Patches to Interactive Week patent article

It's always fun to open up lwn and find a link to article about yourself near the top.

This article contains a number of small bugs. None of the quotes are verbatim. In the spirit of free software, this document contains patches to these bugs. Original text is in italics.

Raph Levien is the model of an independent inventor. The Berkeley, Calif., software developer says he was probably 10 years old when he filed for his first patent.

I live in Benicia now. I was 10 when I sent the disclosure for my first patent to the patent attorney. The patent wasn't actually filed until I was 12.

Levien's patent on a technique for protecting memory in computer systems from crashes turned out to be quite a success. He licensed it and started a career of dreaming up ideas and locking them away with patents.

The patent was a simple way to add virtual-memory like memory access control to small CPU's like the 8085. It was primarily intended as a way to ship systems with secrets embedded in them. Protection from crashes wasn't a goal.

"I've gotten a substantial amount of money for my patent portfolio over the years," he says. "It's been a very reasonable income to pay myself."

Not a verbatim quote - what I said was more along the lines of "the patent royalties over the years have been a very reasonable income".

So it was something of a surprise when Levien posted a notice on his Web site earlier this year, offering a free license to anyone who uses his patented ideas in software protected by the Gnu Public License. In other words, anyone who shares "open source" or free software doesn't have to worry about Levien suing for patent infringement.

Wrong. The patent grant is for GPL software only. If you want to release software under a non-GPL open source license, I'd love to have a discussion with you about royalty payments.

Also, it shouldn't have come as a surprise to people who were following what I was up to. I'd been helping out free software projects on patent issues for a while, including analyzing patents. I'd also announced that I was going to do this at Linux World Expo last spring :)

The news was greeted with warm applause by the open source community, a loose-knit coalition of people who deliberately share the source code to their software. The open source software world has been worrying about patents, because patents give people the legal right to stop others from freely using an idea or process.

Levien's solution marks just one approach that is emerging from the open source community to address patent issues, though most are still in the planning stages. Some people are advocating the creation of "patent pools" that would cross-license patents with everyone, while others talk about creating "mutual disarmament" agreements to reduce the legal complications of patents. Then there are those pushing for tighter scrutiny of the patent process in the first place.

Including myself. The examination process for software patents is a sham - probably half of the patents I've read are clearly invalid to anybody who knows the art.

Open source developers are concerned that Microsoft or another company might develop a patent on a fundamental Internet protocol or mechanism, effectively preventing other software developers from incorporating it. In some cases, patents can be exceptionally powerful and constrain how everyone uses the Web. Unisys, for instance, owns a patent on a method for compressing data that is incorporated into the GIF image format. In recent years, the company has been trying to enforce the patent - which potentially affects everything from Web browsers to Web sites that use GIF images.

"Might" is of course putting this far too mildly. There are a large number of patents right now that pose threats to free software projects. Many are invalid but would require an expensive fight to prove that. Others are likely to be valid under current patent law. Either way, free software loses. I believe that the public interest as a whole also loses.

Traditional patent licenses, which are negotiated between corporations, often return a percentage of the fee to the inventor. But royalty agreements like these just aren't compatible with a system that requires everyone to share their source code freely.

In patents assigned to corporations, the inventor usually doesn't see a dime of the royalties. Many corporations do have incentive programs to reward employees for filing patents, however.

In any case, it's important to separate out economic constraints from fundamental incompatibilities. The traditional model of free software is volunteer and unfunded. In this model, there simply aren't the resources for developers to license patents, fight bad patents, or even in most cases to file their own patents.

But that's only the traditional model. There are now a lot of businesses whose fortunes are tied to free software. As a concrete example, Red Hat ships TrueType rasterizers with their distributions. Apple owns at least two patents that seem to cover TrueType rasterizers. I think it would be entirely reasonable for Apple to ask Red Hat for royalties, and for Red Hat to pay them.

IBM is one of the largest, if not the largest, holders of patents in the world. They also have a substantial commitment to Linux and other free software technologies. If someone came after the Linux kernel with a patent infringement claim, IBM could choose to settle the claim by cross-licensing some of the patents in their vast portfolio, with no money changing hands. Such a deal could be consistent both with free software and patent practice.

