Patent Pools and Pledges - Panacea or Placebo?

Commentary by Florian Mueller - Last year, a lot of noise was made by OSRM, the city of Munich, and Steve Ballmer about the risk posed to open source by software patents. This year, we've seen a variety of initiatives by companies that "donated" patents to "protect" open source, and organizations like the OSDL and the Open Invention Network now try to pool such patent pledges.

I've been wary of those patent pledges and pools from the beginning. There's an awful lot of dishonesty in these all-too-obvious attempts to curry favor with the community and reassure customers. While the jury is still out on some of those projects, none of them has so far delivered a single compelling reason for me to believe that they're really going to be more than a placebo. Some make it sound like these pools are a bulletproof vest for open source, but it's more like you have a coin in your pocket and hope that a bullet will be deflected by it. Too bad the coin isn't even in a place where someone would usually shoot you.

While I do agree that open source should protect itself as best as possible within the legal framework that exists, cheap PR plays are not a substitute for a real solution. The pledges that I've seen so far had all sorts of shortcomings:

- Some pledged patents are of little or no value. Among IBM's 500 patents "contributed" in January, there were some that had nothing to do withsoftware, and many were up for renewal soon, with no guarantee that they'd actually be renewed.

- The pledges typically just relate to particular open source licenses (sometimes rarely-used ones) or projects, such as the Linux kernel (which is only a small part of a standard Linux configuration).

- Some pledges are revocable or haveloopholes such as vague conditions under which the patent holder can sue you anyhow.

- So far the quantities of patents involved have been negligible compared to the total number of issued software patents, and even to the number held by the "generous donors." Even in the long run, there'll be hundreds of thousands of software patents in the world that aren't subject to any pledge. In his speeches, Richard Stallman likens software patents to mines in a park: If there are 90,000 mines in the park instead of 100,000, it's still far from being a safe place to walk.

Even if you don't look the gift-horse in the mouth, there are fundamental problems that even the best pledges can't solve:

- You can't practically go about your programming job by always looking up a patent pledge database whether it contains just the algorithms you need. I don't think any programmer would seriously do that! And even if algorithms A and B are covered by patents in a pool, there may be a patent C that covers your particular combination of A and B, and that patent C may not be in the pool.

- Pledges which exclusively relate to open source aren't too valuable. Software under the BSD license is used in closed-source projects all the time. A project like PostgreSQL, which already felt forced to replace a caching algorithm due to an IBM patent, couldn't just base its development decisions on open source considerations alone. Then there are dual-licensing models for GPL software (MySQL is a well-known example) and companies that sell closed source software to finance their open source development efforts.

- These pledges are only made by organizations that don't intend to sue open source projects anyway. Patent holders who are potentially hostile, be it for strategic or purely financial motivations, won't pledge anything. It's nice to firm up the commitment of your allies not to act against you, but it doesn't reduce the number of enemies.

- If a company promises not to sue open source projects over a certain set of patents, it still doesn't mean that those patents can be used by open source projects for retaliatory purposes. However, the patent game is one of mutually assured destruction, like in the Cold War. If NATO hadn't had a single nuclear weapon, and the Soviet Union had promised not to use something like 5% of its nuclear arsenal, then we probably wouldn't live in freedom now.

- Companies usually can't even make their patents available for the purpose of building a counterthreat because those patents are already subject to existing cross-licensing agreements. If a new entity (such as the Open Invention Network) started acquiring unencumbered patents, then one day they might be able to grant a license to a company like Microsoft in exchange for a covenant not to sue Linux with its own patents. That could indeed make a major difference (even if only for Linux), yet it wouldn't help against trolls that have no products of their own. And a strategic aggressor could secretly arrange for such a troll to do the job.

No matter how you look at it, the only way to reliably solve the problem is at the political level: through legislation that excludes pure program logic from the scope of patentable subject matter. Sure, there's no shortage of people out there who say this can't be done, but they're all wrong. We've been toldmany times that the European software patent directive would come one way or the other -- until we got the European Parliament to reject it by a landslide of 648-32. The German Bundestag and Spanish Senado unanimously backed our central demands. The new German government has just vowed to counter, at the international level, "the trend to seal off markets, among other things by means of patent law." The time is ripe for legislative action.

