Wednesday, June 02, 2004

On the Uses and Abuses of Patents

Via Slashdot, yet another example of the egregious abuse of a government-granted monopoly.

Network Associates Inc. (NAI) has been granted a broad U.S. patent for technology covering "various computer program products, systems and methods" for filtering unwanted e-mail messages, it said Tuesday.

The antivirus software company said that U.S. patent no. 6,732,157 encompasses use of multiple spam-filtering techniques such as compound filters, paragraph hashing, and Bayes rules. These techniques have been proposed and used by a number of antispam technology proponents and it was not clear from the filing how the patent would affect competitors to NAI's McAfee antispam offerings.

"To me this looks like a pretty broad patent," said Rob Tosti, partner in the Patent and Intellectual Property Practice Group of Testa, Hurwitz & Thibeault, LLP in Boston.

Tosti said that using Bayes Rules to calculate the probability of a message being spam was an idea that began to gain momentum in mid-2002. And although NAI applied for the patent in December of 2002, the granted patent presumes that the company has a valid claim.
That this patent is a clearcut case of the USPTO not doing its job right is attested by the following passage from the Slashdot article.
The patent covers "compound filters, paragraph hashing, and Bayes rules" and was filed in December of 2002 ... Paul Graham's "A Plan for Spam" was published August 2002."
One certainly has to give Network Associates' management credit for having serious chutzpah, if for nothing else. There's more than enough evidence out there to shred this patent apart should they ever have the gumption to attempt to enforce it, but IP litigation is hardly cheap, especially when one is going up against a company with pockets as deep as this one. In the meantime, there's no telling just how much havoc this nonsensical patent grant will cause.

Unlike a lot of fans of open-source/free software, I'm sympathetic to the case for software patents, as the last thing one wants is to undergo tremendous sacrifices to develop a great new idea just so BigCo can come along and rip off one's ideas for nothing; even so, I think that this is one field in which traditional patent procedures simply cannot be expected to work without causing difficulties so severe they outweigh the likely benefits. The 20 year patent grant that is standard in other fields is almost certainly much too long in a field like this one, while the USPTO's lackadaisical attitude towards searching for prior art is simply unacceptable. As for "business method" patents, I'd argue that those should simply be abolished outright: only innovative technical ideas ought to be worth patenting, not something as obvious as, say, enabling users to purchase items with a single click ...