Comment: Who benefits from IBM and the USPTO's 'open' bid to up patent quality?

Posted on 12 Jan 2006 at 12:14

The US Patent and Trademark Office's announcement yesterday that it is to team up with major open-source vendors, the Open Source Development Labs and VA Software's SourceForge.net, to launch initiatives to improve patent quality, looks shiny and positive on the face of it.

Open Critics

The free and open-source software (FOSS) community has been one of the strongest critics of software patents, and in particular some of the more dubious patents issued by the USPTO. So any involvement it can get in fixing things at the USPTO is to be welcomed.

But the FOSS community is not just critical of poor patents. It is critical of software patents full stop. Indeed the GNU General Public License - the keystone licence under which the Linux operating system is destributed - prohibits the inclusion of patented code which places restrictions on it use - financial or otherwise.

When a draft EU Directive that would harmonise patent law across Europe came up for debate, the FOSS community was at the vanguard of opinion that the directive should formally outlaw the patenting of software.

But with these announcements from the USPTO, this same community, with its passionate hatred of software patents, will be in a privileged position to submit code to the USPTO and review patents under some brand new schemes.

Too much information

For example, the USPTO's Open Patent Review is an alert service that tells subscribers of patent applications within a specified field, and will enable subscribers 'to provide feedback to the patent office on existing prior art that may not have been discovered by the applicant or examiner', according to IBM's press release.

Greg Aharonian, who runs the patnews service, reacted angrily to the announcement. 'The alert service, is already offered by others. Anyone really caring about new issued patents can do much of this themselves,' he told us. 'Many in the open source community hate patents to the extent that they never bothered to learn much about the mechanics of patent examination. How they hell are they going to be able to provide pertinent information? Heck, most of them still don't know what a priority date is for prior art. Second, unless this is well managed, it will create more problems than it solves, in that it will create more questionable information [that] someone at the PTO has to sift through to see if there is anything of value.'

A USPTO spokesperson, however, responded: 'Developers and innovators of any technology (inventors generally) tend to know best what is known about their art. To that extent, they can point to where they looked to deal with the problems they have tried to solve. But inventors per se are generally not well versed in how to apply prior art per 35 USC 102 and 103, or other statutes.'

But there's evidence that making any more information available will be at best futile, or even detrimental to patent quality if more of the USPTO's resources are spent in assessing a flood of material submitted as prior art. In spite of an existing database with more than 4m computing abstracts, in the 10 years between 1994 and 2004, the average number of non-patent prior art examples cited per patent never rose above two in many of the software patent classifications. And in 2004, 57 per cent cited none at all.

Given the open-source community's anti-software patent stance, this is particularly relevant, as much of the open-source technology submitted as prior art may not be patented in the first place. Yet it doesn't look likely that there will be much interest in it anyway from the patent examiners, given their track record.