Andrew Torrance Before I sold my soul and became a lawyer, I was a biologist and got a PhD. I find it hard to take a strong position for and against intellectual property. I would like to see the evidence. I would like to present today what I think is a new line of evidence about patents. Eventually I hope to expand this to include other forms of intellectual property. I think the first talk and my talk are going to dovetail and come to similar conclusions. I am a visiting professor at a university. I will be working at the Sloan School later. My home is the University of Kansas School of Law. I am going for high status on American Airlines this year. So, I am talking about experimental evidence to support open innovation. I am a patent attorney. I have been doing this project with Bill M. Tomlinson. They have a Social Code Group. This is an on-going project. We have been working on it for five years. There are very able coders like Bryant Davies and Nitin Shantharam to create our experimental system. We call it the Patent Game. It's a game that people can play that simulates the patent system. It's made to evoke the idea that there is an aspect of gamesmanship and maybe politics in its not so nice form that underlies the patent system. I am going to show you the results of 2 experiments that we did. The first was to test the hypothesis that patents spur innovation. And I will give you what I thought my conclusion would be. Before I started the experiment, I was sure that patents spurred innovation. I was hoping to be the first person that would prove that patents spur innovation. I was very excited about this. My collaborator thought that patents were evil- and the programmers were shocked that I believed in patents at all. And you'll see that based on this game, it's going to fail. Property rights by way of injunctive relief then if you prevail in litigation then the state will come down to you in full force. Liability rights represent a system where if you find that you violate patent rights, you can continue to infringe if you wish, but you still have to pay a royalty back to the owner of the rights. A property right in patents and liability rights in patents might have different results. All of my papers are freely available at ssrn. There's a reason for this (they aren't worth much). Feel free to download them if you like. There's also a Google Tech Talk that I did last year. It's a much much longer, I won't say richer but possibly more boring, 90 minute exposition on our method and results and what the implications are. This was given at Google a year ago before they fell in love with patents. At the time, I think they really liked the fact that our results were a little skeptical of patents. I don't know what the result would be today, or if I would be invited. There were 17,000 patents from Motorola so this might change their culture. Patents have very strong legal support. Congress will have the power to secure the exclusive rights to inventors for limited times.. etc. The patent act has been around for a long time. It has just been amended for the first time. It's traditional to claim that without the prospect of patent protection that we will get inefficiently few innovations. It's a claim that has been made for hundreds of years. This is a strong claim and if it happened to not to be true then it might limit the power of this because congress shall have power to promote the useful arts.. and what happens if this system doesn't really work? Shall it have power if it promotes progress? Or shall it definitely be patents and only patents? In law we are footnote happy. My research assistant came back with nothing. I thought this was hard to believe. This person was great with everything else, but not this well known proposition? What? I was having trouble finding evidence that people thought this. I finally found a report from "Patents in the Knowledge-Based Economoy". The literature on the impact of patents on innovation must be considered emergent. It basically means there isn't evidence yet, and it's out there and we will keep looking. Euphemisms all over the place. There's theory and empericism.. and I thought, maybe I could be the footnote. My collaborator is the world leading scholar of simulation and games. We decided to take a simulation approach. Couple of advantages. They can enable rigorous manipulation of experimental conditions. They allow for precise data collection and we collect every single keystroke. We collect so much data that if I tried to mine that data from now until I die I would not finish. Interactive simulations are a subset and we decided to have interactive simulations that would use human. They use human behavior to model human behavior. They generate a lot of data. You can use existing data to set parameters- you can feed the real world into the model. You can directly test alternative hypothesis and all of these are music to my ears. I thought, this was great, scientific expeirments in the law. That's great. Another principle that my coauthor thinks is important says that if the game is compelling and fun, players may actually forget about the underlying simulation. You want them to become immersed in the experience. You don't want them thinking "I am trying to prove this". You are trying to make as much money as possible in our simulations. Immersion is an advantage. You only need to explain the rules. We explained what the functions of the game were, and we allowed people to play. We didn't tell them why they were playing, or what we hoped to find. We had fairly good sample size. This is an untapped data source, people just don't do this stuff but they are starting to. So we wanted to simulate the patent system and test hypothesis about patent rights. How do they spur innovation? Do outcomes spur innovation in different ways? This is reaching pretty high into the sky. But this is ultimately what we all want to know, right? We have a database that allows us to implement almost any patent system function. We don't write all functions all at once. He wanted me to write down all of the features that I wanted. I gave him a list of 77 features. He said, okay, you can have eight and I got him up to nine. We have a fairly rich representation. We use an open source language to program this and I don't know if this predisposes it to open source bias.. we have autorefreshing, pul ldowns, etc. The gaol for the user is to win the game. There