Microsoft patent application 11/296194 suggests a method for distributing music files that ensures copyright holders get their royalties, and started out the way most patents do. An application was filed to the US Patent and Trademark Office in December 2005, and sat there, unresolved, for more than a year. But last week, 11/296194 took a new path. The application was posted on a website, and visitors were asked to submit opinions and evidence to answer the question: is it new?

R. J. Spalding's 1889 patent for a flying machine: arguments over prior art haven't got any simpler since. Credit: BETTMANN/CORBIS

Within two days of posting the patent, there were two responses. “How could anyone consider this non-obvious?” wrote a computer technician from Liberty, Montana. A Hewlett-Packard software engineer disagreed: “I believe the application has merit.”

Welcome to 'Peer to Patent', the patent office's pilot project launched on 15 June to bring patent evaluation to the masses (see http://www.peertopatent.org). The debate over the Microsoft patent is just beginning; registered visitors to the site will have 16 weeks to submit comments. The project aims to help the overstretched, backlogged and beleaguered US patent office search for 'prior art' — evidence that a patent's claims have already been patented or are already in common use. The hope is that tapping the expertise of hundreds of reviewers will improve the rate of prior-art discovery, speeding up the process for patent examiners while creating solid patents more likely to withstand future litigation.

Beth Noveck: wants to open up patent decision-making.

Peer to Patent is a pet project of New York Law School's Beth Noveck, who sees it as a step towards bringing an open-source approach to otherwise closed government decisions. The project has now been in operation for five weeks and already has 1,000 registered users, eight patents under consideration and more than 30 submissions of evidence showing prior art. It has well-heeled sponsors, including General Electric, IBM, Microsoft and Hewlett-Packard, who hope it can help to unclog the process for granting information-technology-related patents. Peer to Patent has also piqued the interest of the UK Intellectual Property Office, which plans to launch a similar programme in the next year or so, Noveck says.

John Doll, the US commissioner for patents, says he is optimistic about scaling up the programme, and sees expansion to include the biotechnology sector as a logical next step. “That would be a natural outgrowth,” he says. “In the sciences you'll see a lot of interest.”

Observers say that Peer to Patent has got off to an auspicious start. But although most welcome the programme as a sign that the US patent office is open to reform, some wonder how much of an impact it can really have. “It is an interesting experiment that seems to me worth trying — although I would be very surprised if it proved to be robust enough to affect the patent system more than marginally,” says Dan Burk, a specialist in patent law at the University of Minnesota in Minneapolis.

The United States has traditionally opposed opening patent examinations to the public, citing the possibility of influence from competing interests or of overburdening the patent examiner with irrelevant claims, says Stephen Kunin, a former deputy commissioner at the US patent office, now at Oblon, Spivak, an intellectual-property firm based in Alexandria, Virginia. The Peer to Patent programme has built in several safeguards against petty interference in the process, however.

Suggestions of prior art must be accompanied by solid evidence, and the registered users are asked to agree to or reject a submission's addition to a top-ten list. Only the ten strongest suggestions of prior art are then forwarded to the examiner at the patent office. The applicant can view comments, make preliminary amendments and ask for an interview with the examiner before action is taken on the application.

For now, the programme is limited to 250 computer software and hardware patent applications. Some worry that it will be difficult to scale it up to handle a more significant caseload. “If it works for 100 patents, can we expand it to anything like a major subset of the 400,000 filed each year?” asks Mark Lemley, director of the law, science and technology programme at Stanford University, California.

Others note that biotechnology or pharmaceutical companies are far less enthusiastic about speeding up the patent process than are computer companies (see Nature 437, 1230; 2005), and doubt if they will embrace Peer to Patent. Biotech companies, in particular, rely heavily on their patents in their business model, and have been resistant to radical changes to the patent system. “I would suspect they would be much more reluctant to participate,” says Arti Rai, a law professor at Duke University in Durham, North Carolina.

It will take time to establish if the pilot project is speeding up approval and improving patent quality. Will the skills of hundreds of unpaid Peer to Patent volunteers match those of patent litigators once a patent is approved? “It's when somebody's willing to spend $2 million to $3 million on a lawsuit that people get really good at finding prior art,” observes Burk.