[Cache from http://perens.com/Articles/Patents.html; please use this canonical URL/source if possible.]

Software Patents vs. Free Software

Permission is granted to copy and distribute this document, and to incorporate it in full or partial form in a news or editorial presentation, modifying the formatting as appropriate. Don't modify it in a way that would make me appear to have a different opinion than what I wrote.

The opinions in this document are mine, not those of my employer.

There's been a lot of press about my activities regarding software patents and the summit meeting that I've called on Free Software and The Law. Since the press tends to simplify things and sometimes misinterprets them, it's time for me to talk about the situation of Software Patents and Free Software in my own words.

The original letters patent were the orders of a king. These early legal documents were often used to grant special privileges to the kings friends. Many letters patent granted a monopoly in a particular business to one family, forever. Anyone else who went into the same business would have been defying the orders of the king, so off with their heads!

The U.S. is important to this discussion because it's the cradle of software and business-method patents. Most other nations don't allow such things to be patented, although certain companies are lobbying to change that. U.S. law, being a descendent of English Common Law, has preserved the patent, though in a more limited sense than its royal application. The Constitution of the United States states:

Congress Shall Have Power To [...] promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries

This is the entire justification for the existence of the U.S. copyright and patent system. Both exist to promote the progress of science and the useful arts: technology, literature, and so on. If the patent and copyright system do not have the effect of promoting progress, they aren't constitutional. Note that the constitution also says limited times. Patents and copyrights are meant to expire - if they don't, they aren't constitutional.

So, we might consider whether or not software patents are actually promoting progress, or if they might even be hindering it. Surprisingly, there is no hard evidence that software and business method patents promote progress. There don't seem to be any good scientific studies on the issue.

The U.S. patent system rewards an inventor with a temporary monopoly on their invention, in exchange for full public disclosure of the invention. This is supposed to be an incentive for inventors to disclose their work instead of keeping it secret. The Free Software paradigm seems to be a better means than a patent to get people to disclose their ideas. The large body of free software developed in just the past few years shows us that Free Software has influenced many people to publish, and of course Free Software avoids the monopoly inherent in a patent. Perhaps that alone is sufficient reason for a critical re-examination of the idea of software patents.

When a patent expires, after 20 years, the invention goes into the public domain, which means that it's everyone's property. Thus, you can use the principle of an expired patent for free, you won't have to pay the inventor a royalty. You are only allowed to patent something that you really did invent, that hasn't been publicly used or disclosed before. In the U.S., you get a year to file your application after you create an invention. There are big penalties for lying on your application, for example, for falsifying the date of an invention.

Large companies like HP most often use patents defensively. Since they know that other companies will apply for patents and then sue for patent infringement, a company that wants to defend itself files its own patents to use against its competitors. This creates a détente between the company and its competitors - each could sue the other in a similar way, so neither one does. The competing companies generally enter a cross-licensing agreement, licensing all of each others patents. Each company thus negates the other company's power to sue over its patents. Large companies are generally cross-licensed with many other large companies, including their own competitors, and thus their power to sue each other over patent infringement is mostly nullified.

So, since the big companies patents all are used to nullify each other, the patents really have zero value to the companies. Why, then, do they bother with patents at all? I think the primary reason is a typical feature of détente: if you stop behaving offensively before the other guys do, you lose. If one large company stopped filing new patents, the other companies would no longer have a reason to cross-license with that company, and then the other companies would start filing patent-infringement lawsuits.

There are a few other reasons that large companies continue to file patents: some of them want to use them to fight a competitor who isn't cross-licensed, for example a Free Software developer who writes a competing program. Some of them use patents to give investors an impression that the company has value. However, the vast majority of software patents, some say as high as 95% of them, are actually invalid due to the existence of prior art. Prior art means that someone else has invented the same thing before, and has published or disclosed the invention in some way, thus invalidating all future patents on that invention. Because software patents are so often invalid, the investor is being deceived. When I've taken this up with companies who are looking for investment and list their patents as major assets, I usually get an answer like yes, we know that is a problem for software patents in general, but ours are enforcible. Yeah, right.

So, who profits from all of this if the large companies don't? Not the small and individual inventors, they can't afford to sue to collect on their own patents. The system does keep a lot of patent lawyers in business, costing companies Billions every year. It may also keep some research scientists in business, since companies sponsor research departments so that they can have a steady flow of inventions upon which to file patents.

Of course, getting companies to sponsor research is a good thing, and you might even think it justifies the patent system. However, wouldn't it be better if the research was targeted to benefit the company and its customers rather than to provide ammunition for an eternal stalemate of a patent war? And wouldn't it be better if the inventions of publicly-funded research organizations like universities were in the public domain, since the public has paid for them? Instead, they are tied up with patents and sold. Some universities claim that they need to be able to sell that patent monopoly, or they won't be able to entice anyone to develop their ideas commercially. Although this is a dubious claim, a U.S. law actually mandates patenting as part of the "technology transfer" of publicly-funded research.

