One Small Leap for Open Source, One Giant Leap for Mankind
press.redhat.com
press.redhat.com
Why are mechanical engineers able to patent their devices, but computer engineers are not?
I understand the ramifications of software patents. But the problems with software patents are the same for patents in every other field. Why is software special?
Why are we narrowing the conversation to software? Patents for software, like any other field require research and development, they save time and money, they make things more efficient, they improve life for countless individuals, they do require risk, innovation, and yes reward to survive.
That's what patents are for. They are to reward the risk takers. In the case of software patents, the reward goes to the programmers who take the risk to write all the code that will be patented. They aren't "abstract" ideas, they are lines of code that are concretely typed into a computer. They convert information from one form to another. They control the physical world in previously unidentified ways.
To say software patents are invalid or software can't be patented really puts software programmers at a disadvantage relative to other fields of science and innovation. Why should pharmaceutical companies, civil engineers, mechanical engineers, and all other disciplines have the right to patent their inventions -- but not software engineers?
A method and system for crossing a body of water via a constructed object. The customer starts on one side of the water and walks across the object to the other side.
Your average software patent is roughly equivalent to a patent on the concept of a bridge. If there are useful software patents, they've been overshadowed by the large number of abusive ones.
Some famous law-suits were about patents that are far less trivial than your bridge example.
For example, Pixar holds a patent on Monte Carlo sampling, and has sued or threatened a number of companies that used any sort of randomization in their rendering algorithms. This despite the fact that Monte Carlo methods have been known and used for many decades before Pixar was started. The issue here is not that the concept of Monte Carlo integration is trivial, but instead that Pixar's contribution to it is so tiny compared to the previous state of the art that it should not give them a "right" to own the whole field.
Similarly, some lawsuits were about compression algorithms if I remeber right. Same issue here - it's not that a given compression algorithm is trivial, it is more the fact that it is probably a tiny increment on the top of existing compression techniques, and anybody that also builds on the known ideas should not be threatened by loosely related patents.
Of course, patents are a problem that should be addressed, but I don't think trying to prove that software patents are somehow different than others is the right way.
But they are fundamentally different in that software is already protected by copyright the same way any other creative work is. Copyright and patent protection are two sides of the same coin that are there in, according to the US Constitution (European theory on the origin of IP rights is different), to provide an incentive to create new things. Unlike a chemical formula or a mechanical invention, the software itself is already protected by copyright, just like a novel or a painting is. It's basically the only thing that I can think of that's protected by both copyright and patent protection.
In many ways, patenting a software system is far more akin to patenting a novel or a how-to book (I'm patenting the idea of a hero that fights an evil villain, and I'm patenting the idea of explaining to people how to fix a leaky faucet).
Again, patents are constitutionally granted on a pragmatic basis rather than on the grounds of some inherent right, so I think it's a perfectly valid to attack them on the grounds that they do more harm than good, and it's also perfectly valid to point out that they are categorically different than patents on mechanical or chemical inventions.
Actually, Pixar has a patent on certain uses of randomization in rendering. In patent-speak, randomization is the mechanism and rendering is the result.
I mention patent-speak because patents are the use of mechanism to accomplish result.
Would you object to a patent on certain uses of a lever to close a gate? (Ignoring novelty issues - closing gates with levers is well known now, while using randomization in rendering was presumably new when Pixar did their thing.) In patent-speak, the lever is the mechanism while closing the gate is the result.
If not, then you're arguing that a lever as a mechanism is somehow different from randomization as a mechanism. If so, you're pretty much arguing against all patents (using mechanism to accomplish result).
It has been long known that MC can be applied to computing any integral over any domain, so computing the integral of color over a pixel is simply a specialization of a general technique, and should not be patented.
Similarly, in akeefer's post above, storing insurance claims in a relational database is a specialization of the well known process of storing anything in a database, so should not be patented.
Patents only work when in old industries?
I personally favor the establishment of some kind of universal license fee that prevents patent holders from hoarding an innovation or charging ridiculous prices for others to use it. That way those who created a concept still get their just due while not being able to prevent anyone from using their patents to innovate.