In most cases, economics weigh against nice patent licences for free software projects - in fact, I can't think of any such deals which have been successfully executed. But this is a separate issue from fundamental incompatibility.

Some companies avoid payments by swapping patents. However, the wide-open cooperation of free software development makes it difficult, if not impossible, for the community to generate its own patents, says Eben Moglen, a law professor at Columbia University and the volunteer general counsel at the Free Software Foundation. The organization is responsible for the creation of the Gnu Public License.

I can only hope Prof. Moglen was misquoted here. It's hard for me to see how cooperation makes it difficult to produce patents.

Richard Stallman, one of the intellectual leaders of the open source movement, has been trying to organize a patent pool that would operate similarly to the Gnu Public License. If you used patented ideas from the pool, you would need to contribute your own patents to the pool. This system would draw in more and more patents, he says - but only if it had good enough bait at the beginning.

Richard Stallman is, of course, the founder and a prominent leader of the modern free software movement, not the open source movement. There is a difference.

"It is hard to get such a pool started," Stallman says. "When it is small, it puts little pressure on anyone to contribute patents to the pool. I was never able to find anyone with patents who actually wanted to contribute."

Mark Shewmaker, an open source programmer and devotee, is also trying to create a patent pool through his Open Patents License. He says one of his greatest challenges is making sure that the pool does not run afoul of antitrust law designed to prevent companies from pooling their patents to create a monopoly.

The goal of Open Patents is to "create a cross-license agreement in which participants can use each other's patents," Shewmaker says. "Businesses and individuals can use patents from any specific patent pool, but only when they meet the nonmonopolistic requirements inherent in that pool."

At this point, Shewmaker's Open Patents License is still being reviewed and reformulated, as different patent attorneys offer suggestions.

Meanwhile, Eric Raymond, another open source advocate, has been trying to push what he calls the "Chinese finger trap proposal." Patent licensees could be locked into protecting open source projects. Raymond proposes that some major company with both a strong portfolio of patents and an abiding interest in open source software prohibit its licensees from suing open source developers. That is, if a company like IBM adopted such a policy and third parties wanted to license its patents, they must forgo enforcing their own patents when open source projects use the ideas.

So far, however, his idea hasn't met with any takers.

On a different front, others are trying to reform the patent system itself. Here, the free software community has found a number of like-minded allies. Congress has signaled that it wants to consider how the patent system might be adjusted to prevent overly broad or not particularly novel patents.

"In the future, we're going to see an organized attempt to make it possible to challenge patents on a 'lay-down' kind of basis," Columbia's Moglen predicts.

Some sites have cropped up that allow people to post papers, articles or products that might prove that someone thought of an idea before. The recently launched BountyQuest, for example, offers cash rewards to people who discover prior art - evidence that shows that a patent is either not new or is obvious, which means it's not eligible for patent protection. Businesses and individuals with an interest in invalidating a patent can post the bounty.

But there's no simple answer to the problem of using patents in open source software. Levien recognizes the paradox: On one hand, he made money from forcing everyone who used his patented ideas to give him royalties. On the other, he shared the source code of several programs and recognized how the cooperation helped him and others. The two models were in conflict. Levien asked himself, "How can I reconcile my patents with my open source development?" The result was the royalty-free license for those who also share their source code.

s/open source/free software/ s/license/grant/ s/.$/ under the GPL./

For now, it seems that kind of generosity is the best the open source community can hope for.

Well, there are lots of other things that can be done as well. First, when faced with a choice between patented and unpatented technology, choose unpatented. If people had been more vigilant about this, the entire GIF fiasco could have been avoided. There are a number of analogous issues before us today, including MP3 vs Vorbis, the development of a high quality autohinter to replace proprietary manual hinting techniques such as TrueType, and others.

Respond to requests for public input when possible, including the recent consultation of the UK patent office.

The spirit of free software is not merely about benefitting from the generosity of others. Actively doing your part to make things better is an integral part of the process. As free software meets today's consumer culture, many people lose sight of that. Don't be one of them.