Eyes Above The Waves

Robert O'Callahan. Christian. Repatriate Kiwi. Hacker.

Friday 14 January 2011

Playing The Game

The Problem

Innovation in software happens roughly like this:

  1. Come up with some good ideas (perhaps 1% of the technical work).
  2. Implement them in software, test the software, deploy the software (99% of the technical work).
  3. Succeed in the marketplace (even harder than the technical work).

The US patent system offers anyone who completes step 1 a monopoly on the use of the ideas. Once other groups to complete step 3, the patent holder can extract a slice of their revenues. This system creates strong incentives to stop at step 1, obtain patents and extract money from others who pursue steps 2 and 3: you avoid the costs and risk of steps 2 and 3, and even more importantly you avoid the risk of being pursued by other patent holders. (The latter risk is particularly acute in software, where in step 2 you will routinely create and use many implementation ideas that have been patented by others.) Groups pursuing this strategy are called "non-practicing entities" or more colloquially, "patent trolls".

Currently most groups aren't following this strategy. One reason is tradition. Another reason is that most technically adept people desire to implement technology for people to use. But nevertheless "non-practicing entities" are proliferating and that will continue while this incentive structure persists. In fact, as NPEs grow, the risk of taking steps 2 and 3 grows, so more groups will choose to be NPEs --- a vicious cycle.

Obviously a patent system which discourages creation and deployment of software technology is failing its goals, but the problem is not obvious to everyone (many people deny it exists) and even if it's obvious, vested interests prevent it from being fixed.

The Solution

I think we need to make the problem obvious to everyone by following the incentives to the hilt, proving by demonstration that the system fails. I would like to see a more rapid increase in NPEs. I would like to see a huge jump in patent litigation. I would like to see most of the profits of practicing entities diverted to NPEs and litigation. This would encourage them to abandon their support for the current system.

Some of this is happening naturally, but there are a few things that should be done differently by NPEs that want to reform the system (let's call them "ethical NPEs"). Some wealthy PEs have bought cheap protection from larger NPEs such as Intellectual Ventures by investing in them. Ethical NPEs should not make such deals. Some NPEs sell their patents to other entites; obviously, ethical NPEs should not sell their patent assets to non-ethical entities. To focus effort on reforming the system and minimise collateral damage, ethical NPEs should focus on pursuing PEs who do not support appropriate reforms. I also suggest that ethical NPEs should focus exclusively on the US market. If the US reforms, other countries will follow, and while the US software industry is crippled we can hopefully keep advancing technology elsewhere. A nationalist sentiment may aid reform efforts too.

Possibly the legal system will be unable to cope with the load. If so, that will add to the pressure for reform. As I understand it, a logjam of litigation would mainly impact patent cases, so massive litigation by ethical NPEs would have a nice side effect: litigation against PEs who support patent reform would be crowded out and delayed by litigation against PEs who do not.

This strategy risks creating a large pool of NPEs and lawyers who act in their own vested interests to perpetuate the current system. But that's happening anyway; hopefully they can be stigmatised.


I'm not currently in a position to practice what I preach here --- my employment agreement with Mozilla makes it impossible, and Mozilla is more important to me than ever. Hopefully other people will pick up the slack!

The ideas here are not new. Florian Mueller's "Fair Troll" model is closely related. I think I read that a while ago and forgot the source until "dave" reminded me in comments. Thanks!


I do believe you should talk to Stephan Kinsella.
Some previous discussion of this idea, sometimes called the "Fair Troll" model:
I think some examples would be very helpful here. While I think software patents are less interesting than dirt, I don't see practical negative effects. I also don't think that the profits of software companies are largely related to patents (even though the lawyers think so). Rather it's the reverse: companies with profits attract lawyers who cause patents to be written and bought.
One of the fundamental problems with patents is that they obtain value only through litigation. Having a patent issue 'officially' has no practical meaning. But if a patent is upheld in a court, it's value skyrockets because then the lawyers know they can sue and win. This system strongly favors organizations with lawyers, so much so that organizations with only lawyers (your NPEs) are favor over organizations with none.
Aryeh Gregor
There's a negative feedback here too. More patent trolls means less innovation, which means less money for the patent trolls to feed off of. There's some threshold at which there are so many patent trolls that there's no longer much profit in it, since someone else will get there first. So it's not really a vicious cycle.
I agree that the only plausible way to change things is to demonstrate how horribly broken the system is. The way to do that would be to have a credible threat of two large competing companies actually seeking injunctions against each other for all their patents, instead of cross-licensing. If both continued to the end, they'd be destroyed, so normally such a threat wouldn't be credible.
So to make it credible, have a bunch of organizations irrevocably pool their patents on condition that none of them use them to sue each other, but any of them can sue anyone outside the pool, using any of the patents in the pool. Get a bunch of small companies together and you could get a decent pool. Then invite the patent trolls in. If the pool gets big enough, large companies would be so beleaguered by the patent suits from it that they'd be forced to join, contributing their patents to the pool. Since most of the actual lawsuits would probably be filed by patent trolls, countersuits would be impossible, and there'd be a credible threat of any company that didn't join being sued into the ground.
Of course, I have no idea if this idea is actually workable. There would be lots of details to work out -- for starters, we probably don't want to kill *all* patents. But if it did work, it would stop large companies from enforcing their patents. Then they'd lobby to cripple patent law to stop attacks by patent trolls. All solved in a nice grassroots fashion. If Mozilla were interested, it could start such an organization and pitch in a lot of frivolous patents that it asked its employees to file for, but probably it would be too unpopular or ethically dubious for organizations like Mozilla to want to take part unless forced.
(I'm pretty sure I've mentioned this idea before on your blog, as well as elsewhere.)
Robert O'Callahan
johnjbarton: just about every example of software patent litigation going on right now (and there are a lot) is a "practical negative effect".
Aryeh: the vicious cycle has a bottom, yes --- most do. But the bottom is a very bad state.
Your idea sounds good but I suspect you might run into antitrust issues.
Bruce Hoult
I've been saying for a while that I'm quite happy for every obvious software idea to be patented, and the sooner the better -- preferably in the 80's so they've expired now.
Do you have any comments on what Paul Allen is up to? When I first heard about the current patent lawsuits my immediate reaction was that he was implementing your strategy. But maybe my bias towards thinking of him as "the good Microsoft guy" is naive.
This is a beautiful post. And Paul Allen is a Dick.