IBM Breaks Patent Record, Wants Reform 130
An anonymous reader writes "IBM set the record for most patents granted in a year for 2006. At the same time, IBM points out that small companies earn more patents per capita than larger enterprises and pushes for reform to address shortcomings in the process of patenting business methods:
'The prevalence of patent applications that are of low quality or poorly written have led to backlogs of historic proportions, and the granting of patents protecting ideas that are not new, are overly broad, or obvious.' And the company has been committing itself to a new patent policy: 'Key tenets of the policy are that patent quality is the responsibility of the applicant; that patent applications should be open to public examination and that patent ownership should be transparent; and that business methods without technical content should not be patentable.'"
How about? (Score:3, Interesting)
It's easy when you're on top (Score:4, Interesting)
I'm glad they're being reasonable about the advice they're giving, especially about "not patenting a business process with no technical content", but they're already so far ahead that they can afford to be reasonable. I don't see patent reform on the horizon as a result; IBM has too much invested in the current system to take a really progressive stance on "Intellectual Property" ownership.
mandelbr0t
Re:Oh, I get it (Score:5, Interesting)
Re:Ability to revoke patents? (Score:5, Interesting)
http://www.uspto.gov/web/offices/pac/mpep/documen
The whole system is fairly transparent and the new proposed peer review system would be a great opportunity for you to provide all this prior art that you claim exsists so the the examiners can have access to it. They get a limited amount of time to try to find something and if they can't find anything there are limits as to the legal definition of obviousness that can be applied to reject an application.
Re:It's easy when you're on top (Score:4, Interesting)
For example, you can at the same time spend a lot on home security and lobby the government to solve the crime problem. Security solves the problem now, but that doesn't mean you wouldn't prefer not to need it at all.
So perhaps they patent so much because in the current situation it's what makes the most sense, but would prefer not to have to.
IBM tries to pursue only quality patents (Score:5, Interesting)
As an IBM employee who has been through the internal patent vetting process, I can tell you that IBM isn't just spouting off about the importance of patent quality, IBM actually does try to ensure that it only seeks patent protection for really worthy ideas.
A colleague and I (mostly him) came up with an interesting approach for quickly and accurately finding the subject's face in an photo taken for an ID card. The idea is that rather than having to carefully adjust the camera before taking the photo, or having to carefully crop the photo afterwards, it's much more efficient to have a fixed camera that covers a sufficiently large field of view that all subjects, no matter what size, will be in the image and then have software automatically identify their head within the image and crop and rescale to get an image of just the head, with the right size and aspect ratio.
I'll admit that it's not any sort of blinding insight, but there were some very clever bits in the way my colleague made it work and made it fast. Not just algorithmic details, either, but some fundamentally good ideas. Further, after we'd implemented it we discovered that there doesn't seem to be another ID photo solution on the planet that works remotely as well. Most of them don't even try to automatically zoom and crop, and those that do suck at it. To the point the 80% of the time the user has to manually adjust the crop.
So, since IBM offers bonuses for patents, we figured that it had enough novelty in it to be patentable, particularly given the crap that gets patented. We filled out the paperwork and got ready for the review board to rubberstamp our application, or maybe point out a few legal niceties that had to be corrected.
They shut us down cold. "Not novel enough". "The usage may have some originality, but the basic ideas are all commonplace". "It's too obvious".
They told us we could work on it and re-present if we wanted, but they were pretty clear that unless we found some more, better, newer ideas, IBM would not pursue acquisition of a patent on our invention. That impressed me, actually, even though I was disappointed to be missing out on the bonus.
I can't say that *all* of IBM's patents are high quality. In fact I'd be surprised if a few dogs don't slip through. But IBM really does try ensure that it only patents real, novel, non-obvious inventions. Probably mostly to avoid paying any more employee bonuses than they have to, but the industry would clearly be a better place if more companies held themselves to the same standards.
Re:How about "automatic revocation"? (Score:2, Interesting)
Overall I think these ideas are on the right track, but it will be very, very hard to stop people from abusing any patent system, regardless of what that system entails.
Ridiculous hyperbole, you lose. (Score:3, Interesting)
Forbes says [forbes.com] that:
In 2002, IBM spent $4.75 billion on research and development. That's more, in dollars, than Microsoft, Intel, Hewlett-Packard and Sun Microsystems.
Infoworld says [infoworld.com]:
IBM filed more patent applications than any other company with the U.S. Patent and Trademark Office (USPTO) in 2005 to once again lead the annual ranking put out by the U.S. Department of Commerce office.
The company filed for 2,941 patents in 2005, which is down from 3,248 applications in 2004 but still well ahead of second-ranked Canon, which filed 1,828 applications,
Assuming the figures don't change too much annually, the average cost of an IBM patent is about $1.5M per patent. And IBM is a hardware company. I'm confident if you looked at software patents alone, that figure would be a lot less.
Re:Ridiculous hyperbole, you lose. (Score:2, Interesting)
Maybe patents should be given a value or life span according to the amount of work that went into it.
Say, for instance, someone built a small shell script* that could parse all known patents and generate all possible combinations of inventions or methods that haven't already been thought up. Then, let's say there is a useful-enough-to-patent** function that could weed out the chaffe and submit the rest as full blown patent pending requests.
I wouldn't really want to give that person a very long monopoly on every new invention in the whole world.*** It seems that once the time/money he'd invested in his invention machine was paid off (plus some profit) it would do more harm than good to let him charge everyone a tax for thinking.
(* or Lisp or Prolog)
(** do you think this is how IBM comes up with so many ideas?
(*** unless I am that person >8^)