Some members of the legal profession claim that software patents are an unalterable fate because theyhave a vested interest in sustaining the system. It's a tall order, but definitely possible, to change the legislative framework in our favor. In every parliamentary democracy.

Especially in the field of software, the patent regime no longer serves the public interest. In a perfect democracy, software patents would already be history. In the suboptimal democracies in which we live, there are special interests that oppose changes. Those have influence and deep pockets, but at the end of the day the most valuable currency in politics is voter popularity.

If all the companies who have pledged patents to open source, or who have contributed to those pools, decided to seriously campaign for legislation that abolishes software patents, then the problem could be solved for good. As long as they don't do that, they're not for real. Some may even have a hidden agenda of creating patent pools to gain effective control over the open source universe. We've got to watch out.

Florian founded the NoSoftwarePatents.com campaign. For his political efforts against software patents, he has been named as one of the "top 50 most influential people in intellectual property" according to Managing Intellectual Property magazine and is a candidate for the title of European of the Year.

Comments

Accurate Aliteration, Absolutely

[saskboy]

"However, the patent game is one of mutually assured destruction, like in the Cold War. If NATO hadn't had a single nuclear weapon, and the Soviet Union had promised not to use something like 5% of its nuclear arsenal, then we probably wouldn't live in freedom now."

The topic title taught us terminology.

Patent MAD seems like the only way to get to be a big company these days. With the dot com boom over, what else could possibly take a small company and make it competitive with a giant like ebay.com? Skype got bought out by who else? Ebay.com

A giant like eBay with their mountains of patented software stands to make a fortune every time someone else tries to write some of their own software to sell something on the Internet.

Re:Accurate Aliteration, Absolutely

[sterno]

"However, the patent game is one of mutually assured destruction, like in the Cold War. If NATO hadn't had a single nuclear weapon, and the Soviet Union had promised not to use something like 5% of its nuclear arsenal, then we probably wouldn't live in freedom now."

Interestingly, the patent world is evoling in a manner similar to the geopolitical one. Patents, historically, have been about detente and mutually assured destruction. Big companies would use the patents to wield power in the market, negotiating cross-licensing with eachother, and keeping their grip on the market place. The super powers in stalemate.

But today you've got the patent equivalent of Al Qaedas. People with single patents or small patent portfolios who contribute no product and base their entire business plan on lawsuits and licensing. Having a mass arsenal of patents does you no good against these companies because they don't make anything that you could sue them for. Small companies that legitimately want to make products are stomped by the big guns and the big guns are held ransom by law firms in IT clothing. This leads to an overall decline in innovation.

The patent system is in need of a drastic overhaul. My thoughts:

1) Require that patents only be able to extract license fees equivalent to a simple formula:

    R&D Cost + Legal Fees * 20%

Perhaps add some interest equation in there, but my point is this. If you want to research something, you'll be guaranteed licensing to cover your costs. If you have to sue to get compensation, that will be covered. Then you'll be guaranteed a profit margin of 20%. Once you've recouped the money, the patented item becomes publicly available.

2) Vastly increase the hurdles to getting and maintaining a patent. This would include exponentially increasing fees per year to maintain a patent (proceeds from this being directed back into the patent office to afford better review of the patent applications).

fire with fire

[Douglas Simmons]

If your company gets sued because someone successfuly patented the double click which is used in your Linux servers, because the patent is so friggin' ridiculous, couldn't you sue the patent office? Take contracts, for example: If you were running a jetskiing operation and you had people sign a ten page fine print agreement basically saying not to sue for any reason, and for kicks you removed the throttle spring so they can't slow down (not using any bomb like in that movie), contracts may not hold up. And I'm not talking just contracts that include something illegal to be entirely void, just contracts that cross the line of reality a bit too far. Apply that to patents, and you got a viable case. Right? Any precedents one way or the other?

That's quite a statement

[LaughingCoder]

In a perfect democracy, software patents would already be history.