I'm sure there are good examiners and bad ones. Perhaps this time we just got one who was very thorough, but having looked at a number of software patents in my field, I've yet to see anything that I could consider trivial.
This is akin to the folks who claim we should go back to physical gold as money because fiat money has been abused. Fiat money isn't evil, it's incredibly useful, but we've failed to stem the corruption in our government that allows it to be abused.
My point is that the abuse of a system does not make that system inherently evil, it just means we need to learn from the mistakes and mature as a society. Lean towards overturning the lame software patents for a couple decades and tighten up the standards for new ones.
I would not have much trouble with a patent on the concept of public/private keys, but patents today are too expensive to get, obscure to read and cover extremely simple material.
In addition software patents lasts way the equivalent of several hundred years because the field moves so fast, imagine that basic concept behind the steam engine was still patented today.
In-so-far as software patents do not achieve that aim, they are worthless.
(the same, btw, could be said of copyrights, but not, interestingly, of many other kinds of patents)
Patents for software do not save time and money, nor do they make things run more efficiently, nor do they improve life for countless individuals. This is simply not the case.
Also, patents are not made in specific lines of code - patents would be a smaller problems if they did. They do patent abstract ideas.
To put simply, the main difference is that (besides the fact that most software patents are bogus) since software is pretty much a set of ideas and nothing else (as opposed to hardware where you have a few ideas and a lot of work worrying with beauty, physical constraints, durability, etc) a good team of programmers will have a few hundred good ideas in the course of writing any non-trivial piece of software - which means that patents can kill free software (since it is unlikely that people working for free would want to license software patents), and also make commercial software projects impossible (even if you have the money, having to license patents from 100 different companies would surely be discouraging - specially for a small company).
I didn't say patents do that, I said software does that.
I do not believe software is "just a set of ideas." If that was the case, why do we need engineers to take an "idea" for a software system to implement that system? Building an idea with software takes time, experimentation, trial and error. Mistakes are made. Algorithms are optimized.
Do you really believe that software vendors don't worry about beauty and durability of their software product the same way manufacturers of physical products do? In a lot of cases, we have "designed obsolescence" of mechanical products, so I can't avoid disagreeing with your statement there.
The whole point of patents is to share the knowledge learned in the process of going from an idea to an invention that implements the idea. The problem with sharing the knowledge is that it becomes more difficult to capitalize on all the effort that went into building the invention. You let others know of the invention and how it was built so they can use that knowledge to further improve the world, while also protecting their ability to feed themselves while they work on more ideas to improve the world.
I know many believe software should be free, but the ability to kill free software is one of the benefits of patents. If a company or an individual makes some software worth patenting and then someone decides to give away that same software for free, then the patent holder should be able to kill that free software.
I don't believe that free is somehow more moral than not free. I don't see anything immoral about selling software. In fact, the motives of many who give away free software could be argued less moral than those who sell it.
Why is there an association between people who sell software and "bad" software vendors? A lot of proprietary software vendors are good people who want to create companies and employ more software developers to write more code and make the world a better place. I don't believe not showing all of your cards, or code as the case may be, makes you a bad person or a bad company.
I think that the historical record is pretty clear that software patents are a huge net negative for the industry and do much more to harm innovation than they do to protect it.
How many startups wouldn't be started if people couldn't patent their software, and merely had to have it protected by copyright? Hardly any (probably just ones that deal with things like video encoding). How many startups will be sued out of existence by big corporations or patent trolls due to ludicrous patents, and how many startups won't be started at all because the landscape is already a patent minefield? A whole lot more.
Patenting isn't an inherent right, it's a pragmatic right granted by the government in order to encourage innovation and openness by providing people with temporary monopoly rights in exchange for sharing their inventions. Software patents serve neither to foster innovation nor to foster information sharing, since software patents themselves are pretty much universally unreadable, unenlightening, obvious garbage.
