[cromwellian]

Software patents are an abomination. I could stomach copyright on software, because at least one is still permitted independent implementation and expression. I can't copy your code, but I can rewrite it.

Software patents on the other hand put a fence around ideas themselves. You can't draw a cursor using XOR by implementing it yourself, period, for 20 years.

The patent system is deeply broken, and it doesn't even stop big players anyway. Really, Apple successfully sued Samsung, did it stop Samsung from taking over half the market? Does $1 billion in fines really matter or Apple or Samsung over the long term? By the time these cases are settled, it has already played out in the consumer marketplace anyway. You can't defeat consumer success with patent attacks. Microsoft's Android revenue shakedown won't replace the death of Windows if it happens, and it won't make Windows Phone/Surface RT a winner.

It's a game only lawyers, IP trolls, or paid industry shills love.

[leeoniya]

"Get a dozen sharp programmers together, give them all a hard problem to work on, and a bunch of them will come up with solutions that would probably be patentable, and be similar enough that the first programmer to file the patent could sue the others for patent infringement." -- John Carmack

from http://harmful.cat-v.org/software/patents/carmack

[jandrewrogers]

This is obviously not true in many cases. There are countless algorithm problems where decades and thousands of pages of published literature on the problem go by before some researcher manages to make a material advance. Computer science is full of cases like this.

If all of the interesting problems were so trivial to solve, they would not be "problems" by definition. The idea that smart programmers could solve these problems any time they wanted to if they put their minds to it is unrealistic. They are considered "hard problems" precisely because countless smart programmers have failed to find a solution after no small amount of effort.

[ScottBurson]

Well -- there are hard problems -- and then there are hard problems.

It seems that a lot of patents are being granted for solutions to problems that are only moderately difficult, if even that. Certainly, I have seen a number for which the solution seems quite standard given the problem. I think what happens here is that new technologies create new problems, which though not deep, are novel. These are solved in quite standard ways, in many cases; but since the problem is novel, technically, so is the solution, so the PTO grants a patent for it. And a large fraction of the patents being issued are of this nature.

I think the solution is to restrict patents to problems that are demonstrably hard -- of the kind that you describe, where there is objective evidence (in the published literature, for example) that people have thought about the problem for some time without solving it.

[fspeech]

The key here is that patents are supposed to be non obvious to those "skilled in the arts" and herein lies the rub: who are those skllled in the arts? In my previous career (wireless and IC) I saw most patents as trivial for someone skilled in the arts: they were mostly byproducts of routine engineering design work. They were novel only in the sense that you are probably the first to work on a specific form. Yet they could appear rather deep and profound to the untrained eyes as you need a lot of training to be skilled in the arts in the first place. Can patent examiners truly judge the novelty of these patent applications?

And the funny thing is: if you come up with some real algorithm for some real hard problem it is unlikely that your algorithm can be protected by a patent, certainly not in its general form. The true algorithmic nature of your solution is likely too abstract and mathematical for a patent as patents do not protect laws of nature (physics and math qualify as laws of nature).

[rayiner]

> It's a game only lawyers, IP trolls, or paid industry shills love.

I think it's a major problem when something like 60% of all IP litigation is over software patents, a relatively obscure area of the law until recently, but I disagree with this statement. I worked for two entrepreneurs who founded real R&D companies and considered patents to be a basic protection that allowed them to operate a small R&D company independently of a massive manufacturing company. These guys couldn't be further from the description of "lawyer" or "IP troll" or "paid industry shill." They're PhD's who have spent their entire careers inventing.

If you characterize everyone who relies on the existing patent system as a "lawyer" or "troll" or "industry shill" you'll make no progress in achieving the reforms of the system that need to happen. You cannot figure out how to solve a problem while understanding only one side of the issue.

[cromwellian]

"R&D company" that ships no produces seems a bizarre concept to me. If you're doing pure research and you don't want to manufacture anything, then outsource it, or partner with someone who can and sign them to an iron-clad contract protecting a trade secret.

I'm be curious if anyone can even name any "pure R&D" software companies that don't ship products, yet created meaningful advancements. Even in the realm of hardware, did RAMBUS really need patent protection?

[rayiner]

Trade secrets you share with other companies are fragile rights. Do you want to invest a ton of money into a company built on a property right that could evaporate at any moment? Re: meaningful advancements, I think MPEG is a decent example, as well as some wireless companies (it's really hard to draw a clean line when it comes to wireless, because a lot of techniques could easily be implemented on a general purpose computer or in an ASIC).

As for RAMBUS, the memory industry is an interesting example. They all work on developing memory technologies, then pool the patents. The patents function to prevent free-riding on the collective effort by any particular company. The wireless industry works similarly (LTE, etc).

I don't mean to argue that the law should protect particular business models. But I think in the abstract, it's useful to be able to bundle up the fruits of some R&D into something that can be the subject of market transactions. I think it's worked amazingly for the semiconductor industry in the model of ARM, for example. The law can provide a mechanism for doing that. I don't think trade secrets are quite it, and I think we're really seeing the limitations of patents for that in the software realm, but I think there needs to be something. Otherwise, it'll be a race to see who can outsource most effectively to China because that's what matters when ideas can't be protected.

[euroclydon]

ARM Holdings. From Wikipedia:

Unlike other microprocessor corporations such as AMD, Intel, Freescale (formerly Motorola) and Renesas (formerly Hitachi and Mitsubishi Electric),[28] ARM only licenses its technology as intellectual property (IP), rather than manufacturing its own CPUs.

[thomaslangston]

Not a software company.

[smnrchrds]

He was referring to "Even in the realm of hardware, did RAMBUS really need patent protection?". There are definitely exceptions in hardware business.

[throwawaykf02]

> I'm be curious if anyone can even name any "pure R&D" software companies that don't ship products, yet created meaningful advancements.

There are many examples (in addition to ARM) -- just think "research labs". Be they government-sponsored, private or academic, their primary function is R&D, not shipping products. One example of a private lab, off the top of my head, is SRI.

Also, though you may claim they don't count as "companies" -- universities.

Edit: Additionally, in the realm of hardware and semiconductors, most companies these days are actually fabless [1]. There is a study claiming that patents facilitated the entry of specialized design firms into the market, although incumbent industry executives didn't look on them too favorably [2].

BTW, Rambus is not the best example. They created a lot of controversy when they asserted their standards-essential patents, the same thing Motorola is doing these days. Whether they "really needed patents" depends on which side you're looking from. But for the industry in general (and I am horribly oversimplifying very nuanced conclusions), the answer seems to be yes for startups, and a mixed bag for big ones [2, 3].

1. http://en.wikipedia.org/wiki/Fabless_manufacturing

2. http://www.techpolicy.com/Articles/P/Patent-Paradox-Revisite....

3. Various other papers by Rosemary Ziedonis at http://www.techpolicy.com/Academics/Ziedonis.aspx

[sspiff]

rayiner wasn't talking about software. Software companies have an easy time "producing" their products. It's much harder in the hardware sector or when you invent car technology. Getting into those markets is expensive. Forging a deal with a car manufacturer, for instance, would put you in an extremely weak negotiating position. Patents give you some leverage and a defense mechanism when they go and use it without your permission.

So I also think ARM is a valid example of companies that wouldn't be alive if not for the patent system.

All that said, I'm not in favor of patents at all. Thinking of something should not give you ownership over a broad concept. Information should be shared freely, to improve upon and so everyone can enjoy the invention.

[TylerE]

I disagree with this sort of extremism. What should not be patentable are basic "way of doing business" patents.

On the other hand, if someone slaves away for years and develops a truly new and unique compression technique, I don't see why that shouldn't be patentable. How is that any different from, say, a steel foundry developing a new unique alloy?

[skybrian]

What's the difference between that and a mathematician slaving away for years to solve a hard problem? And yet, we don't grant patents on math.

One person's time isn't that big an investment (relatively speaking) and funding is available.

[TylerE]

Because you're patenting a concrete implementation.

[skybrian]

That's not as black and white as it seems. An implementation of a compression algorithm is typically a particular library written in a particular computer language. A specification of a compression algorithm is typically described in a specification document such as an RFC. We don't call an RFC an implementation.

[axman6]

What is patented is a method of compressing data using algorithm X. If you can thing of something else which wouldn't be considered compressing data using algorithm X, then you're free to use algorithm X for this new and inventive purpose.

[TylerE]

Some reason, if someone comes up with a highly valuable proprietary algorithm, I don't seem them issuing an RFC on it.

[skybrian]

Now you're nitpicking. The only reason I mentioned RPC's is as an well-known example of a specification, to contrast that with a concrete implementation.

My point is that an algorithm is more general than any particular implementation. They aren't concrete. They're abstract. (And of course if patents only covered one concrete implementation then nobody would bother getting them.)

[otikik]

I would argue that advancement in software is too damn important for the humanity as a whole to be hindered by the preferences of that person. Software will ultimately change us. Shape us.

I think that software will advance more rapidly without patents than with them, even if that means having less immediate financial benefit for the individuals involved in the discoveries.

[cromwellian]

BTW, for Florian Mueller who might be reading, here is your paymaster's official published position on software patents:

http://www.ibiblio.org/patents/txt/020294.txt

[flixic]

It is, as John Siracusa called it, "Sport of Kings". He elaborated a lot on the topic in episode 69 of Hypercritical: http://5by5.tv/hypercritical/69

[joe_the_user]

I agree.

But to take note of the stakes here, banning software patents fully would eventually end patents as such. The problem for the patents in general is that nearly all machine methods are going toward a general purpose machine, a computer, telling a robot what to do. So all innovations here are going to eventually be expressed as software (even if they involve innovations chemical reactions or factory automation).

I think it will be better ending this system but there are a lot of companies whose intellectual capital is tied up in patents and these will fight hard to postpone this end.

[gwu78]

"It's a game only..."

What about governments who collect the fees?

For example, how much have software patents ("junk patents") contributed to the USPTO's balance sheet? Is this relevant? In this position, can the USPTO self-regulate? I ask this not rhetorically, but sincerely.

If the small countries that restrict software patents had the market size of the US, it would not surprise me if they allowed more unrestricted patenting of software. But China stands as a counterpoint to the idea that market size is the differentiator.

Perhaps it is the willingness of US consumers to pay higher prices. This enables patentees to pass the expenses of patent office fees, court fees and lawyer fees to the consumer (even when the patents granted were not practiced to make the product: they're "junk"!).

[jgeerts]

You got my vote on the first sentence.

[Bjoern]

I don't think the original patent idea was bad, only the recent years mutated it into an abomination.

A long time ago this model made sense. Inventors would invent and would be protected to benefit exclusively for a very short period of time (e.g. one year). After the grace period, it would be a free for all again, and people would move on.

Having patent durations of >1 year is obviously ridiculous, particularly 100+years protections etc.

[jobu]

My biggest gripe is that software gets double protection - copyright and patents. I would be ok with software patents if that meant you had provide the source code as part of the patent application, and if approved it would go into the public domain.

[throwawaykf02]

It's not really double protection, because patents and copyright cover different things. Copyright only protects the exact expression of what some software does. Patents cover the functionality of that software. In fact, since software per se is not claimed, the claimed functionality could be implemented entirely in hardware and still be covered by the patent.

No amount of copyright can protect the functionality, because 1) there are countless ways of implementing the same functionality and copyright only protects your specific way, and 2) copyright cannot legally cover functionality anyway.

As such, providing source code in a patent is of little value. (Although I have seen patents containing source code.) As long as you can reimplement the method being claimed by reading the patent with little undue experimentation, it has met its "enablement" requirement.