-based will be part of the design process by her definition; coding turned over to a compiler is production. This theory of design allows design to be consistent across fields.

Samuelson: Software folks do use the term “design” differently. There’s no variable production cost left when you get to compiling.

Suzanne Scotchmer: For software, production is copying.

Baldwin: right. The cost of copying is now nearly zero, but it wasn’t always so. [in response to question] Communication cost is the cost of communicating anything from one agent to another. You used to have to feed the messenger who’d take your message by horse; now you send an email: this is a decrease in communication cost even if you assign part of the infrastructure cost to the particular communication. Globally, both communication and design costs have fallen by orders of magnitude, though this is not uniform.

Single user innovator has no costs of communication (or if single user is firm, has lower in-firm communication costs than outside). So design cost is the key for deciding whether to invest. Very tacit designs thus must be single-user: example—a new course for a professor, where it’s simply prospectively incommunicable how you want the course to go—you have to create it to have the course; you can’t delegate to teaching assistants.

Producer has to communicate to users that it has an artifact for sale; this is part of communications costs. Sears Roebuck = example of communication cost decreasing, where catalog goes all over the US, notifying them of their options.

Open collaborative innovation: each contributor figures out the part she wants and can do at reasonable cost, and the part that would be nice to have but that she can’t do herself at reasonable cost. Each contributor has the option to communicate and probabalistically expects some (use) value. (Expected value of contribution can be negative, for example for a hedge fund that would expect its strategy to be appropriated if disclosed.) A church recipe book: we can make a meal using each other’s recipes. The total feasible cost is much higher than one person could afford acting alone, if she had to test all the recipes herself. If each contributes, you blow through that design cost limit.

Combining the models, we see that single-use innovators occupy some exclusive space (where communications costs are very high), and some which overlaps with producer innovators (who also have their own exclusive space, with moderate communication and design costs) and with open collaborative innovators (who also have their own exclusive space, where design cost is very high). Producer who has users do 9/10 of the design will do better than producer who tries to do all the design itself.

Von Hippel would say single innovator-users have three big advantages over producer innovators: apply sticky info about their own needs, no transaction costs, mostly freely reveal. Baldwin: Big disadvantages: may develop redundant designs, lack capacity to undertake large design projects, and most of the time they have higher unit production costs. Von Hippel would say: Letting producer-innovators run wild was a devil’s bargain because they created monopoly and secrecy, plus transaction costs; we don’t have to tolerate these any more. Baldwin: yes, but open innovation only has the capacity to take on certain large projects. Giving something away may threaten someone else’s rent, so it’s no surprise to see producer innovators seeking to claim territory.

Scotchmer: why can’t all these benefits be achieved within the firm? (Doesn’t Yochai Benkler go through this argument in some detail?)

Baldwin: in order to get actual work done, you need transaction-free zone. Contracts can be the boundary but you can’t be contracting within the zone. That’s Coase’s theory of the firm. Because of complementary use values, people discovered that it was viable and valuable to have a transaction free zone with no contractual boundaries.

Inventorship and Authorship by Communities

Andrew Torrance: starting from the proposition that user collaboration/innovation is significant, what legal changes are indicated?

Molly van Houweling: note that in the realm of user-generated copyrightable material, copyright is easy to get—e.g. my Facebook status—leading to difficult questions of who owns what; whereas patents are hard to get and so it’s less often worth it for, say, mountain bike racing innovations, so the issue of ownership simply comes up less often.

Wendy Gordon: there was a case involving collaborative theater—after a while, nobody can remember who contributed what; the judge said that everyone could own something, but each had to be able to identify their own contributions, which didn’t help.

Von Hippel: if something is developed collaboratively, there’s often very little documentary evidence to dispute a later commercializing company’s ownership claims.

Samuelson: patent scope as a lever: if the commercialization is merely an improvement, then some claims may be invalid, or might narrow construction of the claims. Pioneering inventions get a very broad scope; showing prior innovation might affect scope of patent, consistent with current law.

Strandburg: patent, we ask questions similar to those asked about traditional knowledge. Under current doctrine, it would be hard to define inventors, because everyone made a tiny contribution. Makes it hard for them to get a patent, but do we care? Is there a problem? Because this stuff is out there and not very well documented, it can be copied/stolen (traditional knowledge again): vulnerable to attack from outside patenters. Copyright is different because there, if you are not a joint author, then you don’t have a right and you can’t keep another from taking ownership.

Me: I may be misunderstanding Strandburg but I don’t get the distinction she’s offering between copyright and patent here. Especially given the power of legal threats, whether or not one could win a case. Are we just offering different prototypes of patentable/copyrightable works, the former more likely to be the product of large-scale collaboration? Maybe we think there’s a difference between patent trolls and strike suits in copyright? That is, patent trolls represent bad patents (patents that should not exist) whereas strike suits are more likely to involve good copyrights, just claimed by the wrong person (one not entitled to make a claim)?