To answer your question. Mint.com patented their software. Yext.com has patents. Lots of startups patent their software and the algorithms that give their companies a competitive advantage.
My argument is that software isn't a special case in the field of innovation. If someone can patent a solar panel that extracts 10x more energy from the same area, then why shouldn't a company be allowed to patent a software algorithm that can compress data stored on disk to 10% of its original size? Ultimately, math and science are behind both inventions.
I also do not agree that it is pretty clear that software patents are a huge net negative. You'll have to provide evidence for that. If you do that, you'll also have to show that patents for software are more negative than patents for mechanical devices or designs in all the other fields if you think software patents should be disallowed, but not patents in other fields.
The question is, why don't we disallow ALL patents? Think of all the lives that would be saved if pharmaceuticals could be sold at generic prices!
There's something really wrong with the philosophy, religion perhaps that motivates a group of software engineers to think their work isn't valuable and shouldn't be protected and isn't worth paying for. The arguments just don't seem rational and my experience with the world tends to discount all the arguments in favor of degrading the field of software engineering to that of garage bands just hacking out code because they love it. Eventually those hackers run out of money and need to eat.
Perhaps there is some idea that software engineers should just really really love what they do! They should love it so much they don't care if they get paid. It's as if getting paid for what you do makes it work. That's not true though. Athletes get paid to do what they love. Lots of doctors and lawyers love what they do. I think it's totally fine to get paid to do what you love.
On balance, patents do more harm than good to companies and their engineers.
The companies you mentioned don't patent for competitive advantage, they patent so that when they get sued by someone else they can countersue. That's what everyone does; if you don't have patents and someone sues you, you can't fight back.
Patents on pharmaceuticals do encourage innovation, because the one essential element is the chemical formula (to a lesser extent, the process to produce that chemical), which would easily be copied without protection. A given software application consists of thousands of lines of code and thousands of algorithmic decision (often hundreds of thousands or millions); if you knew that someone was using an XOR in order to flip bits, it doesn't exactly let you copy their system. My argument is that patents on software don't encourage innovation because they don't increase the protection that's already there, and the threat of other people's patents is constantly looming and stifles innovation.
Look, I'm a software engineer, and I know a lot of other engineers, so I'm not discounting my own or anyone's work or saying it's not valuable and that people shouldn't get paid for it. What I'm saying is that patenting isn't a useful or appropriate mechanism of protection, that copyright combined with the inherent difficult of duplicating the thousands of features in someone else's application is barrier enough, and that patents actually hurt entrepreneurs and software companies and make it harder for engineers to get paid for their work.
You have effectively killed any sort of technical conversation if everyone must protect trade secrets. You have literally killed any advancement in software if everyone must have a the secret dies with me sort of attitude about every innovative step taken by every software developer out there.
The patent system needs to be reformed but if done away with will leave us with little more than Damascus steel.
I'm not sure why you'd think that software advancement would be hindered without patents, or that conversation would somehow die out. Patents for software are notoriously useless to read, and only an infinitesimal part of any system is patented anyway, and then generally only as a landmine against future competitors or to enforce mutually-assured destruction. The vast majority of an application's "secret sauce" is protected by trade secrets, not by patents, and copyright guards against the actual theft of resources. I don't think that patents in the software world really do much to open up what would otherwise be a trade secret, with the possible exception of things like compression algorithms. I think that's fundamentally different from how patents function in, say, the pharmaceutical industry, where patent protection does encourage sharing of information.
A normal patent is for a specific way to reach a result. I can patent my specific "illumination device" (light bulb), but I can't patent "illumination of a space" (lighting a room).
To bring up a well known softare example, Amazon has a patent on one-click shopping. There is no way to create another way to let customers buy things with a single click without infringing that patent, because the patent is for the result (order & payment of goods with one human-computer interaction) not the way it's implemented.
"The Congress shall have Power To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries."
In other words, intellectual property rights are artificial constructs that Congress is allowed to grant in order to promote progress.