What could be the reasoning behind this statement? If a perfect democracy means that the issue would be put up to a vote, with the majority opinion carrying the day, I'm not quite as sure as you that software patents would be history. Consider all those people who have a vested interest in maintaining the status quo - we know how they would vote. Now consider the software developers themselves (a tiny demographic) - I would guess that most would vote to abolish software patents, but some (maybe even many) would vote to keep them in place (after all, they write code and may not want others to use it freely). That leaves the big blob of people who know nothing about the issue. How would they vote? I would suggest they would be easily swayed by "campaign" advertising. Now, who has the money to run the campaign - the ones with the vested interest or the software developers?

Now perhaps you believe that a perfect democracy does not leave room for campaigning or advertising to sway opinions. If that were the case, given the large unwashed mass's lack of understanding of the issue one would have to assume their vote would split 50/50. So, in the end I don't see a perfect democracy changing anything.

Public Patents

[Doc Ruby]

Seems to me that a patent pool that accepts patents under a license obligating the pool not to enforce the patent, and offering anyone a free, unrestricted license to use the patent, is harmless. It's also indistinguishable from "public domain" in operation, but has the valuable bonus of an interested party which can actively fight others who try to privatize the patented IP. The public domain is a great legal principle, but the economics fail to protect the public - "tragedy of the commons".

Funding the "public patent" pool is a challenge, but a sufficiently diversified and pushy pool with lots of profitable licenses can probably pressure "association members" for enough money to operate. Or sell "value added services", like IP legal advice and "IP market info", like licensing activity. Maybe just a virtual organization, cheap in normal operation, that pulls dues from licensees only when they vote to challenge a rival IP claim to their free patent. If legal, possibly after the fact, when someone licenses a patent that had meanwhile been conflicted by a new, private patent. The economics of that operation might see a "reverse claim jumping" industry, where people spot unlicensed public patents in conflict with newer private patents. They license the public patent, then fund the suit against the private rival, which seizes the private income for the pool, compensating the new public licensee.

The power of this way of thinking, a real libertarian method for fighting private IP hoarders, offers a lot of new operations in the public interest, entirely consistent with private capitalism, while excluding monopolism. Let the good times roll!

A solution to consider

[Teppy]

What about adding a clause like this to both LGPL 3 and GPL 3:

By using this software you agree never to initiate a software patent lawsuit against any person or company. If you do initiate such a lawsuit, the license fee is $1M per year, retroactively, and you are defined as "aggressive". There is one exception to this rule: Anyone may initiate software patent lawsuits against others who are aggressive.

This has the effect of even making life difficult for litigation companies: As more and more software includes at least LGPL code, over time they would have to run their companies without the use of software of any type.

Re:about that...

[arkanes]

1) Wouldn't prior art mean that C isn't legal unless the person/company holds patents A and B (in other words, couldn't patent holders A (inclusive) or B go after patent holder C for infringment?)

No, you can patent innovations that derive from other patents, including combinations.

2) In closed source, how can it be determined (legally) that someone is infringing on another's patented algorithm?

(Software) patents cover behaviors, not just implementations. If your patent covers a method of embedding data in documents, you can generally tell from the applications behavior if it infringes - at least enought to start a lawsuit and begin discovery.

That being said, I don't know a thing about patents, and they kinda scare me (I picture a patent-boogeyman when I think about it) or is that what they are supposed to do?

Pretty much, yeah.

My quick & dirty solution to our patent troubles:

1. Shorten the terms. You get 5 years for free, and you pay logarithmically increasing fees after that. By holding a patent, you're withholding knowldge from the public good, and that should be taxable.
2. Mandatory licensing. A patent is there to encourage you to profit off your invention, to make it more likely that you will produce things based on it and thus benefit the public good. If you can't make a usefull product based off your license, then someone else can. This has the side effect of removing patent MAD.
3. Require implementation. You don't have to do it when you file - it can be in the "patent pending" phase - but when that time is up you either need a practical implementation or you lose your patent. Speculative patents are stupid.
4. Reduce burden of proof to legally challenge a patent - the patent office is clearly not capable of fully validation patents, so the legal presumption that it is needs to be removed.
5. Independent invention should be de-facto evidence of obviousness and the burden should be on the patent holder to demonstrate it's non-obviousness. If the patent is still ruled non-obvious, the independent inventor gets a reduce rate on the mandatory license fee.
6. Increase the number of challenges to patent validity based on form. A person "skilled in the art" should be able to accurately re-create a patent soley from the patent application. Expert testimony to the contrary should be a major blow against the validity of a patent.
7. Patent protection cannot be claimed on something that would be wholy protected by other IP law. No storyline patents, because any implementation of your storyline would be protected by copyright. Most software patents go away for the same reason. Patented hardware that required software to run is still okay, because the copyright covers the software and hardware covers the innovation in hardware. This would still allow some business and process patents.

