3 ms·
"If I read them correctly the whole business of deciding if something is obvious is seriously broken in that it seems to be limited to combining claims from exi
by hristov 14y ago
"If I read them correctly the whole business of deciding if something is obvious is seriously broken in that it seems to be limited to combining claims from existing patents."
Obviousness is not based on existing patents only. It is based on anything that is published before the relevant date. Thus, examiners should and sometimes do use all kinds of different publications to determine obviousness. But the problem is that the patent office is accustomed to do prior art searches based on existing patents. Thus, internally the PTO is set up to favor prior art searches based on patents rather than other publications. This generally makes sense for other slower moving fields of technology (e.g. mechanical devices, farming equipment, etc.) or for fields where everything gets immediately patented (biotech) but it makes no sense in software.
Another problem is that software was considered generally unpatentable until a landmark supreme court case in 1995. At that time people started filing for software patents and the patent office was generally set up to mostly use patents as prior art searches. Well, there were no software patents then, because software was not patentable until then. The unsurprising result was that a lot of very obvious software patents got through.
Thus, my pet theory on software patents is that there is nothing inherently wrong with them, they were just not correctly administered by the PTO. If the USPTO hires proper competent computer scientists as examiners and encourages the searching of all available prior art, then we would get much fewer and much higher quality software patents.
And I do agree with your idea of opening the review process. The pto is trying to make tepid steps in that direction but they are obviously scared of a flood of comments that may happen if they open up the gates and them having to pay examiners to evaluate every single comment from every crazy person on the net.
- jedbrown 14y agoJournal editors don't have time to personally review every submitted manuscript, yet that peer review process is relatively successful. Journals generally do not compensate reviewers (though I have been compensated for reviewing book proposals---in the form of my choice of free book from the publisher). In any case, what if the patent examiner selected a small number of experts in the specific field and sent the proposal out for review, along with some compensation for quality reviews (positive or negative). A senior USPTO examiner makes on the order of $100k/year (http://usptocareers.gov/Pages/Misc/SalaryRates.aspx http://usptocareers.gov/Pages/Misc/SalaryRates.aspx) and spends 1-3 days reviewing each application. After overhead, that's more than $500-$1500 in personnel expenses per patent. What if, after an initial filter, the application was sent to three experts at $200 each. It seems like this could keep the costs similar to what they are now, but dramatically reduced the number of false negatives for prior art.
- astrodust 14y agoIf nothing of any consequence happened before 1995 then yes, this would be a valid way of doing this, but in reality it's insane. The amount of ground-breaking computer science done in the 1950s continues to amaze me. They were inventing things that only today are seeing many concrete applications. There are so few things that are unique enough to even merit a patent, and fewer of those things that should be patented. If every computer science paper on every new technique were patented you would be unable to do anything useful with a computer without paying millions in licensing fees. I think for the greater good of the industry software patents should be eliminated entirely. If you need to protect something, keep it a trade secret.