In the UK, however, the case law works differently, and IP law is more of an outgrowth of the theory of natural rights that a person naturally has a right to whatever it is that they've created, be it something physical or something intangible.
From what I have read and learn so far, the patent system and copyright system are redundant. The free market already rewarded risk-takers in the form of first-mover advantage, among other things.
I could summarize you what I learned about the history of innovation.(The early American book publishing industry and their relationship with British authors, James Watt's true contribution to steam technology, the fashion industry, etc). But I think it is better for me to point you a book that have all of that and explain it better than I can.
http://www.dklevine.com/general/intellectual/againstfinal.ht...
One thing to keep in mind is that, say, mechanical engineers works aren't covered by copyright. They need to apply, pay, and only get it for a short period of time. The advantage they get is that it's general...
Right now, software developers have both, which doesn't seem fair to me.
Besides, right now copyright is way stronger than patents... Though I don't think this should be the case (I'd like to have, say, 5-10 year copyright).
You are absolutely right when you say that the purpose of patents is to reward the risk takers. Except they actually don't —in my opinion. They tend to favour the big, established players, which cross-licence anyway, nullifying the incentives of patents.
Paul Graham himself basically said the same thing: you can't reasonably reject some kind of patent, and accept the others, if only because you can't set clear legal limits.
Needles to say, I'm against patents. All of them. Rick Falkvinge convinced me.
In the pharmaceutical sector they make new drug discovery economically possible. Were it not for patents we would not have a tenth of the new drugs and treatments available to us today. Marketing costs (an unreasonable expenditure IMHO) and testing costs are completely outrageous and there are so many promising drugs that wash out in stage two or stage three trials that companies need the assured profits of a success to compensate for the risks and costs.
There are many fields that would probably be improved by limiting patents or shortening their lifespan, but in pharmaceuticals the lifespan of patents probably needs to be upped five or ten years...
Now, what kind of drugs pharmaceutical companies will do research on? Cure for rare illnesses? Cure for illnesses widespread in poor countries? This is good for their image, but not their wallet. Far more lucrative are cosmetics, and widely useful medicine in developed countries. Just see the ads, they are likely to match their sells.
I may be wrong on each point, but if I'm right, pharmaceutical companies could reduce their costs down to a third of what it is now, while remaining just as useful. Get rid of patents, and the costs could go down further.
Now if pharmaceutical companies really can't survive in the current world without patents, this still isn't a reason not to ditch them. Specifically, nothing prevents direct state intervention (like spending some of the taxpayers' money). If you are against that, remember this is already happening (at least in the EU), and that patents are a quite intrusive form of state intervention.
> I may be wrong on each point,
You are.
Anyway, I am quite ignorant. Do you have any references I could read so I have a better understanding of the problem? Thank you.
Cosmetics? What cosmetics does Abbott make?
To the extent that the argument has any connection to reality, it's a combintation of irrelevant and wrong.
If you don't want to pay the costs of "cosmetics", don't buy them.
To the extent that there's any cross-subsidization in Pharma, it goes from the "non-essential but popular" to the critical. In other words, you've got it backwards.
Not so fast. You made claims that depended on knowledge. You suggested policy.
And now it turns out that you knew that don't know what you're talking about....
Anyway, I stumbled upon a better source than myself: http://www.dklevine.com/general/intellectual/againstfinal.ht... (a case against intellectual monopoly). Chapter 9 is particularly relevant here. If you have any counter-argument, I would be glad to hear them.
Marketing includes giving away free/discounted drugs, telling doctors when a drug works, and telling people that the problem that they have might be curable.
Which one do you want to cut?
* It must be novel.
* It must not be mathematics. Maths cannot be patented.
* It must be non-obvious.
Most software patents violate one or more of the above three conditions.
That isn't what patents are for. The intent of patents is to get inventors to disclose their inventions to the public for the benefit of all. Rewarding the inventor is the incentive that make this happen, makes it work, not the goal.
