Friday, May 15, 2015

Comparing Next-Best Policies in the Patent-Troll Debate

by Neil H. Buchanan

In two recent Dorf on Law posts (here and here), I offered some thoughts on the supposed problem of "patent trolls," which are companies that buy inventors' patents and then sue (or threaten to sue) purported violators of those patents.  Supporters of proposals to change the system claim that trolls are extracting illegitimate gains from innocent parties, mostly (but not entirely) by negotiating nuisance settlements with companies that would rather not take the risk of losing big in court.

Echoing comments by Professor Dorf, who wrote about this subject here two years ago, my initial conclusion was that the problems with the patent system seem to have nothing to do with the trolls.  In particular, it appears that the U.S. Patent and Trademark Office (PTO) grants patents that are too broad, and that the courts have done little or nothing to narrow the reach of patents, such that it has become almost inevitable that firms will have violated some patent at some point, even if they had no intent to do so.  Or, at least, I am willing to take those assertions as fact for the purposes of this post.  And if all of that is true, then fixing those problems would make the trolls go away.

In part of my post yesterday, I pursued a line of thought suggested by Professor Dorf's earlier post, and by a comment on my May 1 post, both of which noted that a key problem is the absence of a searchable database of patents, making the lack-of-fair-notice problem especially severe.  I observed that this could possibly point to a partial solution: If a searchable database could be created, then people would be on notice that certain patents exist.  Even if such a database could be created, however, I nonetheless observed that this would be at best a deeply flawed approach, because it "cannot reasonably become one of the standard operating procedures for non-tech companies" to have to consult the database.  Providing notice, in other words, would still imply a legally onerous burden on innocent parties.

In the comments on yesterday's post, readers pointed out that public (PTO) and private (Google) databases already exist.  This is not the point, however, because the question is whether the databases are useful in solving the problem at hand.  Everyone appears to agree that the current versions are not useful in that regard, and some readers suggested that it would be simply impossible to create a truly complete patent database that could provide the kind of notice to potential violators (or, more optimistically, potential licensees) that would be needed to address the problem.

All of that is fine by me, because I had already questioned whether such a database, even if it could exist, would actually do much (if any) good.  I then noted that, in any event, "the court system is there -- in all of its imperfect glory -- to sort it out, and to assess proper penalties."  The transaction costs and imperfections in the legal system are always troubling, but it was difficult to see how one could continue to have a system of patents without some judicial or quasi-judicial method of resolving possible violations.

Of course, some people might simply argue that we really should not have patents at all.  If the problem is that people are innocently stepping on overly broad patents, and the system cannot bring itself to issue only narrow patents, then maybe the solution is to eliminate patents.  That way, there would be no court cases at all, because there would be no legal property to protect.  This could be done as a formal matter (via constitutional amendment) or as a functional reality (via legislative, administrative, and judicial action).

Let me be clear: I am not endorsing (or rejecting) an end-all-patents solution.  I am simply describing it as a possible response to a situation in which a more nuanced solution continually eludes us.  But let us think about some less extreme solutions as well.

In yesterday's comments, some readers argued that a system in which people can litigate patent violations creates a severe ex ante vs. ex post problem.  If a potential violator had known what trouble would come, she might (if she were not inclined to license the technology) have been able to come up with a very cheap alternative to using that technology.  Once the process is underway, however, starting all over again can be prohibitively expensive, which provides room for patent holders (or the trolls to whom they sold their patent rights) to demand punishingly high damages, to which the innocent violator must agree.

This is certainly a problem, but it arises again from lack of notice, which effectively assumes that the violation is innocent.  Again, this screams out that the real problem is that patents are being granted for "obvious" ideas (else we would not have so many people accidentally reinventing all of these wheels).  But if the more extreme options of actually fixing or eliminating patents are off the table, what are the second-best options?

One such option is, indeed, the legislation to which I referred in my May 1 post, and which was described in the NYT article to which I was responding:
"The Senate bill, written mainly by Charles E. Schumer, a New York Democrat, and John Cornyn, a Texas Republican, would place restraints on demand letters, essentially threats to settle or be sued; require information on who actually owns a patent, making it more difficult to hide behind 'shell' companies; and require losing parties to pay legal fees if the court decides the suit was not 'objectively reasonable,' a deterrent to frivolous suits."
This second-best option, then, essentially says this: "We cannot fix the patent system, so we will try to reduce the number of lawsuits by making the net payoff from such lawsuits probabilistically lower."  And the way to make the payoffs lower is to threaten a patent holder with a financial penalty for bringing losing suits, when those suits were objectively unreasonable.  (That last requirement is a change from the House version of the bill, which had a simple loser-pays provision.)

Note that, because this is a second-best option, it still leaves plenty of innocent violators bloodied.  If it is true that people can accidentally end up violating patents, then it will still be objectively reasonable for a patent holder to assert in court that her patent has been violated -- unless the court is effectively being asked to treat the "objectively reasonable" inquiry as an invitation to impose a judicial solution on all of the underlying problems in the patent system (overbreadth, etc.).  And if that latter possibility is what is going on here, then this is a truly dangerous way to legislate, because it would essentially turn all of patent law into whatever the judges say it is, ex post, with virtually no guidance or limitation.