are social and financial incentives to winning. We made inventing into a series of elemnets that you can put together. There are combinatinos of elemnets are what we call inventions. You cna make thinsg and sell it on the market, or you can patent things and then later sell it on the market. There are prior arts out there in the system. People can license their patents, to buy them, to license in, and we also take into account infringement and enforcement. So if you have a patent, and someone infringes, you can take a look and decide to sue and hire some lawyers, and then the other team signs on some lawyers, and it relies on probabilities on prevailing in federal courts. We use real data to inform our court system. You can get damages up to 3x what you have been assessed liable to. There's some infringement called willful infringement because you multiply the damages by 3 if you are willfully infringing. There's also injunctive relief. We have one system where this is perfect.. if you are liable for infringing, then you get an injunction, you can't make something again. There are costs in money, costs for prosecution, litigation, patent creation; there are opportunity costs, like less time inventing and more time patenting.. large companies have huge departments that can sue while other departments invent. There are multiple users versus no users and computer-vs-computer. This is an example of the interface for the game. There's a design box, an inventing box, and you drag those letters, you can .. but you can drag say a C-D-E down and you can make that, it's wroth a certain amount. Different distributions of value can be assigned. So sometimes it's worth something different in certain games.. you have to explore the inventing space differently in each game. You can license and sell in each case. We can take this information off the screen, in some games we give people complete information and in tohers no information. There are a number of strategies that you can imagine using. You can play it safe. You don't infringe on other rights. You can just play it safe, stay away from patents. If you infringe accidentally, you can say sorry. You can be fast and loose and you just make what you want and if you get sued you'll deal with it. There are other strategies, you might make patents in the beginning, not make much money, but later pull in the crops. You can strategically patent a whole bunch of things and later act as a patent troll and sue sue sue. There are licensing strategies, enforcement strategies, all sorts of mixed strategies. Players really liked playing. Some of them would call me at night and say they want to play. There was lots of wooping and chest thumping. Players invent selectively and intelligently. Sometimes they license their works out, sometimes they will buy and sell. They often litigate. They usually get excited in an angry way about infringing. The legal rules seem to mean something to them. Our hypothesis was that the prospect of patent protection for inventions promotes the market more than the other case.. we ddid IRB-approved trials. We had a relatively easy ride through the IRB approval process. We had a system that had patents; we had something that wasn't open source - you could patent or affirmatively make them unpatentable, you can make an anti-patent - this is not like copyright or copyleft or open source and it was our trial. And then a pure commons where there was no patenst available. Here's what we got.. for innovation, the number of unique inventions during a 30 year period. As you'll see, the pure patent system had roughly 80. The patent and "open source" system was mostly 80 too. The pure commons produced quite a bit more, almost up to 100. This was significant in a statistically significant way. We also looked at productivity. The total number of inventions whether or not there were repeats.. and here the patent/open source system fall backs as opposed to pure commons. Pure commons doubled. And pure commons completely outperformed the other systems when you consider wealth. Some possible explanations! Maybe it's just a terrible simulation. Maybe it's garbage. Maybe in general it's a bad system. Maybe unless the parameters are set, and maybe we set them wrongly. Maybe we didn't choose the right players. Or the alternative is that.. despite the received wisdom, commons based innovation works better than patents. Cows in Switzerland. Alwuern Ostrim won the Nobel in economics about the tragedy of the commons. If you put cows in a village, they will eat the grass until there's only dirt. Maybe for some reason the English were unable to keep their cows from destroying the commons. For some reason, there are many commons that are green year-round.. there are legal mechanisms to make sure there's not too much green munching. For a long time, she was a voice in the wilderness agains the traditional view that everything will always gobble up. Maybe the tragedy of the anti-commons, or the benefit of the commons, maybe there's something to it. A second experiment that we did, we thought maybe we could tease out some more of this phenemono.. we looekd at injunctive relief, liability rules, property rules, etc. This came out of a paper - in law it is famous - Property Rules, Liability Rules, and Inalienability - One View of the Cathedral. It's a fabulous paper that sets out the idea that depending on the legal rule you will get different outcomes. I did a fellowship a few years ago, and some of the faculties were Austrian. A bunch of Austrian economics folks. They wanted to test our system using the property/liability rule rubric. So property rules and entitlement.. removing an entitlement from a holder must buy it where the value of the entitlement is agreed by the seller. Liability rule denies the holder of the asset to exclude others or indeed to keep the asset.. or is helpless to resist others from taking the property by paying its fair value determined by some neutral party (the courts). There is strong support for strong property rules in the courts. Eppstein in Chicago said that in any comprehensive legal structure, property rules should dominant liability rules. The dominance of property rules over liability rules is universal over physical things. So we thought we should look at this at an intellectual property format. Property rules by injunctive relief would promote patents more than by damages. Patent protection traditionallly says