Electronic patent databases invent difficulties

Found on New Scientist on Saturday, 02 April 2005
Browse Various

If reinventing the wheel is ridiculous, being able to patent such a device is even more so. Yet that is what happened in 2001 when patent attorney John Keogh won a patent for a "circular transportation facilitation device" from the patent office in Australia. He did this to flag up deficiencies in the country’s new patenting system.

To qualify for a patent, an invention is supposed to be new, useful and non-obvious. Patent examiners establish novelty by making sure the invention is not either already patented, or in use commercially without patent protection.

They do this by checking for so-called "prior art" in existing patent applications and granted patents, as well as looking for non-patent prior art in sources such as trade magazines, websites and journals. However, examiners – particularly in the US – are often criticised for not checking thoroughly enough for non-patent prior art.

Univentio, the patent-information company he runs in the Netherlands, has discovered that Espacenet, the European online patent database, is missing 322,000 UK Patent Office documents, plus 186,000 and 17,000 patents respectively from the French and German offices.

The Leeds Patent Information Unit, the largest regional library of its type in the UK, was unable to relocate a large percentage of its hard-copy collection when it moved offices in 2004, says Lagemaat, while the Dutch patent office has already begun disposing of its hard copies of patents.

This looks pretty messed up; patents get lost or are destroyed before they are digitised. The granting process itself is also messy: "prior-art" and "non-obvious" was never heard. How else could have Keogh patented the wheel? What about patents for the the "y-axis", "browser navigation" or "sudo"? There exists prior-art (and all that is pretty obvious too).