Bilski Oral Argument at the US Supreme Court (Software Patent Case)
fsf.org
fsf.org
1. The USPTO has in fact tightened up significantly on the granting of software patents for the past five years, meaning that the flaky ones (like One-Click) are historical artifacts and not indicative of current practice.
2. There is nothing in the law that can reasonably be interpreted to impose a ban on software patents and this will not happen in Bilski.
3. Abuses in the field are rampant, with patents mills acting much like extortionists running a protection racket.
4. In the long run, the problems will be largely self-correcting, as the same blizzard of software patents that now clogs the system will eventually become prior art that will serve as the best defense against trivial patents of the future that would otherwise hinder innovation.
Take the points for what they are worth. Personally, I think Bilski's claims will go down, as will the absurdly expansionist interpretations proferred by his lawyer, but this will not put an end to software patents. It is Congress which is constitutionally charged with defining patentable subject matter and it will take nothing less than an act of Congress to impose definitive reform in this area.
[0]: http://arstechnica.com/tech-policy/news/2010/03/controversia...
Not true; I saw a patent the other day for a system call being transformed from (in one application of the patent) "send(data, host_port)" to "send([data1, data2, ...], [host_port1, host_port2, ...])"
What if you are a mechanical engineer who wants to construct a new invention? How will you know if someone holds a patent claim on it already? What if you are a chemical engineer who wants to invent a new substance and someone has a patent claim on that?
The entire universe is nothing but math. It's all math. Statics and Dynamics is no more or less math than encryption algorithms. Everything that can be patented moves physical objects around the universe, whether they are magnetic poles on a spinning disk or they are levers, motors, and dirt and there are myriad patents on backhoes, cranes, and forklifts.
Math is all there is. Software is no more or less math than a lightbulb. Everything is math, because math is just a language used to describe the physical world around us.
The problems stated here are not unique to software, they exist in all fields where innovation is essential. Either do away with all patents or do away with no patents and reform the entire system.
Software is no more or less important than the planetary gear system in the Toyota Prius. Would the world be better if Ford, GM, and Honda could use the planetary drive system in their Hybrid cars? Maybe not! Maybe we wouldn't have a planetary drive at all! Why would Toyota spend so much money on R&D to build a great hybrid system if they knew Ford was just going to take their hard earned and expensive invention and make a bazillion dollars on it.
Ford can use the planetary drive if they pay to license it. Software developers can use patented algorithms if they pay to license them too -- that's the whole point -- protect the innovators who pour their hearts, minds, and souls into making the world a better place.
Unless there is a profitable mechanism like patents to protect intellectual property, the risk/reward ratio becomes prohibitive and I am not sure innovation won't be hurt.
I'm not saying we shouldn't do away with patents. Let's be pragmatic and see what happens, but what I do say is that software is no different than hardware. Equal opportunity under the law, for software engineers or mechanical engineers.
Why is there a difference? The more I look around the world, the more discrimination I see wrt software engineers. I think a lot of it is due to the abstraction. People can't see software. They can't see the for loops and while loops, but there is no difference between a gear that moves around in a circle and a loop construct in software, so why are people saying they should be treated differently under the law?
Also, I'd like to suggest one reason to distinguish software patents from others: they hurt more, and they kill certain types of development. If a mechanical engineer has to license a patent and pay a certain amount per widget, that might cut into their profit margin or raise the cost per widget, but it won't necessarily kill the product (unless the patent owner outright refuses to license). If a software engineer has to license a patent for a FOSS project, even a fraction of a cent per copy would kill the project, because it fundamentally can't work with a FOSS license.
Note that I've worked with mechanical engineers who give away their designs under similar terms, at which point they'd face exactly the same issues; however, this represents a far rarer case than in software.
That said, I do personally agree that patents in general ought to go away. However, I think that software patents do a lot more damage than some other kinds, and therefore it makes sense to focus on them first; they also represent a newer and more tenuous concept that we can more easily fight.
It's the concept of this ethereal stuff that bugs me. The IsNot patent and the One Click show the system is broken. Heck, look at the xor patent. I think the easiest way to have wrote the law is that it must be a physical object to be patented. Otherwise protect it with copyright.
