For more than four years, Microsoft has claimed that
Linux infringes its patents. True or false? Here’s a clue:
Have you ever seen the Boogeyman?
Steve Ballmer insists that Linux infringes Microsoft patents and that Microsoft’s agreement with Novell proves his point. Meanwhile, Novell insists that Linux does not infringe Microsoft patents and that its agreement with Microsoft was prompted purely by logical, business cases. Like some high-tech plea agreement, Novell states:
Patent concerns did not drive our entry into this agreement. Novell makes no admission that its Linux and open source offerings infringe on any other parties’ patents. Our position has not changed as a result of this agreement.
So where does the truth lay? No one knows for sure. Well, perhaps Microsoft believes it knows, but the company has been making the same claims for more than four years without presenting any proof. Can someone say “SCO?” Ignoring Microsoft, what’s the deal with software patents? How big a threat do they pose to open source? And is that threat any different than other other risk posed by closed-source code?
By my count, there are in excess of 230,000 issued and active “software” patents in the United States today. (Although a true assessment of what constitutes a software patent would require a more detailed investigation, my estimate is based on the number of issued and active U.S. patents listed in patent classes 345 and 700-725. Operating system software, for example, would likely fall into one or more U.S. patent classes that are included in this group.) The largest holders of “software” patents are, in descending order, IBM, Hewlett-Packard, Hitachi, Intel, Microsoft, Fujitsu, Sony, NEC, Sun Microsystems, and Mitsubishi.
When you look at that group of 230,000 “software” patents, it is fair to say, as a general rule, that no one company’s patent holdings are more likely or less likely to tread on Linux than the next company’s. That is to say, it is just as likely an equivalent percentage of IBM patents tread on Linux as HP patents, Microsoft patents, or NEC patents.
When you look at the patent landscape through this lens, the “threat” posed by patents against Linux is proportional across the board, and the “threat” posed by any one patent holder is equal to the ratio of their total “software” patents to all “software” patents. Hence, measured on that basis, Microsoft holds only 2 percent of U.S. “software” patents, so one can presume that Microsoft’s patents represent only 2 percent of the potential threat. By contrast, the six companies that are members of the Open Invention Network, — IBM, Sony, Philips, Novell, Red Hat, and NEC — represent more than 13 percent of U.S. “software” patents, and they pose no threat to Linux at all.
With 15 percent of all patents owned by ten companies, the remaining 85 percent of U.S. “software” patents lay in the hands of other parties. Now, many of those other parties have no particular reason to want to threaten Linux, as they have a strong working relationship with the open source community. This includes companies like Toshiba, Motorola, and AMD, all large “software” patent holders.
Instead, the real threat to any software producer, whether open source or closed-source, comes from those patent holders who don’t actually produce anything and are only looking to extract royalty income and license fees from the actual producers. Some refer to these entities as “patent trolls,” and the bigger you are, the more of a nuisance the trolls can become. Just ask Microsoft: At any point in time, Redmond’s attorneys are defending more patent infringement suits than any other company in the software industry.
There is little to be done about these non-producing entities, and the more commercially successful open source software becomes, the more likely companies producing and distributing such software will come under attack from non-producing entities. Patent trolls are a simple fact of life.
But the landscape isn’t altogether as bleak as it may seem. Although courts generally accord patents a presumption of being valid, the landscape for “software” patents is a bit different. Almost two-thirds of the active “software” patents (and about 60 percent of Microsoft’s “software” patents) were issued between 1996 and 2004, a period when the United States Patent and Trademark Office (USPTO) was suffering from funding and staffing shortages, and was simply not prepared to handle the massive influx of applications covering software and business methods. Hence, the quality of the “software” patents issued during that time tends to be more suspect, and those “software” patents are far more likely than patents in other areas of art to be found overly broad (and subject to having their claims narrowed on reexamination) or altogether invalid.
Finally, there is the issue of the threat posed by such patents to either individual developers or end-users. Individual developers, those not working for a substantial company, are not attractive targets for a patent holder. There is little likelihood of a substantial damage award in such cases, and even if you enjoin one developer, such an injunction would not necessarily apply to other developers. Going after individual developers would be a bit like the game “Whack-A-Mole.” It’s what makes the Microsoft promise seem so hollow.
An end-user doesn’t make an attractive target either — for the very reasons SCO learned: You don’t want to sue your own customer for patent infringement. It’d bad for sales.
Of course, open source developers shouldn’t trivialize patents. At the same time, the reality of the patent landscape is not as fearsome as some would like you to believe.