Slashdot Mirror
EFF Proposes a Working Code Requirement For Software Patents
Juha Saarinen sends news that the Electronic Frontier Foundation has proposed a fix for software patents in general and patent trolls in particular: requiring applicants to provide specifics about their solution. They say the applications should include working code, or at least "detailed, line-by-line notations explaining how their code works."
"And if they do get a patent, they should be limited to the invention they claimed. We think software patents are bad news, and incredibly harmful to our society and economy. We wish we didn’t have to deal with them at all. But by fixing the functional claiming problem, and limiting patentees to a narrow invention that they actually came up with, we would also limit the amount of harm those patents could cause. The Patent Office does not (yet) have the power to get rid of software patents entirely, but it can fix the functional claiming problem."
This would also allow people who did license the patent to use the code rather than having to re-invent it themselves from scratch, or for that matter people coming along when it had expired...
Hang on isn't that half the point of the patent system? To grant a monopoly on an invention for a limited time in return for providing sufficient information in sufficient detail such that said invention can be replicated when the patent expires?
The question could be asked, why isn't Copyright protection sufficient for your code?
If someone else independently can implement code that does the same thing, then it is obvious it should not be eligible for a patent.
I'll see your senator, and I'll raise you two judges.
The majority of the software patents I've seen are simply a statement of a problem, not a solution. They do nothing to promote the progress of science, and as such should not be considered valid patents, and the laws should be changed to make them not valid patents.
Following the EFF's proposition would be a good way to make sure that your patent states a solution and promotes the progress of science.
If I have been able to see further than others, it is because I bought a pair of binoculars.
This doesn't necessarily do away with patent assertion entities (trolls). Many PAEs are not actually the original inventors/assignees of the patents, but rather buy them later on and begin filing infringement lawsuits. This requirement would, however, reduce the number of startups-turned-trolls who filed and were granted patents but never followed through with development. It also might make the provisional application system more useful, allowing start-ups to file provisionals to establish priority, but also requiring them to develop a working prototype before granting a utility patent.
It seems to me like a lot of these patents are about interfaces, not algorithms or code. If you're patenting rounding buttons or that the software allows you to purchase with one click, the code is probably trivial. It's either a visual or interface idea.
separate the patent fees into a Prelim and a Filing fee. Then if you want to file a prelim patent (to get cover while you do the code) you then pay the Filing Fee to upgrade it to a Full Patent ap. If it gets rejected you can refile but you have to pay the Filing Fee each time (so try 12 times you pay the prelim fee and 12 filing fees).
This will also cut down on the "shotgun" approach to patents.
Oh and a Big Reform would be to autodrop patents in the form of X on %new platform% or X And Y Together (unless you bind them in a novel manner)
Any person using FTFY or editing my postings agrees to a US$50.00 charge
It's not a complete solution to patent reform. It's just a small part. It raises the bar. Instead of just filing a patent on some idea, you at least must have something that works.
I worked as an 'expert' defense witness on a patent case where the patent was just a 'paper patent'. (This was because I had co-developed a product and still had a shrink wrapped boxed copy from way back when.) The patent had many internal inconsistencies. The parties ended up 'settling' after I provided a working implementation that had been sold and offered for sale over a decade earlier. I assume the 'settling' meant the troll went back under the bridge.
If this specific patent had been required to have a working implementation, it would never have been granted in its current form.
I'll see your senator, and I'll raise you two judges.
What the EFF is specifically decrying is functional claiming. That is, an element of a claim described not in terms of what it is but in terms of its function. For example, instead of claiming a nail or a screw, claiming "a means for mechanically fastening" or, even more broadly, "a fastening means." This is allowed under 35 U.S.C. 112 6, now 112(f):
In other words, you can use functional claiming, but you have to explain what you mean in the specification, and you only get what you describe there, plus equivalents. If you don't describe what you mean then the claim is invalid as indefinite.
In the case of software, generically claiming an algorithm (e.g. "sorting means") requires describing in the specification what you mean by that. There has to be some corresponding structure. Failure to properly support a functional claim has bitten many a patentee. This happened just recently in a software patent case involving Google as a defendant, for example. Function Media, L.L.C. v. Google Inc.. Overall, the popularity of functional claiming has been dropping like a rock. From ~1992 to 2011 the percentage of newly issued patents with at least one "means for" claim fell from ~45% to ~10%. It has since fallen to almost 5%.
I am skeptical that a new advanced idea or new or improved algorithm that advances the state of the art would take man-decades to turn into working example code. Are there in fact, any such examples of this?
It sounds like you're talking something you would work an entire career on from the time you're young until the time you're old to have something working. ("man-decades") Even if we're talking 2 man-decades == 20 man years, that was done as 20 people for 1 year, it does not seem unreasonable to have a working model. That had been true for non software inventions. I remain skeptical it would take 20 people a year to build a simplistic working model of even an advanced idea. The working model doesn't have to be the most efficient implementation of the idea. It just has to illustrate an advance in the state of the art.
Yes, I know you are identifying a possible problem caused by the proposed requirement of working code. But the EFF proposal is to fix a much more serious and widespread problem that is real and not hypothetical.
I'll see your senator, and I'll raise you two judges.