that the inventions would be hamphered and that fewer inventions will be generated. Strong patents with strong injunctive relief. We got IRB approval. We did four treatments. We did trials. And same duration for the data as the previous trials. Oh and by the way, our subjects are paid to participate and also paid a bounty if they win and this comes out of experimental economics literature that says that we should pay them to show up and then pay them to win so that they will really really try to accelerate at winning. We used the same proxies - innovation, productivity, wealth. So, in innovation, or unique inventions, interestingly again, the system with no patent system does really really well. But which among the damages - injunctive relief by property rules or liability rules? Injunctive relief and injunctions are the same.. but damages are different, and neither is in the stratosphere. Under productivity, the total number of inventions made.. and injunctions stays low, and damages does well and starts to approach the commons system (but it's still statistically different). Under wealth, the reuslts are similar... letting people infringe, but they have to pay for it, and some reasonable cost, may be better than telling people that they must not infringe and then allowing courts to draconially preventing it from happen. It's highest without rules, which is consistent with all of our expeirments. It's relatively high with liability, low with property/damages rule, and with both rules it's the lowest. Perhaps liability rule spur more innovation than property rules. This challenges the traditional view that injunctive relief is generating innovation. In 2006, SCOTUS dealt with this issue. If you had a patent and there's a finding of liability, you automatically had an injunction - no inquiry was put into it. You would get an injunction immediately. This is different from other forms of law where you have to consider the effects of such action of such power legal relief against someone. Usually you don't get an injunction at all. SCOTUS decided on this in 2006 and decided that patents should not have special cosnideration in injunctive relief. And they would not allow injunctive relief and that the analysis is mandatory and in some cases injunctive relief would not be appropriate but damages would usually still be granted. We went from 100% for injunctive relief and now there's 70% getting injunctive relie. The choice of legal rules might effect wealth outcome. Possible explanations? Maybe my system sucks, maybe my players suck, or the time or parameters are all wrong. This is all possible and we will continue to explore these. And maybe patent damages spur innovation more.. and liability rules outperform property rules in patent law. We're continuing to look into this. We're always looking into our theoretical framework. We're experimenting with our design of experimetns. We have five versions or designs of the game. Maybe prosecution costs- maybe if the costs were higher or lower that might effect outcomes. It costs $3-$6M to enforce patents. Patent term, how long it is, how .. the new patent law will make prior art more strictly applied and it will be harder to get a patent now if you don't rush to the patent office immediately. Information costs... knowing what others are doing.. user characteristics.. maybe different groups of users.. interpretation of results.. we're trying to be careful. The conclusions are from within a simulation, maybe they don't apply to the real patent system. Based on all sorts of other evidence, it's intriguing that it lines up with all of the critics that say that the patent system doesn't work like most people think. Can we find situations in which the patent system is doing well? I think there is a hint of this in liability rules. Maybe we ought to let people make other inventions, but pay them a reasonable royalty instead of an absolute right to say no. So in conclusion, there's a mysterious and dark lack of evidence about the current patent system and its effectiveness. Experiments suggest that patents may harm innovation at least under the conditions that we have set so far. Some people like trademarks... An increasing chorus of voices, like Lessig and others, are voicing the concern that we do not nderstand intellectual property and that the constitution might be wrong about the beneficial effects. We think we need to explore this thoroughly. Biotechnology is especially fond of patents. It was based on a SCOTUS case in 1980 that allowed the patenting of anything under the sun that was made by man - genes, proteins, etc. Even that has come under sustained attack, like ACLU vs. Myriad.. where the court found that you can't patent genes. That was overturned by the court of appeals, but it still remains to be seen if SCOTUS will take this up and there's a good case that the result will not be positive for gene patent owners. Most commentators think that SCOTUS will not take the other prometheus case up if they were going to confirm the patentability of biotech. DOJ, one of the defendants in the ACLU vs. Myriad case.. they were suing DOJ and the USPTO that they do not have the right to issue patents on genes. They filed a friend of the court brief, even though you're the defendant, then sure you can't patent genes at certain times. They sent an amicus brief..they said naturally source DNA should not be the source of patenting. The America Invents Act has an interesting provision... section 33 of it says that you cannot patent a human organism or anything directed to or incompassing a human organism. That might end up including genes, stem cells, methods of diagnosis or treatment, all the bread/butter of the biotech industries. Would that be bad for biotech or could ti be good? Could it foster open biotech and might that lead to a better biotech? Rather than luxuries to be tolerated within the law, open science and open innovation might be vital to innovation and technology. Until this time, they have been seen as the eccentric cousin and very well meaning, and in certain circumstances producing good results. The evidence in favor of intellectula property is lacking enough that it might make us question those that sideline open science. Maybe open science should be the mainstream mechanism for generating innovation. Maybe patents are the luxury that should be tolerated in certain circumstances and hopefully not tolerated at all. Thank you.