Assuming that the legislation is not a stealth rewriting of all of patent law, then, the most that we can say from a utilitarian standpoint is that it might be a worthwhile tradeoff, if the value (monetary and otherwise) of the frivolous suits that it would deter is greater than the value (monetary and otherwise) of the meritorious suits that the loser-pays possibility will deter (plus any concomitant loss in incentive to innovate, due to adopting a system that protects patentable ideas less vigorously).  Because even slam-dunk suits can go sideways, after all, it is not only bad actors who are deterred by loser-pays provisions.  It is not, in other words, just "the lawyers" who will lose from fee-shifting provisions, but some people with valid claims who nonetheless would no longer find it prudent to risk an adverse court ruling.

Are there other next-best options?  Although, as I made abundantly clear earlier in this post, I actually did not endorse the "mandate a truly searchable database" policy choice in my post yesterday, let us consider what role such a database (as necessarily imperfect as it surely would be) could serve as a second-best solution.  As I noted, "The ability to set up such a database, moreover, would create an impetus to fix the other big issue ... .  Making the ability to sue violators contingent on a patent's having been properly registered on a public database would all but force the players in the system to confront and fix the overbreadth problem."

In other words, an advantage of this second-best solution is that the very process of posting information on a database could make it easier to identify similar patents, and to begin to narrow what one can claim as a violation of one's patent.  But again, let us assume that I am wrong, and that such an imperfect database would not force anyone to confront the deeper problems with overly broad patents, and so on.  Even so, if the new rule is, "You cannot sue parties that were not given proper notice, and proper notice now requires listing on this centralized database," then we are more directly attacking the ex ante/ex post problem that I described above.

Moreover, the concerns that I raised yesterday, regarding who could reasonably be expected to consult such a database, could also be integrated into any such reform.  If the concern (as reported in the NYT article) is that restaurants, hotels, and other non-tech companies are receiving demand letters and being sued, then we are really saying that there is an entire category of parties who should never be sued for patent violations, because they are always innocent when they violate patents.  Only the people who might actually be wrongly and deliberately profiting from stealing the ideas of others should be sue-able.

Do I like that set of reforms?  Not really.  It is, by design, an attempt to figure out how to eliminate the egregious stuff that might currently be happening, without attempting to deal with the underlying problems.  And my ambivalence here is largely motivated by my sense that the political debate -- even on this very technical set of legal issues -- is being driven by anecdotes rather than data, and that we have no real sense of the degree of the problem.

Note also that a "searchable database as notice" second-best solution is likely to kill off quite a bit of trolling.  No more holding up non-tech firms.  No more suing for violations of patents that are not reasonably searchable.  Importantly, it is much more difficult to imagine truly wronged parties who will not get justice under this system (compared to those who would be scared off by fee-shifting provisions).

To be very, very, VERY clear, I am not endorsing any solution.  I am simply saying that it remains unclear whether there is a serious problem that needs to be addressed in the first place, that the anti-lawyer legislation has costs as well as benefits (neither of which are obviously large or small, relative to each other), and that there are other imperfect measures that could discourage what we apparently want to discourage.


Michael C. Dorf said...

Here's another possibility: Change patent law to make it more like copyright law, by requiring actual copying for a successful infringement action. To be sure, this adds a layer of transaction cost in litigation--the need to prove copying, typically through indirect evidence. But one could deploy presumptions that would themselves address the overbreadth problem, such as this one: Presume no copying where the patent at issue is not being practiced.

Greg said...

I was about to suggest EXACTLY the same solution that Prof. Dorf just suggested. The bulk of the problems with patent law come from the lack of a requirement for actual copying, when in fact most infringement comes from two people coming up with the same solution to the same problem, it's just that one of them encountered (or anticipated) the problem first.
I would argue that this is really an obviousness problem, not an over-breadth problem.

Justin said...

Agree with Professor Dorf and Greg, although I would prefer reforms also get us out of the language of property and instead have its own discrete rules. Ultimately, while not exactly either, infringement should be more like "theft of trade secret" (albeit a trade secret that is not entirely secret, nor should it be) than "theft of property."

Neil H. Buchanan said...

Nice to see consensus emerge! And not at all surprising that this better idea (certainly better than both of the next-bests that I considered in today's post) is not what Congress is considering.

And, yes, Greg is right about my sloppy use of overbreadth. It probably would have been better to do a global search-and-replace, putting in "overbreadth and/or obviousness" wherever I had referred only to overbreadth. I don't think that any of my points were unique to either concept.

My bottom line is that I'd vote "no" on the current anti-trolling legislation, simply because it is such a bad fit for the supposed problem. If the underling problem is really a problem (and I suspect that it might be), then some better option (such as Professor Dorf's) is readily available. The status quo ante is better than a false solution that will preclude other solutions (because people will argue that the anti-trolling legislation hasn't yet had a chance to work).

john gabhart said...

There could be more leeway given to judges in patent cases, in that,
#1 judge could find a suit to be in malice and award damages to the defendant, under plaintiff and attorney license.
#2 or, there were some grounds of infringement, but not enough to warrant remedy.
#3 or, remedy awarded.

Something has to be done.