A trend over the past decade has been a new kind of patent parasite company that exists only to sue other companies. The parasite buys up patents and then files infringement lawsuits against other companies with deep pockets and a lot to lose. The parasite offers licensing terms of a few tens or hundreds of thousands of dollars, versus a cost of Millions for the victim to successfully defend themselves in a lawsuit. The victim generally caves in and pays the license fee, even though they could eventually have proven the patent to be invalid in court. So, this is another tremendous drain of company funds caused by the patent system. Parasites don't enter cross-licensing agreements, so they represent one of the worst offensive uses of patents.

The cross-licensing defense works well if you have a lot of patents to use as ammunition, as a large company might, but the Free Software community has only a handful of patents held by a few individual developers, and thus ends up being an innocent bystander injured by a war of giants. Since it costs tens or hundreds of thousands of dollars in legal fees to successfully file an enforcible patent, we don't expect to get many more of them. So, scratch the cross-licensing defense for now.

Since most software and business method patents are actually invalid, why don't we just prove that in court? It often takes Millions of dollars and years of time to win a single patent case. I don't know many Free Software developers who could afford to fight even one patent, and we are faced with hundreds of thousands of them.

Then, why don't we just "fix" the patent system? Well, the best way of fixing it might simply be to dispose of software and business method patents entirely, but most of the proposals to "fix" the patent system simply try to make the problem only a little bit better. For example, there are plans to provide patent examiners with more prior art so that they will reject more of the invalid patents, but not nearly enough to solve the problem. Such proposals can only be treated as interim strategies, poor defenses to tide us over until we can really solve the problem.

Arguments for killing software and business-method patents abound. I'll list a few:

• If the existence of software and business system patents doesn't actually help to advance science and the useful arts, then they are constitutionally unjustified.
• Software patents are all on applications of an extremely versatile hardware device called a "computer". But then isn't the computer, not its software, the thing that should have been patented?
• The 20-year term of software patents is generally much longer than the useful lifetime of the invention. This circumvents the constitutional requirement that patents have a limited term, because the invention is so obsolete as to be without any remaining value at all by the time it reaches the public domain.
• The monopoly that business-method patents create on a particular form of business can constitute a violation of the anti-trust laws.
• If other countries start accepting software and business method patents, they will in effect be creating a U.S.A. Tax paid by their own citizens. Those countries would just take all of the patent royalties paid by their own companies and send them to the U.S.A., where the software and business-method patent owners predominantly are today.

Ironically, some of the biggest patent holders are the Free Software Community's own partners, companies like IBM and HP that have aggressively incorporated GNU/Linux into their business plans and expect significant revenue from it before long. IBM is said to hold 10% of software patents, and HP is one of the largest patent holders in general. It's important for us to start a dialogue with these and other partners. That's why I am calling a summit meeting on Free Software and The Law. Patents will be a full-day topic, with DMCA, UCITA, license law, and other legal issues covered on a subsequent day. Unfortunately, this isn't a public meeting. When asked about that, Richard Stallman said inviting a public audience to a negotiation isn't the best way to get an agreement. However, I've invited a broad slate of Free Software representatives, covering all of our various viewpoints, and will invite a few more.

The meeting will simply be a discussion and a negotiation, none of it is set in stone, but there are some things that the Free Software community representatives might be likely to ask for. For example, we might ask for some assurance that our corporate partners aren't going to sue us. When Eric Raymond and I asked this of an IBM representative recently, the answer was we're not prosecuting our patents against Open Source developers. HP hasn't sued any of us either, although they haven't stated their policy in so many words. But perhaps the Free Software developers, enthusiastic partners of IBM, HP, and other companies, deserve a little more formal assurance of that policy and its continuation.

Another issue we might discuss is how to defend a Free Software developer when he or she is sued for patent infringement. It might be fair to ask our partners, who have much deeper pockets than ours, to help vest a fund for our defense with an organization like FSF or EFF. Or perhaps we should deal with each lawsuit as it comes up? With the DeCSS lawsuits, we found that EFF can defend less than one case per year. We'll need more than that.

Since our software has proven so useful and effective in our partner companies, I think the Free Software community would be negotiating from a position of strength. We've proven our worth to our partners, and now we deserve to understand how far they'll go for us. Hopefully, that will come out in the summit discussion.

So, you can see that we have a lot to talk about, both at the summit and in the Free Software community in general. The summit is being held on August 31 and September 1, in San Francisco, immediately after the LinuxWorld conference there. Thus, many of the key players will be in town. HP has generously promised enough funds to provide the meeting facilities and catering, and to cover travel and lodging of some free software representatives who can not pay their own way. HP says they are open to having equal co-sponsors, so that the conference won't appear to be predominantly HP's. SSC has promised to cover travel and lodging expenses for Richard Stallman, and other co-sponsors will probably step up before the meeting.

I hope I've cleared up some mis-perception of what I've been working on, and if this essay has been educational for you, I consider that an achievement in itself. Of course, I'm open to your ideas. If you have comments or questions, please feel free to write me at bruce@perens.com or use the phone number listed on my web site.

Many Thanks

Bruce Perens