For it to work, the bar for "nonobviousness" must be set infinitely higher than it is now. I'm not entirely opposed to software patents. For the RSA algorithm, for example, I could see the validity of claiming it's a patentable invention. But there are a few key requirements:
a) Patents can only cover process, not end results. If you patent a way to make steel, I'm free to develop another way if I don't want to or am able to license your method. In the same way, if you patent a way to "process gestures on a multi-touch device", I should be free to implement another, independent way of accomplishing the same result. You must not be able to patent the idea of accepting input through gestures. Doing so will severely stifle innovation, because there is no longer any incentive for anyone to put in the work to improve the idea; the first, half-assed patented idea then becomes the way or the highway...
b) For the system to work in practice, there must be a way that one can reasonably find out whether something is patented. That means the bar must be very high, so that the number is lower, and the overly abstract, overbroad patent descriptions currently must be replaced with something much more precise. I think the bar should be high enough that the numbers of patents approved number in the hundreds, max, per year. For it to be patented, it should be freaking genius. Your jaw should drop from pure awe when you read the description.
If the above requirements are not met, the best way to proceed is to abolish the whole idea.
As an aside, I'm amused how many people who normally are adamant about the supreme efficiency of free markets then completely turn around and use arguments like yours above about "why should they develop this when someone will just steal the idea". What is the fundamental difference between patents and preventing someone from opening a coffee shop in a good location because you opened one there first? Why don't you think free competition will produce the most efficient outcome when it comes to patents?
Patents are a reward for both the inventor and society. If we reduce the reward for innovation, we reduce innovation. Especially very expensive innovation that may yield products that are easy to copy.
I suppose my big concern is that for the individual inventor, there really isn't any protection at all besides patents. An individual isn't big enough to mass produce products or protect them or market them more than the big corps, but the individual can sell the patent or get investors to help him/her fight infringement.
Yes, the megacorps get most of the benefit now, but that is largely because most inventors work for the megacorps. But a patent only costs a few hundred dollars and is well within reach of individuals who are willing to do the research and put the effort into building a patent to protect an idea. Patents are the only protection left for joe or jane inventor. No megacorp is going to pay for your idea or your invention if they can just build it themselves.
As for free markets, I don't believe in free markets, or at least, I don't believe the market is free -- or ever will be free. It's not free because special interests want to protect their access to the market. I think the market should be regulated, especially when the public has to bail out systems that collapse due to a lack of regulation. But I digress...
Design patents cover end results. If I patent a design, you can't use the same design -- even with a different process to get there.
There is a reasonable way to find out if something is patented, the USPTO has a patent search. It could be better, but no one ever says, "Hey, I should build this, let me search the USPTO first!" They just start coding or building.
It may be too easy to get a patent and the USPTO may not have the experience to evaluate nonobviousness -- unfortunately, Einstein doesn't work there anymore. Perhaps the govt should pay more for people who have the experience to adequately evaluate claims and charge more to get the patent approved.
Why is that your big concern? Would you like to take a guess at what percentage of GDP that example represents in practice? (hint, single digit is too high).
I don't think the corporation vs. corporation battle is as significant as the corporation vs. the little guy -- David vs. Goliath. Patents in this case are David's slingshot. Or shield maybe.
I don't worry about giving corporations any additional leverage. If we want to spur innovation, we have to enable more innovators to exist and profit from their inventions. Megacorps don't need help -- startups and individuals do.
This may be the case in some fields, but it clearly isn't for most software patent cases. There is very little chance (for this exact reason) that HTC's or Google's engineers read Apple's patents. And yet, they implemented something that Apple considered similar enough to sue over. Outside of a very special cases (like encryption algorithms), it seems that nearly all software patent infringement is accidental, the result of people trying to solve the same problem and happening to run into the same solution. Combine that with an incredible store of over-broad and obvious patents, and you find a world where it is virtually impossible to write anything without violating somebody's patent (actually, if you are writing gui applications, this is literally true).