Please Go Away

[thebdj]

This message is directed to Florian Mueller, hopefully a more reasonable individual then most the /. anti-patent zealots.
Your comment about IBMs patents is borderline FUD. While some of the patents may have nothing to do with software, those that do and are up for "renewal" (in the US you just pay your maintenance fee at 3, 7, and 11 years post grant) are usually only renewed if the continued protection is sought. In this case, by turning over those patents and not paying fees, they become fully public; however, most people in the public don't understand this system and simple add 20 yrs to the filing date and forget about it. This shows those patents are clearly available for individuals to use.

Another comment is the one about being limited to only certain licenses or as you say specific projects like Linux. Now clarify something for me, does that mean that individuals can then only release code under certain licenses? If so, which ones? Why is this really bad if the license limitations prevent people from using restrictive licenses. Ones specific to project (ie the Linux Kernel) may only really apply to those one things, and may not have much or any scope outside said project(s).

You bring up the combination patent idea where A and B are in the system, but your combination exists as Patent C. If there exists the proper measures as set out by SCOTUS and CAFC that A and B could be combined for C the patent can invalidated in court. If it cannot be invalidated, then you might actually have a problem...but in most cases an invalidation can be done unless combining A and B would be totally unheard of.

You mention that this doesn't reduce the number of enemies you might face. However, having your own patents at hand is good in the event of a court case, because you have your own protection there against their threats, particularly if you can show that their patent is either invalid or that your creation is along the lines of this patent and not theirs. It is also good because companies change regimes. Just because IBM is friendly with Open Source now does not mean they will be in 10 or 15 years.

The troll is a seriously problem, but the key to the troll is they are out to make a lot of money. The fact is that in the software world open-source does not necessarily have the money. There are bigger and better targets worth much more money. Large companies like Microsoft, Adobe, Sun, Google, etc. are far juicier targets then Redhat, Novell, etc. IBM is a nice big target, but in many peoples minds almost too big. Microsoft has been too quick to settle in the past (possibly because they know they will lose) with various cases (with maybe the exception of EOLAS). However, IBM is fighting SCO tooth and nail and not just sending them on their way to get rid of the pest. When IBM wins this case, they will be much less tempting targets because people will understand that IBM is not going anywhere. Oh and IBM has one of the biggest (actually it might be the biggest) patent portfolio in the US.

I think software patents are not all the gross evil that everyone sees them, and that certain well written ones are just fine especially in cases where a hardware or software implementation of a process can be used with equal efficiency. I do believe the problem with the US system lies in the fact that the addition was not made by lawmakers or the Patent Office, but by a panel of judges known as SCOTUS, actually the USPTO has tried their best to limit software patents and more particularly business method patents as best as they can without overstepping the SCOTUS ruling.

I just hope that you have a bit more of an open-mind then the zealots I see speaking most frequently on this topic, and have enough since to understand the need for rational debate and discussion and not just one-sided bashing.

Re:Competition and Collaboration

[Anonymous Coward]

I don't think it's caused by faith in competition, at least not in the traditional capitalist sense of producing something valuable for the lowest price. It's more like a lack of faith in this competition. The patent system discourages real competition, granting monopolies on production.

In this system, the successful corporation will be the one that most deftly competes for the government's protection -- not the one that most deftly competes to create the best product.

I think that real competition (for products, not for protection) is complementary to collaboration. Witness the widespread success of open-source software. The main threats to both collaboration and competition are government-enforced monopolies, created through measures like patents and the DMCA.