Slashdot Mirror
Federal Circuit Appeals Court Limits Business-Method Patents
Zordak writes "The Court of Appeals for the Federal Circuit has just issued its much-anticipated opinion in In Re Bilski [PDF]. This was a re-visit of the State Street issue of what constitutes patentable subject matter (including whether software and business methods are patentable). In summary, the court has affirmed and strengthened the 'machine-or-transformation' test, upholding the patent office's rejection of claims on a method for hedging risk in the field of commodities trading. Although the court refused to exclude software patents categorically, it is likely that the reasoning of this decision will be used to reject many software patents (note that some of the dissenting judges would have completely overturned State Street and tossed out all software and business method patents). Although not as sweeping as some had hoped for, it is certain that this decision, along with the Supreme Court's KSR decision last year, will lay a difficult mine field for those who want to patent software and business methods."
The problem is that Congress decided not to fund the USPTO and allow it to fund itself through application fees, processing fees, and all whatever other fees the USPTO wants to charge. So, the more patents they review and grant, the more money they make. This was the primary cause of the patent explosion - it suddenly became in the Patent and Trademark Office's best interests to allow and grant as many patents as possible. Take the $500 application fee (or whatever it is), charge a review fee, charge a patent issuance fee, and leave it up to the courts sort out whether or not a patent on a method of removing snot from a nose with a finger is a valid patent or not. That's what's been going on. My solution is for congress to fund the USPTO, but that they should only be allowed to issue, say, 100 patents per year. All patent applications are held and reviewed and at the end of the year, the top 100 are granted patents. Frankly, 100 is too generous, I'd rather it be 20 to 30. These days there are nowhere close to 100 truly innovative and nonobvious inventions per year. Maybe 5 or 6, at the most. So 20 patents per year is more than adequate, and 100 is allowing crap that shouldn't otherwise be patentable. Once new patents become a rarity, the number of patent applications will drop exponentially. It should not be a valid business model to own a portfolio of ridiculous patents and sit around waiting for someone to infringe them. That does not foster innovation - quite the opposite.
So, give the USPTO $100,000 or so in federal funding, grant 100 patents per year (a board can pick the top 100), and by all means don't give the USPTO financial incentives to grant as many patents as it can, only to leave the actual work of determining patent validity up to the judiciary. Our courts are clogged enough without people suing over patent infringement for nosepicking methods.
Stupid people make stupid things profitable.
I recently took an IP law class (taught by a patent attorney) and he mentioned this fact. He also said that the best way to get a patent through is to contact the patent reviewer assigned to your application near the end of the month, agree to a rejection on specific grounds (even if they are minor, simply to help them fill their quota), and then resubmit the next month with a response to those issues. The reviewer assigned to your case is always the same, so they will recognize your application and approve it on its second time through as a quid pro quo. So yes, there's no denying there are issues and opportunities for corruption within the approval process, but it's actually more natural for the reviewers to reject rather than approve to make a dent in their caseload. Also, the grounds for objection (obviousness, for example) can be very subjective so it is possible to reject many patents based on a combination of several pieces of prior art and the assumption that it could have been discovered or invented by anyone with ordinary skill in the art, even if this might be unlikely in reality.
I'm always positive; it's my nature.
I'm not proposing eliminating software patents, so I don't need to provide such a line. I don't actually have a problem with software patents, so long as they are actually new, useful, nonobvious, and accompanied by a written description that adequately discloses the invention. The problem is not software patents: the problem is bad patents, and I can make several recommendations for cutting down on those.
First, the presumption of patent validity should be eliminated, making it easier to overturn bad patents. The large numbers of patents that are invalidated shows that the presumption of validity no longer makes sense.
Second, the patent office should make it easier for those with computer science backgrounds to become patent agents and patent attorneys. By bringing in more people educated in the field, better patents will be written and bad patents will be more easily overturned.
Third, courts and examiners should (and have started to) make more of the fact that all claims (even, or perhaps especially, broad ones) must be backed up by disclosure in the specification. As an example, if I disclose only one compression algorithm in my specification, then I should not be able to claim something as broad as "compressing data by eliminating redundant information." Instead, I should only be able to claim the algorithm I disclosed. Too many overly broad patents are granted, and this is especially common among software patents because the examining corps doesn't have enough competent computer scientists to weed out the chaff.