The authority for patent laws ultimately comes from the Constitution and finds its implementation in statutes passed by Congress. There is nothing in either the Constitution or in the existing patent statutes that expressly allows courts to limit process or business method patents (the type at issue in the case before the Supreme Court) to those that pass the so-called "machine-or-transformation" test imposed by the lower court. Thus, it is by no means assured that such a relatively narrow test will be upheld. If it is not, then the looser standards likely will be back in play, meaning that a broad variety of process patents (include software-related) will be capable of being granted.
For those interested, this case has generated huge interest in the tech world and, as of a couple of months ago, at least 44 such companies had filed briefs with the Court (for what it is worth, most of them favor rejecting the lower court's ruling and thereby letting more process patents be issued). A good summary of the positions appears here: http://www.patentlyo.com/patent/2009/08/briefs-in-bilski.htm....
I am not saying that software patents shouldn't be abolished - just that such a result is highly unlikely to come from any ruling in this case.
You're probably right. The best way for a person to help get software patents abolished, IMO, would be to join their local Pirate Party -- http://www.pp-international.net/
The _real_ way to get software patents taken care of is, as with any other matter of public policy, to convince large swaths of major, consistent contributors to the campaigns of your incumbent representatives to press the issue.
This will probably be difficult because people with money want to do evil with these patents and they will just counter your pressure, though there is some hope that the big players will decide that software patents are too dangerous altogether if a few more injunctions like the one against the sale of Microsoft Word come out (see i4i v. Microsoft). But right now, Microsoft is trying to use its patents to neuter its open-source competitors, so they may value that too highly. We'll have to see!
> Heh, no it's not. That'll just get you marked a radical and summarily ignored by anyone with power. Not that I agree with that, but that's how things are.
The Pirate Party already has one MEP. If (probably when) the Lisbon Treaty is passed, they'll have two. The Pirate Party will probably be in power in several countries over the next 10 years (as part of coalitions).
> The _real_ way to get software patents taken care of is, as with any other matter of public policy, to convince large swaths of major, consistent contributors to the campaigns of your incumbent representatives to press the issue.
This might be true in the USA under their current political system. But as you imply from what you say, the USA isn't properly a democracy, it's an oligarchy ruled by corporations. This is in fact a bigger problem than software patents.
It isn't true in Europe, where Pirate Parties have a siginificant chance of getting people elected, due to using more democratic electoral systems.
If Pirate manage to get software patents abolished or neutered in one major jurisdiction, they will probably indirectly win in the rest of the world.
> This will probably be difficult because people with money want to do evil with these patents
I agree.
> But right now, Microsoft is trying to use its patents to neuter its open-source competitors, so they may value that too highly.
That's certainly part of Microsoft's plan for their patents.
Except that they basically didn't exist before the StateStreet case. The court has signaled that it may be willing to reconsider its previous ruling, so there is definitely hope.
So, patents apply to implementations, right? Also, code itself is the design, right? And yet, so far we try to describe the implementation with vague words of mostly English language, and to bend the boundaries by making it as broad as possible. Doesn't seem to work too well.
In real world, you can patent an implementation of a bridge -- effectively the design of its construction if it's novel. You can't patent the underlying, generic bridge-construction principles. Code is design.
I think the principle could apply analogously to software patents by opening the source of a proprietary implementation, and then licensing interoperability with the patented implementation. In other words, that spells like kind of a code-escrow for an exclusive licensing rights for limited time.
The patent would only apply to forks or third-party implementations that are compatible with the patented implementation. If you invent a clever protocol, you could require money for compatible implementations: this would hopefully incentivize opening proprietary implementations.
On the other hand, if someone takes your ideas and builds his own implementation, you would not be entitled to anything: that someone will be required to build his own market share, compete with your implementation and user base from the ground up, and effectively do the same work that you have already done -- just to get on par with you.
Complete codebases and unit tests could be required to make it easy to verify the compatibility or incompatibility of programs potentially based on the patented codebase.
Reviews for novelty would be much easier to arrange if expert programmers could study the working code and compare it to earlier implementations. That is, as opposed to lawyers trembling with the words and semantics of patent applications.