WHY THE USPTO/OPEN-SOURCE PARTNERSHIP WON'T WORK: As everyone begins to absorb the news that the USPTO is partnering with the open-source community to improve the quality of software patents, reactions have ranged from guarded optimism to borderline nihilism. And while the details of this program have not been provided, there is enough information and history to suggest that the effort won't work as advertised. There are multiple reasons for this:
(1) Application overload - under the "patent alert" provision, registered individuals will receive applications meeting a certain criteria (e.g. applications from Microsoft, applications relating to CDMA, etc.). Now, unless you are well-versed in USPTO classification and search methodologies, your in-box or RSS feed can contain dozens of new applications every day. And unless you already know what you are looking for (which would be next to impossible in most cases), you would have to devote at least a few hours every day just to find out what it is you're looking at. For each identified application that is "of interest," an additional number of hours (days?) would be needed to further vet the application and conduct your own search. It's apparent that this arrangement will quickly overwhelm any individual or organization seeking to be an active participant in this program.
(2) Information quality provided to the USPTO - examiners have limited time to review a patent application. With all the resources available to examiners (and there are a lot of them), when additional information is foisted upon the examiner, the less time the examiner will have to interpret and apply the prior art. Search templates have recently been introduced by the USPTO to streamline and structure the search process. By injecting the general public into the search process the entire purpose of these templates becomes frustrated. And, with all due respect, there are lots of people (i.e., the Slashdot crowd) that simply don't understand the patent process and have no concept of anticipation and obviousness. If particular applications attract the ire of such people, lots of examiner time would be wasted and severe delays would inevitably result.
(3) Examiners don't rely non-patent literature anyways - back in 1994, IBM spearheaded the Software Patent Institute (SPI), which is the progenitor of the current prior art proposal, and was designed to "provide courses and prior art about software technology to help improve the patent process." It has been, in effect, a failure, as no one in the USPTO found the SPI to be particularly useful. Furthermore, the USPTO already has over 4 million computer abstracts available to examiners for use during examination. However, as the illustrious Greg Aharonian pointed out, the average number of non-patent references cited in issued software patents is less than two - in 2004, 57% of software patent applications do not cite any non-patent prior art.
(4) The "Patent Quality Index" is practically meaningless - for years, individuals and companies have been trying to establish an objective "quality" rating for patents (see here, here, and here; see also US Patent 6,556,992), but there are questions as to how accurate these ratings are. The actual quality of a patent isn't really tested until it has at least gone through a Markman hearing during litigation. Furthermore, companies are more interested in the potential value of a patent more than they are interested in some third-party's assessment of the patent's "quality." And I am not aware of any metric contained in the quality ratings that take into account the ever-changing positions of the Federal Circuit - what may seem to be a good practice at one time (i.e. objects of the invention, using "the present invention" language, means-plus-function claims) can suddenly change into something different at another time. While these "quality" ratings may provide some useful information, it is doubtful that anyone will treat this more than a whimsical tool.
UPDATE: The Just a Patent Examiner blog has an "inside" perspective on these proposals.
Thursday, January 12, 2006
Subscribe to:
Post Comments (Atom)
Posts
I think that this new initiative by the USPTO falls in line with their overall attempt at increasing patent quality via maximizing efficiencies.
This open source collaboration coincides with the recent proposed changes regarding limiting continuations, identification of primary claims for initial office attention and the recent request for comment regarding statutory subject matter as related to data transformation processes.
It is encouraging that the USPTO is acting on so many fronts to attempt to better the system.
I think that you have provided a valuable identification of some potential pitfalls associated with the proposed collaboration.
It will be interesting to see if any of your concerns are addressed as further details/discussions unfold.
I take exception to your assertion that "patent examiners don't rely on nonpatent literature anyway".
I am a patent agent specializing in business method patents (e.g. class 705). The examiners have gotten noticibly better at finding relevant nonpatent prior art, and it's forming a larger fraction of the prior art cited against our pending applications.
I welcome this. I'd much rather have relevant prior art cited against an application so that the applicant will seek more realistic scope in his/her claims, than have irrelevant patent based prior art cited against claims that justs wastes time.
The better examiners get at finding good prior art, the more confident they will be in their searches, and ultimately the faster we will get to allowable claims.
I will grant that Examiners have been getting better over the years in searching prior art. However, the fact remains that very little non-patent prior art appears in recently-granted software patents.
It may be the case that the USPTO is catching up, and that this progress hasn't been reflected in recently published patents. But the fact that many software patents issue without *any* non-patent references is troubling and should be addressed more publicly by USPTO management.
Post a Comment