US patent mess will get worse before it gets better
Little chance of arresting ludicrous land-grabs
Analysis The discredited US business patent issuing machine is being tweaked, but the signs are that it will not bring about the reform needed. The action plan of the US Patent and Trademark Office's new business methods patent initiative has words of comfort about industry outreach and quality issues, but it is failing to impress those at the patents coal face who know that the USPTO will continue to use its rubber stamp too liberally. The new USPTO programme will introduce a second pair of eyes to give an additional layer of review by a senior examiner. In addition, the guidelines for computer patents will be revised in April. Supposedly, more attention will be paid to prior art - patent jargon for earlier technology and patents related to the application - since in very many applications, there is nothing novel. Apart from silly US patent case law, the situation is made worse by the ponderous USPTO, which has never managed to keep current on technology when that is the essence of its business. When the first software patent applications began to be filed, hardware experts were used because the agency had no software expertise. The situation has not improved greatly. What has happened is that patents for business processes that are neither new or unique are being granted by a gung-ho USPTO that only gives a cursory glance at them before waving them through. Anything man-made The problems for the IT industry begin with the US constitution, which allows patents to be granted to "new, useful, non-obvious" inventions in one of four categories: "process, machine, manufacture, or composition of matter", with the result that the US Supreme Court has decided that patentable subject matter covers "anything under the sun that is made by man". Business methods were excluded until the July 1998 US Federal Circuit Court of Appeals decision in the State Street Bank versus Signature case, where the court found that a computerised method of administering mutual funds was valid, and that it was not necessary for the program to be tied to a physical invention. Another case in April 1999 that had the effect of cementing the decision was AT&T versus Excel over a patent for call message recording for telephone systems, where the same court (which is the specialised appellate court for patent cases) reversed the district court's finding for Excel. In 1998, 700 software patent applications were made that covered financial or business methods, with 300 being concerned with the Internet - so-called cyberpatents. Previously, copyright law had been used to protect computer programs. A few prominent instances of patents taking advantage of these extraordinary decisions are well-known, but there are now thousands of US patents granted or being applied for in a latter-day gold rush. The Amazon patent on a one-click checkout was defended against Barnes & Noble; Bruce Dickens' patent on a work-around for Y2K date changing is fortunately being re-examined; the GeoWorks' wireless data protocol patent is being enforced; and the Priceline reverse-auction patent is being vigorously defended. There are also patents that are plainly based on fantasy, like the hyper-light speed antenna. The international angle The USPTO will be arranging a meeting about e-commerce patents, probably in July, which is a good move if suggestions made are actioned. But before this happens, there is an important Diplomatic Conference for the adoption of the draft Patent Law Treaty at WIPO in Geneva from 11 May to 2 June. The text of the basic proposal is here. Article 6 precludes any national patent office from imposing different requirements to those under the Patent Cooperation Treaty. This concerns the USPTO, but it remains to be seen whether it will have any significant effect on cyberpatent applications. Only by having much better informed examiners is there any hope of improving the system. Whatever changes in patent application and examining procedures come about, the problem remains that a significant number of US patents have been granted in breach of the examining rules. The system is broken, and duff patents have to be removed, by challenging and re-examining each one. It is most unwise that the compensation package for patent examiners is based on how many patents they review. Fortunately, the new USPTO initiative does include provision for enhanced technical training. A change of US law is not needed: the solution is to apply the law. A problem that has yet to be faced is that in quite a few cases patents have been granted for business processes where there is a previous implementation that could not have been patented at the time, since it was before the State Street Bank case. If the USPTO does not start its own re-examination process, those who actually invented the earlier business process should produce the prior art to invalidate the relevant opportunistic patent, and send it to the USPTO. Business procedures used for years on Minitel (for example the shopping cart) have subsequently been patented for the Internet. The USPTO has just become the second Performance-Based Organisation under a concept introduced by Vice President "Mr-Internet" Gore and the National Partnership for Reinventing Government (really). A PBO is supposed to be accountable for results by having clear objectives, specific measurable goals, customer service standards and targets for improved performance. These need to be set out in the public domain, with Commissioner Todd Dickinson's and other senior jobs being on the line if the improvements are not substantially met. Maybe Greg Aharonian, whose Patnews never fails to entertain, should be put in charge of a USPTO quality control programme. At present, Congress is considering a bill to allow the USPTO to keep the fees it gets from applicants. This should give it the money to deal with the mess it has created by failing to reject applications that should never have been granted. For a start, salaries should be increased to attract more and better people. The Register has its own suggestions for sorting out patents mess: Cyberpatents should be limited to around three years, rather than the 17 years they currently enjoy, because they become outdated very quickly. The re-examination of duff patents, and the opponents' legal costs, should be paid by the patent holder if the patent is not upheld on re-examination. Public comment should be solicited via the Internet before a patent is granted. Business process patents should not be recognised outside the US, while the mess is being sorted out. ®
Sponsored: RAID: End of an era?