Strandburg: Patent is not automatic—so if collaborators don’t get together to patent, there is no patent, whereas the copyright will be automatic; and you’re vulnerable to later claims by someone else who claims to have a valid patent. (But that part is true for copyright too: Stouffer suing J.K. Rowling for infringing Larry Potter & the Muggles. I don’t think this is about collaboration on the front end, it’s about vulnerability on the back end.)

Brian Carver: In terms of barriers to user innovation, contracts are also at issue: users can be stifled from the outset if contracts prevail over IP rights/freedoms. When restrictive terms become the norm, you stifle all manner of user innovation, providing monopolies in external markets outside the copyright—Blizzard has a monopoly on its game, which is fine, but now it has a monopoly on matchmaking so that you can find someone else with whom to play.

Stephen Maurer: why not rely more on defensive publishing to keep outsiders from appropriating too much? GPL can also prevent appropriation.

Amy Kapczynski: other stories about why users might have interests in using/innovating—researcher’s exemptions, farmers’ rights, moral interests (as Wendy Gordon has discussed), lack of participation in the market as justification for not making them pay.

Von Hippel: companies are eating their own feedstock. (Their time horizons are not ours.)

Strandburg: what’s wrong with a company patenting an improvement on a user innovation? Is there a problem with the fact that the users aren’t compensated? Maybe there’s an incentive problem; maybe we want users to be allowed to use the next step without paying the company because we want further follow-on innovation.

Jason Schultz: various doctrines about telling the truth about the people involved in an invention/work might be policy levers for finding/rewarding the people who deserve to be rewarded for an innovation: inequitable conduct, false marking, false designation of origin, copyright management information (at least in the way that it could work, though not as the statute has it).

Van Houweling: the standard prospect theory says that if users are good innovators patentees will license them on good terms, but the account of innovation we’re hearing fights that because it indicates that firms/improvers are not necessarily the experts on their own patented matter; if they aren’t the experts, then the stewardship argument doesn’t work very well.

Overlapping Rights, Derivative Works, and Blocking Patent Failure

Pam Samuelson: the way derivative works are defined now: (1) based on a preexisting work—not nuts about that, because what does it mean? But (2) gives nine examples in the statute as the kinds of things contemplated as derivative works: motion picture version of novel, novelization of dramatic play, translations, musical arrangements, etc. If we limited the right to other analogous uses, that would limit the disturbing/unbounded scope of the right. If plaintiffs had to show that the thing they were complaining about was like one of the existing, staid derivatives, and if they had the burden of proof, that would be a cheap fix. We could still get to that by reading this as Congress’s intent, but the courts have gone haywire over “any other form in which the work is recast, transformed or adapted.” Cutting the pictures out of a book = derivative work in the 9th Circuit, which is baloney.

We’ve screwed up all infringement analysis by following the idea that the derivative works right is superfluous to the reproduction right, and then we say that things that can’t possibly be substantially similar are—it is insane to say a Chewbacca toy is substantially similar to Star Wars. We should use common-law evolution to ignite an improvement defense (Wendy Gordon says fair use, but Samuelson wants to make fair use do less work). Baker v. Selden was argued as, in part, an improvement case. The 3 American commentators were all high-protectionists and wrote the improvement precedents out. It’s always the high-protectionists who write the treatises!

Me: I want to ask Strandburg’s question again: what exactly is the problem for improvers? Their problem is not that they can’t get further rights to commercialize. It’s not, despite many fears, that they can’t get compensated for unauthorized commercialization by the original authors. It’s that they get shut down (and maybe that third parties make money from it when the improvers can’t). If we are interested in policy levers, going for something that sounds like the right to get compensation may be so unpalatable that we lose what more user-innovators need, which is the right to be left alone.

Kapczynski: By definition, these people are usually not interested in excluding others.

Gordon: doesn’t see any way to argue that copyright beats patent in terms of blocking rights. The right to be left alone could be achieved with a sharp sword—some sort of blocking right. She would limit the derivative work right by requiring (1) a new instantiation—a book from a movie, as opposed to pictures cut out of a book; (2) use would have to be foreseeable, as the listed uses are; (3) use would have to be economically significant for incentives, as movie rights are. Loves the idea of improvement as a defense from history.

Carroll: Motives for tinkering are heterogenous; don’t funnel stories into efficiency. In copyright, people are talking back to culture. We want a regime that allows that, whether or not these works produce social welfare—they produce personal welfare that has value of its own and should be protected even if my criticism harms your view of an iconic work.

Samuelson: People want to concentrate on free speech, but some derivatives, especially on the software side, aren’t about speech. Innovation and competition matter too—that’s her unbundling fair uses paper. Free speech is important, but not the only thing.

No comments: