got a letter from Bruce Lehman
I got an email from Bruce Lehman, which was very big of him after I criticized him for his claim that I "seem[] to believe you can have a post-industrial economy without any copyrights."
Anyway, Bruce's email (and to be clear, it was sent not just to me, but to me as a "IIPI Supporter" (which has as much connection to the truth as his statement above)) was proud to announce a new section on the IIPI website called a "Discussion Forum."
The discussion forum was inspired by the "debate surrounding the European Union’s (EU) proposed Directive on Computer Implemented Inventions (CII)." As Bruce writes, "It is important to remember that the patentability of computer implemented inventions, or lack there of, will have a profound effect on European industry and competitiveness." Absolutely right, which is why is it so good that the IIPI has opened a discussion forum on their site so people can contribute to the "discussion" about IIPI's strong support for this software patent initiative.
I encourage all to answer Bruce's call: You can find the "CII Discussion Forum" here. And be sure to spread the word!
| Permalink | technorati
Comments (64)
software patents - now you are on the mices' home turf!
the campaign for creativity - run by the lobbyists supporting the EU directive - also has a blog about the directive.
patentanwalt axel horns also has an excellent blog where he provides up-to-date links and commentary on the directive (and other things.)
simon gentry of C4C is very earnest in his lobbying efforts, but he does not quite understand the nuances of the debate. neither do the FFII who oppose him. and neither do the various patent authorities who do not understand that extending the nominal regime of patent protection to software is a very bad idea indeed.
that is not to say that software should not be patentable - indeed it should be and it must be - the issue is how to do this without causing indiscriminate and unintended harm.
the opponents of the directive are raising some valid and important points, but these are being lost in the rancor of the debate and the simplicity of the solution offered by the directive's opponents.
the FFII's solution is embodied in the recent amendment offered by the french socialist MEP michel rocard in which patents would not be allowed for software that handles "the treatment, the manipulation, the representation and the presentation of information".
this would exclude from patent protection such things as speech codecs, file compression methods, digitial decoders, and many other technical inventions that are embodied wholly in software.
the mice see no good argument for excluding from patent protection a wide swath of technology, simply because the inventions in this area happen to be embodied in software. to an engineer software is a tool just like resistors and capacitors are tools.
at the same time we fully understand that there are substantial problems with extending the nominal regime of patent protection to software - not the least of which is the apparent inability of the patent authorities to consistently and properly examine such inventions.
the software patent debate is a little like the copyright debate where one side says do away with the restrictions of copyright, and the other side says even more restrictions are needed. it is our opinion that no software patents are as bad for society as ordinary patents granted on software.
pity the two sides cannot discuss this and seek a compromise. the assumption must first be that software - like any other technical invention - should be patentable. once this threshold is passed, then the debate can be focused on how to properly do this.
in the current, winner-take-all debate everyone looses.
Hmmm ... that's a very interesting tactical choice, to call in a flame-strike. It can of course do notable damage against low-ranking individuals. But I'm not sure how well it does against the discussion forum of a think-tank. Especially a business-oriented think-tank. It'll be instructive to see the results, if any.
Loath as I am to agree with the Mice in public, I feel that I must admit that, at least to some extent, I do.
It does seem to me that whether an invention is implemented in software is not the crucial question, or even a relevant one. I think that it may go back to Stallman's ideologues back in the days of the Lotus look-and-feel patent lawsuit, and the disastrous decision to allow patentability of mathematics, which led to the effective unavailability of public-key encryption for the term of the RSA and Cylink patents. I agreed with him then, but it seems that too many simply took the approach that the problem was software implementation, which was not true in either case.
The patents that we have all seen and strongly protested failed both the novelty and non-obviousness tests, miserably.
I am afraid that - for whatever reason, whether incompetence in the underfunded PTOs, undue influence from large businesses, the relative novelty of the technology, or all of the above - the real problem is that the bar of non-obviousness has been lowered to insignificance, and the standard for prior art gutted, while the financial barriers to patent application are irrationally high. The combined result of all this is that well-funded organizations can obtain frivolous, trade-strangling patents while small inventors are frozen out.
"It does seem to me that whether an invention is implemented in software is not the crucial question"
Ah, I'd say that is indeed a crucial question - because it forms a profound if subtle shift. It comes much, much, closer to patenting ideas, to giving monopolies on abstract concepts.
This is well-trod ground, see for example:
http://lpf.ai.mit.edu/Patents/patents.html
In particular:
http://lpf.ai.mit.edu/Patents/against-software-patents.html
Once can argue that the line *should* be crossed - but, implicitly or explicity, a line *is* being crossed.
As an EU citizen I strongly object that a US lobby group with strong affiliations to the USPTO and the Person who takes responsibility for US software patents interferes in internal EU affairs. This is against the diplomatic code, Mr. Lehman. Is he still a US wipo representative by the way? If it was so I would like file a complaint and sent it to the next US embassy.
All the rhetoric about software patents as patenting ideas is hogwash, unless those patents are invalid under the existing patentability standards.
Software is just another way to implement a machine. It's a machine that operates on blobs of data, which can either represent real-world objects, or be social or intellectual 'objects' in and of themselves. Can you seriously assert, when we buy a song online, play songs via our computers, and manipulate them almost always as computer files, that these are not sufficiently legally cognizable objects such that software transforming or manipulating them should be promoted through patent law? I don't see how.
Early in the history of computers and software, things were not so clear -- there were not always standardized 'objects' or ways of manipulating or representing data. Now there are, and more and more of our vocations and avocations involve manipulating digital content. Denying patent protection to inventions in this supplementary universe is denying patent protection to more and more of our lives, and to things that often would clearly have been implemented as machines if they had been possible before the advent of computers...
People can quibble about terms and where to draw legal lines, but fundamentally, I think it's ridiculous to assert that no software should be patentable.
Three Bline Mice wrote: "this would exclude from patent protection such things as speech codecs, file compression methods, digitial decoders, and many other technical inventions that are embodied wholly in software."
Yes it would, and quite rightly too, because these things are software patents. These things are currently excluded from patentability by the 1972 European Patent Convention, and despite the best efforts of the EPO to push back the boundaries of patentability in Europe to cover software inventions such as those you describe, I think we have a good chance of preserving the very sensible exclusions from patentability that were spelt out by the 1972 Convention.
Ah, I'd say that is indeed a crucial question - because it forms a profound if subtle shift. It comes much, much, closer to patenting ideas, to giving monopolies on abstract concepts.
this has nothing to do with patenting "ideas" or abstract concepts seth finkelstein. this is what the FFII and others keep saying, but this is a canard.
despite the rhetoric from the directive's opponents, that - horror of horrors - US style patenting might be possible in the EU, it is worth nothing that the SCOTUS has maintained for 150 years that ideas and abstract concepts cannot be patented.
in 1854, in the case of le roy v. tatham, the court said:
“a principle, in the abstract, is a fundamental truth; an original cause; a motive; these cannot be patented, as no one can claim in either of them an exclusive right.”
in 1874, in rubber-tip pencil co. v. howard (20 Wall. 498, 507), the court said more succinctly: “[a]n idea of itself is not patentable.''
in 1939, the court clarified in mackay co. v. radio corp. (306 U.S. 86 94) that: “[w]hile a scientific truth, or the mathematical expression of it, is not a patentable invention, a novel and useful structure created with the aid of knowledge of scientific truth may be.''
later, in 1948, in funk bros. seed co. v. kalo co., (333 U.S. 127, 130) the court emphasized that: "[h]e who discovers a hitherto unknown phenomenon of nature has no claim to a monopoly of it which the law recognizes. if there is to be invention from such a discovery, it must come from the application of the law of nature to a new and useful end.”
there are not obscure legal footnotes, but oft-cited, well-settled law in the united states. if anything can be said with clarity, it can be said that ideas and fundamental principles of nature cannot be patented in the united states (or anywhere else.)
no one is suggesting that ideas or abstract concepts can or should be patentable.
it is always better to discuss this a concrete example. a convolutional digital filter is not an idea. it is a technical invention that performs a technical function. the same is true for a digital decoder such as most likely sequence estimator (MLSE) equalizer - used in digital communications. these are inventions that are embodied wholly in software - that is lines of code run on a general purpose computer.
proponents of the directive argue that inventions such as these should be patentable - subject to all other requirements of patentability (new, novel, non-obvious, etc.)
opponets of the directive say that such things should not be patentable - simply because they are implemented in software.
we mice think it is not only ignorant, but downright insulting, to say that there are no inventions in software, or nothing worth patenting.
that being said, even three blind mice can see that there are substantial problems with extending a nominal regime of patent protection to software:
the costs
the enormous cost of acquiring a portfolio of patents, and the cost of avoiding patents, does not fit well into software’s otherwise low cost development model. as a slashdot poster observed, “the amount of money required to perform the due diligence research seems like it would be greater than the amount of money needed to develop the software, or even the total revenues that the software could ever generate.”
the costs of acquiring patents are even greater. as forbes magazine noted:
“... a rule of thumb is that gaining sufficient protection for a patent internationally will cost about €100,000. filing ten related patents to create a protective “picket fence” around the key patents will cost €1 million.”
this ain’t hyperbole. in 2005, it will cost an applicant, using professional assistance, between between €5,000 and €10,000 to obtain a single patent from the EPO.
as patent attorney fees account for a significant portion of the expense, lowering the fees at the EPO will not substantially lower the cost for the applicant. for most individuals and small businesses, acquiring one patent, let alone a portfolio of patents, will remain out of reach even if the government lowers the fees to zero. attempting to save money by obtaining patents without professional assistance is not a practical or recommended solution.
the cost of patenting does not fit into the low cost development model for software.
how is this going to be solved?
the practical difficulty of avoiding infringement
the complex, arcane, and obscure language found in most patents makes recognition of inventions and performing “product clearance” very difficult. even for diligent, responsible companies with deep pockets, it is very difficult to avoid patent infringement.
how is this going to be solved?
there is no “fair use” in patent law
unlike the fair use exemption that enables authors to use short quotations of copyrighted material in a book, there is no “fair use” in patent law. minor infringements for personal use and academic research - long legal under copyright protection – would not be legal under the current patent regime.
how is this going to be solved?
there is no “clean room” avoidance
in patent law, there is no defense of independent development. the “clean room” development of software products that prevent allegations of copyright infringement, offer no protection against charges of patent infringement. even if you independently arrive at the solution, if you infringe on the claims of someone else’s patent, you are potentially liable.
how is this going to be solved?
there is no reverse engineering
while copyright law permits limited infringement to reverse engineer interface information – a sine qua non for application developers – no such equivalent exists in patent law. software is unique in that interoperability between devices is vital. the adoption of the existing regime of patents presents a significant threat to application developers.
how is this going to be solved?
insufficiency of disclosure
in an application for patent the technical description only has to be sufficient in order to support the claims. one can claim a method, or an apparatus, without having to demonstrate the code necessary to implement. in some cases it may be easier to get a patent than to create the code that would infringe it.
how is this going to be solved?
trivial patents
not all software patents are trivial but there is no question that a significant number are. whether or not this is a result of error, incompetence, or oversight the fact remains. there is a broad consensus that “questionable” patents are a problem and the proliferation of such patents presents a clear and present danger to software developers.
how is this going to be solved?
for almost all of the brief history of software development, software developers have not had to concern themselves with patents and have lived under accommodating copyright laws. as copyright is primarily directed towards copying, it has been a relatively permissive regime so far as development is concerned.
the imposition of a nominal regime of patent protection on software represents a sea change in this environment and may bring with it some particularly troublesome consequences. moreover, the imposition of a nominal regime of patents on software development will bring the laws of patents into the realm of individuals and very small companies and this has the potential to completely change the dynamics of software development.
for these reasons alone, it is entirely reasonable that many small and medium sized companies are opposing software patents.
until the directives supports addresss THESE concerns, they are wasting their time with us.
and until the directives opponents accept the patentabilty of software and demand action on the practical problems, they too are wasting our time.
as a permanent resident of the EU, the thing is this: software patents exist in the US. the US is the world's largest market. software developers in the US have no choice but to consider patents when developing their products. software patents developed with a mind to patent issues can be sold anywhere in the world.
software developers in the EU, thinking themselves immune to patent infringement claims, develop software without consideration to patents. these products cannot be sold in the US or Japan and this places the EU at a competitive disadvantage.
pity that the FFII is so attached to their radical agenda that they perpetuate this untenable situation. they are making the EU less competitive. their simplistic solution of prohibiting all patents on software is an insult to the intelligence of rodents and human beings alike. and it avoids the real issues that we have mentioned.
on the other hand, the fédération internationale des conseils en propriété industrielle (FICPI), and the association internationale pour la protection de la propriété intellectuelle (AIPPI), and the IIPI to which professor lessig kindly pointed completey fail to appreciate and address the issues we stated above. they are aloof to these legitimate concerns.
it's the stupid versus the stupid.
and it makes our three heads want to explode!
Yes it would, and quite rightly too, because these things are software patents. These things are currently excluded from patentability by the 1972 European Patent Convention, and despite the best efforts of the EPO to push back the boundaries of patentability in Europe to cover software inventions such as those you describe, I think we have a good chance of preserving the very sensible exclusions from patentability that were spelt out by the 1972 Convention.
well, ian clarke, it seems that you have not studied the EPC.
EPC 52 sets forth in broad sweeping terms what is and what is not patentable. it can be compared to 35 USC § 101, but with some differences:
“(1) European patents shall be granted for any inventions which are susceptible of industrial application, which are new and which involve an inventive step.
(2) The following in particular shall not be regarded as inventions within the meaning of paragraph 1:
(a) discoveries, scientific theories, and mathematical methods;
(b) aesthetic creations;
(c) schemes, rules and methods for performing mental acts, playing games or doing business, and programs for computers;
(d) presentations of information.
(3) The provisions of paragraph 2 shall exclude patentability of the subject-matter or activities referred to in that provision only to the extent to which a European patent application or European patent relates to such subject-matter or activities as such.” (emphasis added)
although EPC 52(2)(c) clearly states that “programs for computers” shall not be regarded as inventions, Article 52(3) says that this only applies “to the extent to which a European patent application… relates to such subject matter or activities as such.”
it is incorrect ian clarke to point to EPC 52(2)(c) and ignore EPC 52(3). this was added to the convention for a very specific purpose - to give the EPO wiggle room to decide what is a computer program and what is a technical invention.
programs for computers are non-technical inventions. technical inventions implemented in software are not programs for computers.
in addition to the EPC, the implementing regulations (i.e, the rules) clearly state that an invention are liberally sprinkled with references to the “technical” character of an invention. rules 27 and 29 which define respectively the form and content of the specification and the claims specifically require that an invention have a technical character.
the question is this should speech codecs be patentable or not?
we happen to believe that they should be patentable because there is much left to discover and invent. patents provide the economic incentive to invest in R&D.
you seem to believe that simply because a speech codec is implemented in software, that is should not be patentable. where, pray tell, is the logic in this?
Three blind mice wrote: "you seem to believe that simply because a speech codec is implemented in software, that is should not be patentable. where, pray tell, is the logic in this?"
You seem to believe that a speech codec, even though it is essentially an abstract idea, should be patentable. Where, prey tell, is the logic in that?
I mean, lets not forget the basis for patent law, which is to provide a motivation for inventors.
I personally know people that have worked on speech codecs, and I can assure you that none of them were motivated by the desire to obtain a patent on the fruits of their labour. In my experience, as a software engineer, few authors of software are motivated by patent law, copyright perhaps, but not patent law. Given this, please explain where the public interest lies in granting them a patent on their speech codec?
TBL, I must be brief - I believe your own list illuminates the problem with sweeping the extensive scrope of software patents under the term "invention". I don't mean the following analogy to be inflammatory, but to perhaps cut to the core idea: Our current understanding of "property rights" now explicitly excludes human beings. In the 18th century, it did include human beings. The legal structure was very accommodating - if a tractor or a horse can be a piece of property, why not a person? From the viewpoint of the mechanics of the law, there was no difference. Yet that formalistic similarity masked a profound social policy decision. One could then refine it and say one isn't owning a "person", but perhaps "a claim on legal rights regarding social membership", which becomes practically a euphemism.
If the patented subject matter is indistinguishable from the abstraction collection of knowledge itself, because any "process" is that same collection - then that's patenting an idea by another name. Which can be implemented in terms of the legal system, just like owning a person. But it is a difference in degree which makes for a difference in kind.
TBL, you point out a lot of the problems existing with software patents, and then you simply say (paraphrasing) "but still, software should be patentable, we just have to get rid of these bugs." I'll go over some of the issues, and make my case on why software patents are 1) unnecessary for impulsing r&d, 2) detrimental to SMEs and individual programmers and 3) detrimental to Europe as a whole.
Patents are nominally created in order to give incentives not only to r&d, but also to the widespread difusion of the results of r&d. You buy a monopoly, paying with disclosure of your research. The problems when applied to software is that most often a software patent is not at all useful for a programmer to write code. Because of patent language being convoluted, you get two undesireable effects: due diligence is impossible (there are too many software patents) and the disclosure you're buying the monopoly with is much less significant than in other fields, where it is far more useful.
There is no fair use in patent law, but this would be simple enough to solve: limit patent coverage to commercial applications. I would not consider this a sufficient solution to the software patent issue, because of all the other remaining obstacles, but it would certainly be a good start.
The clean room issue is just not possible to sort out, within the current patent framework. It is not a bug, it's a feature. You're buying a pretty much unlimited temporary monopoly on the process or product in question, and that's the value that a patent has. If you would allow independent replication of patent results to be uncovered by the patent, what would be the point of obtaining one? The reason why patents are desireable to patent holders is their power to monopolize a certain product or process. In other words, giving a "clean room" exemption would be equivalent, for software at least, to say "we're giving you copyright protection and no more", which to me is not a problem, but to those that defend software should be patentable seems to be.
Reverse engineering is not at present allowed for because it is generally the case that, for most fields of technology, patents are useful documents that tell you what you need to know to interoperate. A patent on a pump, for example, would tell you how the pump is built and interfacing it to other things would be trivial. However, because software patents cover an unlimited space of products (a speech codec can be implemented in an infinite number (speaking theoretically) of computer languages, and use an infinite number of library calling conventions, etc) the patent is not a sufficiently descriptive tool for interoperability purposes. That said, there are plans to include interoperability exemptions in the EU's directive.
You say: "as a permanent resident of the EU, the thing is this: software patents exist in the US. the US is the world's largest market. software developers in the
US have no choice but to consider patents when developing their products. software patents developed with a mind to patent issues can be sold anywhere
in the world."
This is not true: the US is not the biggest market. The EU is bigger by GDP, China is bigger by population. I'm not sure what you mean by biggest market here, so I cannot tell which is greater.
Then you say "software developers in the EU, thinking themselves immune to patent infringement claims, develop software without consideration to patents. these products
cannot be sold in the US or Japan and this places the EU at a competitive disadvantage."
There are several reasons why I disagree with you on this point: 1) European companies could develop and sell these programs in the EU without paying patent royalties and pay patent royalties in the markets where they are due, this would make European companies more competitive since foreign companies living under software patent regimes would also have to pay royalties for products developed for the EU. (I may be wrong about this, jurisdictional issues aren't my forte.) 2) The European software industry is unused to patenting software and dealing with these issues, and would therefore be at a disadvantage if a patent regime were imposed on Europe, In addition, patent royalties would, at least in the short term, impose a strong penalty on European companies and a strong flow of capital would occur towards non-EU companies that have been playing the patent game for many years. How this would benefit EU companies (and citizens) is beyond me. 3) patents impose transaction costs both on innovation (applying and enforcing them) and follow-on innovation (patent royalties) and generally make software a much more difficult product to develop. EU companies and the EU market might benefit from faster innovation and product cycles if this artificial transaction cost were not imposed on Europe. Since foreign prior art is generally admited as an affirmative defence against patents, EU companies would be free to utilize these innovations elsewhere, knowing that EU innovations are patent-free in Europe and the world. 4) These transactional costs of patents would require companies to spend money in something different from creating software and marketing it, cost that would have to be born by European software users.
On the substantive issue "is a speech codec worthy of a patent", while I have no intention of bilittling the work of speech codec programmers, my answer would be no. (Note that the arguments made above are "practical" arguments about the wisdom of instituting software patents in Europe and have little to do with this argument.) The reason why I believe this to be the case is because software 1) is already covered by copyright (unlike other industrial products worthy of patents), 2) software is a mathematical construct (more specifically, it can be described as the member of a grammar, or as a finite-state automaton) and I do not believe that mathematical constructs should be made into monopolies (note this applies even more strongly to algorithm patents such as a speech-codec), 3) because r&d in speech codecs happens and has happened without patent incentives (see speex) and there is no reason to suppose that it would suddenly stop.
Software has very fast cycles, faster than most other industrial products, and innovation is a requirement to stay successful in the market. r&d is a necessary condition to remain competitive. Because of this, I don't believe software "inventions" (mathematical objects exist a priori) should be patentable.
ian clarke: You seem to believe that a speech codec, even though it is essentially an abstract idea, should be patentable. Where, prey tell, is the logic in that?
david: The reason why I believe this to be the case is because software 1) is already covered by copyright (unlike other industrial products worthy of patents), 2) software is a mathematical construct (more specifically, it can be described as the member of a grammar, or as a finite-state automaton) and I do not believe that mathematical constructs should be made into monopolies (note this applies even more strongly to algorithm patents such as a speech-codec), 3) because r&d in speech codecs happens and has happened without patent incentives (see speex) and there is no reason to suppose that it would suddenly stop.
if you use a digital cellular phone, you are enjoying the benefits of a technical device known as a speech codec. without it, you be stuck on analog forever.
compressing speech is an abstract idea, the specific way that this is accomplished is a technical thing, invented by man. that mathematics are used to describe its operation, or that it happens to be implemented in software does not make it fundamentally different "other industrial products worthy of patents." drawing a line between hardware and software is impossible.
furthermore a patent on a particular speech coder does not mean that all speech coders are patented.
as for R&D occurring without investment, have you any idea how much investment went in to the development of the AMR speech coder used in GSM? someone has to pay for all those trips and conferences and meetings, someone has to pay the engineers' salaries, someone has to pay for the programming, someone has to pay for the voice quality tests, someone has to pay for the standardization, etc. those people expect a return on their investment.
as for copyright being suitable protection.... you're kidding us, right?
I personally know people that have worked on speech codecs, and I can assure you that none of them were motivated by the desire to obtain a patent on the fruits of their labour.
and we personally know people who have worked on speech coders who were motivated by the need to earn a living.
david we appreciate your reply, but you are taking the easy way out amigo. the issues we described are not intractable - they are not even being discussed seriously because people like you take the simplistic approach to prohibit patents or are perhaps motivated by other ambitions. a significant problem is (from professor lessig's pointer) what joff wild is saying:
While there are a number of people who have genuine doubts as to the subject matter covered in the CII Directive, it also seems to me that many opponents are not so much concerned with the patentability of software but with patents per se. There is a significant political context to the whole debate that those who support the Directive would be unwise to ignore. Basically, their argument seems to be that the Directive is bad because it is supported by multinationals, and particularly US multinationals. It is the kind of argument that is very appealling to a certain type of European politician.
the "certain type of european politican" to whom he refers is colored green, thinks red, and believes that if not for stalin, marx's ideas would have worked.
Mice:
I believe that there are some problems with your take on this that I would seek to draw your attention to (I'm a software developer in the UK, and I've applied for US software patents).
Firstly, software is not like resistors or capacitors. It is like calculus or statistical methods. Software is a notation for expressing algorithms. It doesn't matter whether the algorithm approximates pi, decompresses a DVD in a competitive new way, or draws a dialog in an application, it is an algorith. Algorithms should not be patentable any more than grammatical phrases should be.
Software is writing, not engineering. Software is algorithms, not tools.
Copyright is the correct protection for software, and indeed given that software is functional that is already extraordinary protection.
Secondly, the workshops run by the UKPO show how difficult defining pure software patents are, and therefore how difficult any compromise that allows software patents as long as they are not on pure software will be for the patent offices. When Groklaw and The Register agree on something you know there are problems.
Thirdly, the first two points come together. Non-pure software patents will be misused.
Got a new 3D disk drive design? Patent it, including the software that regulates the drive head. Something like Bresenham should do it.
Got a new cancer drug dispenser keyed to the patient's prior symptoms? Patent it, including the software that it won't operate without. Something like a markov chain should suffice.
(The two algorithms I mention are simple and fundamental, but wouldn't be obvious to a patent clerk dresses up in legalese and buried in a hardware patent application.)
And finally, any software patentability will be the thin end of the wedge. Look at copyright "harmonisation".
You can best protect and reward the creativity of software developers like me by protecting us from software patents. :-)
we mice think it is not only ignorant, but downright insulting, to say that there are no inventions in software, or nothing worth patenting.
As a software developer I'd like to say that's very sweet of you, but do you believe you should be able to patent that sentence?
Or the schema for that sentence? Or a method to generate it?
How about the sentence:
"Swap the nouns for the verbs and reverse the sentence."
Should you be able to patent that sentence, which is after all functional?
Or this one:
"I before e, except after c."
How about these sentence (or statements):
"Where ou occurs, swap it for 1. Where ee occurs, swap it for 2. Where ai occurs, swap it for 3. Then Swap all full stops for the word 'ITRATILANUM'."
"Where 1 occurs, swap it for ou. Where 2 occurs, swap it for ee. Where 3 occurs, swap it for ai. Then Swap all full stops for the word 'ITRATILANUM'."
This is an encoding/decoding routine.
Then we get to this statement:
"In a sentence, if the word 'ignorant' occurs, follow it with the words 'but downright insulting'. Insert a comma between the word 'ignorant' and the inserted text. This method will increase the rhetorical effect over merely using 'ignorant' alone."
Not technical enough. Let's write a computer program to do it.
s/ignorant/ignorant, but downright insulting/
Or we could convert the words to numbers using ASCII and write the instructions in machine code.
Now of course you will point out that this is pure software rather than embedded software. But the problem is this:
The software doesn't care where it is. It doesn't know what it's being used for.
Let's say the patent on the software is as part of a hardware device, or a complex application. If the patent relies on the software being used in a similar context (how will that be defined?) it's useless, a "pirate" can just read the patent and copy the software for a lawnmower or a file copying utility rather than a pacemaker or an image editing program. If it is not context sensitive, the patent is a pure software patent anyway, just in a "shell" to discguise this fact.
This is why arguing for compromise is so difficult: moving halfway from "no software patents" to "software patents" leaves you with "n software patents". And n will be non-zero...
First of all I am happy to see TBM now finally agreeing that there are huge problems with software patents today, basically repeating many of the arguments myself and many others have used against them. Would I be willing to consider software patents if I felt that all those problems could be addressed? yes, probably. But the thing is that I don't really think many of them can be adressed in a satisfactory manner. I also do believe that a system of software patents where these issues are not addressed is more damaging than no software patents at all. So until a credible proposal for addressing these issues is put on the table, I think the best policy for everyone involved is not to have software patents.
Also the argument used in one of TBM posts about european developers being at an disadvantage compared to US companies due to not being able to sell their products in the US due to patents is just silly. How would these developers have less disadvantage if their software where prohibited by patents also in Europe? Or for that matter why would it be an advantage for an European developer to have to cough up huge amounts of cash to get defensive patents not only in the US but also in Europe? Sounds to me European patents would only double the costs of getting defensive patents. It
This blog post is not about software patents, but about an American lobbying group joining the debate.
While I disagree with Mr. Lehman on software patents (I think they should be abolished immediately, worldwide), I am happy to find his organisation joining our opponents.
Usually, the American software industry lobbying for introducing changes that would result in a multibillion dollar license revenue stream to them from European industry has the good sense to hide behind puppets like the EICTA, pretending it cares for the European industry.
Having outfits like Mr. Lehman's speak up for software patents is great for our side. I hope everyone involved in Europe notes their position clearly. Any proposal to do what America does or wants is DOA (dead on arrival) in European software patents politics. Even the few remaining fans of software patents need to pretend they don't want to copy American standards.
Maybe Mr. Lehman could use his Washington connections to have the American government hint at complaining under the TRIPS agreement if the European Parliament doesn't legalize software patents. That would completely nuke all support for forcing them down Europe's throat, while just getting the IIPI officially involved is certainly helpful, but not decisively so for our side.
Really, Rob, I find your arguments well-intentioned but completely missing the mark.
All of your suggestions are already patentable as you describe them, because they are _methods_. Any method of doing something new, useful, and nonobvious is patentable. Of course, most of the linguistic transformations you describe are obvious or useless or both, in which case they would not be patentable...
At any rate, I am continually flabbergasted by the way that so-called software developers or computer scientists could continue to think that software is merely a static description of mathematical fact. It patently is not. Software is data that acts as a _machine_. We are talking about compilable software here. In combination with a computer, it does work. It processes digital (or real) objects, transforms, manipulates, and changes them.
Really, it's amazing that you could assert that something like Microsoft Word is essentially mathematics. That so ignores the meaningful properties of that program as to be equivalent to saying that all machines are essentially fluctuations in quantum gauge fields. It is a reductio ad absurdum if there ever was one.
The entire purpose of the vast majority of software is to do something with meaningful objects, not to declare mathematical truths. To say that software is merely mathematics is ridiculous. To say that it is ultimately mathematics completely misses the point.
My head is spinning, and I'd like some clarifying comment. First, I'd like to propose a hypothesis that software itself, being the expression of an idea or method, is inherently not patentable. It looks like that is something on which we can all agree.
It sounds like the absolute anti-software-patent position is that nothing should be patentable if it can be implemented in software. On the face of it, I think I disagree, and I believe I can speak for 3bm as pretty strongly disagreeing as well.
I have also heard no disagreement that of all the controversial patents discussed in this global debate, none really passed muster as to novelty and non-obviousness. This is supported by the primary economic argument against software patents, that software developers are hamstrung by concerns that techniques they develop in the normal course of their work may, and probably do, infringe one or more patents. This is exactly what is addressed by the "non-obviousness" requirement.
The fact that the inventions are implemented in software does not cause this problem, but the nature of software, in which the implementation of a method or process involves little more than expressing it in a general-purpose language, means that many times more inventions can be implemented and tested than when physical materials must be used up for every prototype.
I can understand, and even sympathize with, the desire to forbid all software patents under the circumstances. But, as above examined, I believe that, strictly speaking, there is not really any such thing as a software patent, and that the real problem is that the statutory requirements for patentability are not being upheld. This problem predates the very concept of software, as Professor Lessig pointed out regarding the Selden patent at the turn of the last century.
Therefore, my conclusion is that, to fix what we all seem to agree is wrong with the patent system, forbidding the patenting of software-implemented inventions is neither necessary (because all the disputed "software" patents also failed the normal patentability tests) nor sufficient (because the same thing happens with non-software-implemented inventions, albeit perhaps less often).
Other than my conclusion, I don't think I've heard anyone disagree with anything I've said.
I'll also point out that I'm undecided about such things as RSA which, despite being a "fact of nature", was nevertheless extremely novel and definitely not "obvious to an ordinarily-skilled practitioner of the art". On the other hand, I believe that PKP ended up benefitting only the lawyers, as I don't think Diffie, Hellman, Rivest, Shamir, and Adleman really gained all that much from their patents. In particular, I am quite sure that the market as a whole suffered from not being able to implement their algorithms without expensive licensing (and interestingly, they didn't pursue copyright, because they gave away the RSAREF package under a GPL-type license).
More light on this, anyone?
By the way, the patents around public-key cryptography were the other half of the story about the legal troubles of the development of PGP by Phil Zimmermann. The usual, but wrong, way the story is told, is that the bad evil government went after the hero of the Republic because his freedom-loving invention threatened to overthrow the Sith Lords Of The Empire. Or something like that. In reality, the government investigation was sparked by RSA from claims of violation of their intellectual property ("theft!"), notably the public-key cryptography patents.
TBM, I have given a more or less detailed reply to most of the issues you raised, whereas you have dispatched me by saying that my motivations might be impure (reflexive anti-corporate sentiment, or some such) without addressing at all my arguments. You're free not to have to spend your time arguing with me, but I'd like you to tell me, if you say those issues can be dealt with, what solutions exist, in your view, to do so.
You say: "as for R&D occurring without investment, have you any idea how much investment went in to the development of the AMR speech coder used in GSM? someone has
to pay for all those trips and conferences and meetings, someone has to pay the engineers' salaries, someone has to pay for the programming, someone has
to pay for the voice quality tests, someone has to pay for the standardization, etc. those people expect a return on their investment."
I don't believe I said r&d can happen without investment, that would be absurd, even if such investment is of the personal time of the developers in a volunteer fashion. r&d does not come out of air. You have imputed this statement to me, because it is absurd and easy to argue against. What I said is this: "3) because r&d in speech codecs happens and has happened without patent incentives (see speex) and there is no reason to suppose that it would suddenly
stop." I did not say that r&d was free, or that investment was not required. I said that r&d can happen without patent incentives, and this is not an ivory tower kind of argument, it is supported by fact. Development of many algorithms and file formats (jpeg, png, Ogg Vorbis, speex, etc) happens without any patent incentive, and because of the need for software to be innovative in order to stay competitive in the market I'm not at all convinced that such incentive is necessary. In other words, without patent protection, someone would have found out that analogue voice is crap, and voice compression can make a lot of difference, and manage to monetize this one way or another (manufacturing chips for the aforesaid compression, selling implementations for this (first mover advantage counts), etc).
Also, there have been those who have said that software is a machine, that viewing it as mathematics is missing the point, or that the fact that something is expressed mathematically is irrelevant for patentability issues. Why I think this matters, and why it matters a lot, is because mathematical objects (automata, grammar members, collections of lambda calculus expressions) are only a combinatorial result, they are a priori existing objects. Before anyone came up with the GSM compression, the mathematical object of the GSM algorithm existed, the same way pi existed before anyone calculated it. In addition, because of the way patents work, they cut off the opportunities to utilize certain mathematical objects, which in my view is as much a clear case of a common, the heritage of humanity, as one can expose: inexhaustible, pre-existing and infinite, what more do you want? A patent does not only close one given mathematical object, though, it closes an infinite number of them. When you patent an algorithm I cannot write any code that behaves like that algorithm, even if my code is different from yours. I don't understand where this right emanates from, except from the practical matter of giving an incentive to r&d, which I have already discussed why I don't think necessary for software.
This is the main difference I see: consider a pump with tolerance pressure X, a certain shape for the input and the output of the pump and a certain efficiency of pumping. If you get a patent on a design for this pump, I may well be able, using different materials or techniques, to create a drop-in replacement for such a pump. Now consider a compression algorithm: it has a certain domain, a certain compression efficiency and so on. When you get a patent on this algorithm you not only get a patent to this algorithm in the way you implemented it, you get a patent on this algorithm in ways of implementing it you might not even know exist. You completely destroy my ability to create a drop-in replacement for this algorithm (I can make a different algorithm, but it won't interoperate with other devices using yours, while in the pump case I could have both pump designs working together with no problems.) If you still don't think this is a problem, then we have little to talk about. If you think it is a problem then do tell why and how you think it can be solved, while preserving the idea of "software inventions" and patentability.
Really, it's amazing that you could assert that something like Microsoft Word is essentially mathematics.
It's more marketing and law, but let's look at the functional side of it. :-)
What are the objects that Word operates on? Letters, words, paragraphs. That Word works on particular kinds of objects is seemingly obvious.
But those particular objects are represented in the computer's memory as numbers. Not ASCII or Unicode, but numbers nonetheless.
Therefore any algorithm that operates on words and letters is in fact working on numbers.
So what? A machine that makes a better mousetrap is in fact a collection of gears and other moving parts.
Well, there are two problems.
The first is that the software patent, in descibing the software, is very much describing those moving parts, not how you make the moustrap per se. Therefore any machine that moves that way will be covered by the patent, not just any mousetrap-maker.
The second problem is that the software doesn't just operate on moustraps. It is, by virtue of the fact that it works on numeric representations of objects, generic. The representation of the data isn't special, it's all numbers or a structure from Knuth.
So we have the contradiction of a description of a "machine" intended for a specific pupose but that can have no specific purpose.
Therefore patents on software are remarkably generic, and they are generic precisely because of the way computers must deal with the specific.
That so ignores the meaningful properties of that program as to be equivalent to saying that all machines are essentially fluctuations in quantum gauge fields.
The meaningful property of a program is that it works on data represented as numbers. Any real-world attempt to constrain what those numbers are intended to refer to will either allow piracy (being too specific) or extortion (being too broad).
A lazy example that I don't have to search for is the Hypercard-affecting patent that Stallman always trots out. But as far as I understood the legalese in my patents, they could have referred to just about anything.
If patenting methods is what's important, their status as software is unimportant and so we don't need software patents. If patenting systems containing software is important, it's unworkable in practice.
Don't let software's marketing as applications obscure the fact of its implementation as algorithms.
Aha! You do make an important point, David: interoperability.
In your example, there is the problem that the form of the algorithm's result is inextricably connected to the design of the algorithm. This is qualitatively different from Microsoft's pathological patent on Office document file formats, which stands as a classic counterexample since it appears to exist only to stifle interoperability and hence competition.
I can't make a convincing argument one way or another about this, but I can certainly see that a PKP-like control of patents of this nature harms healthy competition. I would also observe that it's not necessarily a smart thing to do, because the market will, and clearly has, punished those who tried - witness the hampered uptake of RSA (I know - I was one who chose not to try to implement it due to the licensing issues).
Still, public-key cryptography was a startling advance, and it came not a moment too soon. But again, this is science, right? The bottom line is that scientists generally depend on each other, and their work always builds incrementally - just as Rivest built on Diffie and Hellman's work. Generally, in the crypto world today at least, patenting an algorithm is the surest path to failure in the market, because everything depends on everyone being able to use it, even if it was entirely innovative. The IETF actively shuns patented work unless there's a guarantee of licensing on reasonable and non-discriminatory (RAND) terms, but with the rise of free software even that requirement is being reconsidered as insufficient.
My take on this is still pretty ambivalent, but given the strong mitigation of the negative marketability, I wouldn't really fuss about patenting truly innovative and valuable algorithms. Practically speaking, any such patents would soon be available for a song, because they'd be commercially worthless.
I believe there is already a good bit of history to support this. Just consider the case of LZ77. When the owners of the patent tried to strongarm the industry, the industry said: fine, we won't use GIF any more, we'll just use JPEG, though it's harder to implement, lossy, and sometimes noisy - it's better compression anyway, and they developed PNG for when it wasn't suitable.
I think it's important to consider these issues, and not to rush to a solution that may throw out the baby with the bathwater.
Brian, you say that because the market, basically, tries to route around patents, it is OK to award them for truly innovative software algorithms like RSA (please note how RSA isn't even software, strictly speaking, but a result of number theory). So, what you're saying is that smart agents will not patent things, because the market will oppose them and if they do they will give licences in order to get their algorithms adopted. There are two issues with this approach. One, like it or not, is Microsoft, or insert your least favourite monopoly. Monopolies generally do not have to interoperate, and can afford to ignore their competitors to quite a great extent. In a world without monopolies I'd be less worried about software patents, although I still think they're a really bad idea. The other issue is you say, basically, that software patents aren't useful because the market will refuse them. What's then the point of awarding them? What's the benefit for society? Note this is not a matter of natural right, patents are clearly social artifacts, awarded on utilitarian grounds alone. So, why should the EU award software patents if 1) they can cause a lot of trouble, as exposed above and 2) even when you would argue they have merit, they would not be commercially significant? If they don't make a commercial difference, the utilitarian grounds go away.
I wrote a program in a language called "perl" that prints "Hello, World!" on the screen:
#!/usr/bin/perl
print "Hello, World!\n";
Let's put aside for the moment that this is a perfectly obvious program. I go out and get a patent on it.
Joe Hacker writes a program in a different language, C:
main()
{
printf("hello, world\n");
}
Can I sue him for patent infringement? Does the fact that I wrote a program to print "hello, world" on the screen in one language mean that no one, anywhere, can write a program that does the same exact thing in a completely different manner? What exactly did I patent? It seems to me that either I've patented the IDEA of printing "hello, world" on the screen, or the METHOD of doing it using perl. If I've patented the METHOD, then Joe should be free to write his program; he's not using my method. In that case, my patent is pretty worthless; there's literally dozens of ways to write a "hello, world" program, or any other application. You can even write a program in perl that looks completely different from the one I wrote above, uses totally different commands and syntax, yet does the exact same thing.
I just don't see how a software patent is going to be enforceable, without allowing patenting of the IDEA that led to the creation of the application. And we don't allow that.
Rob:
Patent law is all about patenting ideas. That's the point, and that's why ideas are not copyrightable, only expressions of those ideas. Patents are different. Patents are about ideas that are not tied to a particular embodiment.
In the case of your example, the idea is the method. The method by which you display the text "hello world" on the screen is the idea. But the method is trivial and obvious because it's all implemented by the compiler and OS or BIOS code in your example.
However, if we were talking about the early days, working at a low level, you could have had a real method claim for displaying bitmapped text on a screen. At the time, no one thought it should be done, or was even useful... especially when we already had punch cards. ;) Think about the steps involved in that process on an early computer in assembly or even hex, and you'll see how a method claim might have been constructed.
Maybe you disagree with the entire premise of the patent system, but your post seems to indicate that you're confusing copyright concepts with patent concepts (and I am not denigrating, just trying to help clarify). In patents, anything you patent had better be an idea, and a good one, or it's most likely not valid.
Regarding RSA and it's alleged harmful impact, I have two things to say:
First, it's not really a harmful impact if it didn't exist before (or wasn't public), unless it would have been developed independently anyway right afterwards. From what I know, this seems unlikely. In most cases, we'll never know.
Second, they were not the best businessmen in my book if they set their licensing fees so high that no one adopted it. If they had to set their fees to zero in order to license it, it wasn't a very valuable patent to start with. If they set them too high, they lost out. These days, it probably makes sense to set very low licensing fees for things like that, and make money by broader adoption.
A few comments:Unease is expressed by various individuals (like Mr. Karl-Friedrich Lenz) concerning the intervention of non-European participants, in particular from the IIPI, in the political debate on the patentability of computer-implemented inventions. I do not share such concerns. Patents are a universal instrument used in a globalised world economy, and I do see absolutely no reason why patent politics should not be openly discussed on an international basis. As pointed out by some pseudonymous participant "three blind mice" it would be naive to think that Europe could prosper as a patent-free island embedded in a globalised economy equipped with a patent law also covering computer implemented inventions:
Yes, the EU is also a large market in sheer numbers, and China easily outnumbers the U.S. and the EU even if taken together but it would be naive to deny the technological superiority of the U.S. in many fields. If numerous products made in the EU and embodying computer-implemented inventions patented in Japan and/or in the US could not be exported to those countries because of lacking patent-awareness of European companies this would surely be a drawback for the European economy. And, not to forget, that many products finally assembled in the EU are composed more or less completely from components imported from other non-EU countries. In a globalised economy, this is quite a normal procedure. If patent license roalyties are due in the countries of origin, the European importers will have to pay them anyway.
There are some participants in the debate coming from the programmer's point of view denying that e.g. a codec device made out of a piece of software running on a suitable processor can ever be seen as an embodiment of a technical invention because of software comprises of abstract data structures. I think that this is a misconception. In patent law, software as a lingustic construct is entirely irrelevant. Only software running on a suitable processor can create effects which have any meaning in terms of patent law. However, when software is executed on a suitable processor, a certain real-world semiotic has to be atttributed to the objects defined within the static linguistic contruct which goes well beyond the syntax and semantics of a programming language. For example, if a piece of software comprises a certain object, say, a variable $VELOCITY, this object is embedded into the syntax of the chosen programming language and as well into the semantics of the run-time behaviour of that language. But only if this object within the static linguistic code is directly or indirectly conncted during the execution of the software to some physical entity reflecting the real world (in this case, a velocity value measured by some sensor and communicated to the computer via some I/O channel or at least a velocity value simulating a corresponding physical entity) then we might have a technical environment justifying the application of the principles of patent law within the context of software.
The pseudonymous "three mices" further say:
Concerning the IIPI, I would say this judgement is premature; they are not long active in this particular debate. However, I could understand if someone would be disappointed due to the intellectual lethargy of some long-standing professional organisations like those mentioned by the said posting (others could easily be added). There should indeed be more room for an open discussion of moderate solutions, eg. reforms on patent law exemptions for source code, initiatives to foster collective licensing models well known from the field of copyright also in the patenting business, and the like.
If the European Parliament should eventually insist on a complete ban of any patents on computer-implementable inventions in accordance with the doctrine of Stallman's Utopia, then the final spurt will happen in the EU Council in order to make sure under all circumstances that such irresponsible and misguided policy will not be blessed by the EU Council in the course of deliberations in a Conciliation Committee made up of representatives of both bodies.
Regarding:
"... denying that e.g. a codec device made out of a piece of software running on a suitable processor can ever be seen as an embodiment of a technical invention ..."
My view is that it can be seen that way, just like a person can be seen as a chattel property (with, again, apologies for the analogy). The system can declare it to be so. Yet the *consequences* are more extensive that the mere declaration would imply, because the implications shouldn't be ignored.
Axel,
Whether the EU institutes or not software patents is a matter that has mostly to do with EU policy, and very little with the prevalent climate elsewhere. That said, India has refused to institute software patents, which seems to indicate that Europe is not alone here (and would most certainly be within TRIPS), and it seems no hindrance to its outsourcing industry, as it is questionable and so far unproven that the lack of software patents has damaged European industry.
Many products are simply local in their nature: they are custom code, or they are created to deal with a particular circumstance. These products, which comprise the majority of software code ever written, would benefit from an absence of software patents in Europe. The products that are due for export can pay software patent royalties if necessary when sold under foreign jurisdictions. This would change little if EU institutes software patents, since patents in Japan and the US will not have force in the EU, and there is no patent reciprocity as there is with copyright, so whoever needs to be paid royalties might well be a different institution depending on the jurisdiction. Something might be patented in the US and not in Japan, in which case Japanese companies must pay royalties if and when they export, but not in their home market. The situation for European companies would be quite similar, they would pay royalties only when it is necessary.
Whether the US has or has not higher technology is not necessarily a relevant point. In fact, software growth is more likely to happen in low-tech countries, where much of the infrastructure that is taken for granted elsewhere does not exist yet. In any case, the EU by itself is a very significant market. Net flows of capital if software patents are instituted would certainly be in favour of the companies that have been playing the patent game before, which are not by at large European companies. I'm yet to see why this is a good thing.
The problem in trying to talk about semiotics and technical effects of software when running in a computer is that all software, by that token, has technical effects. Obviously, and until the metaphysical computer becomes a reality, computers are programmable machines. When software is loaded in their memory, stuff happens. Logic gates have current flowing through them, heat is generated, and so on. When I run a computer program, the very least it will cause is the values written on memory to change, and a certain pattern of activity in the processor. If it has more specific outputs, like, say, it prints messages on the screen, it will alter even more components. So when it is said that software is not to be patentable as such but only insofar as it has a technical effect when loaded into a computer this is entirely vacuous, because all programs have technical effects by this meaning.
In addition, the proposed EU directive institutes bans on publication. This is very clearly a matter of software as a linguistic-mathematical artifact. Publication has not and cannot have in itself a technical effect. If defenders of computer implemented patantability want to be taken seriously they should not ignore the fact that the directive as it is written would permit such patents as the one-click shopping, which is entirely lacking in any technical effect whatsoever (technical effect is supposed to be about industrial application and the use of forces of nature).
Also, I think it is interesting that you consider that the European Parliament, as embodying the will of the people of Europe, is a toy institution to be skipped over and "guided" back to responsible and adequate action by the Council, sovietist body per excellence, constituted by appointed officials. Of course you're free to hold this view about the European institutions, but I do not think it is a view that most European citizens share in.
David,
- At the time being and for all forseeable future, the European Union is not a national state having a Parliament with full powers. This is simply a matter of fact; please read the treaties constituting the EU. A lot of power remains with the (democratically elected) national Governments, i.e. the EU Council.
- Of course, each and every software running on a suitable processor does have technical effects. But the EU directive as adopted by the EU Council just says that not each and every technical effect generated by running software justifies a patent. Only if a specific technical effect created by running some piece of software on a processor makes a contribution to (technical) prior art, a patent can be awarded.
- The patent system is crucial for computer-implemented inventions embodied in embedded systems, e.g. in telecommmunications equipment or automotive cars. Maybe that the role of the patent system could be seen differently for software dedicated to run on a general-purpose PC. However, when restricting the discussion to the material criteria of patentability you will never manage to establish a proper separation between both classes of applications. This could be made, if so desired, only on the side of the legal effects of granted patents. In the copyright world nobody discusses whether an opera, an architectural blueprint or a novel, all these entities taken "as such", should be subject to copyright. They simply are all copyrighted. Problems with the needs of certain classes of users are, in Europe, solved by a well-balanced system of statutory "exceptions" and "compulsory licenses". If there should be specific problems resulting from the encounter of the software world with the patent system, these problems could be solved by adjusting exemptions and compulsory licenses, not by adjusting material criteria of patentability. In my opinion, this is the core flaw underlying the EU Directive. I would not be overly depressed if the Directive should fail entirely - no Directive surely is much better than having a misguided Directive generally preventing patents on computer-implementable inventions as desired by some MEPs in JURI, effectively in accordance with Stallman's Utopia. Without the Directive, the Patent Offices would simply continue to grant patents on computer-implemented and/or computer-implementable inventions in accordance with the respective applicable law.
- In my view, more than five years of political discussion on the consequences of the encounter of the software world with the patent system have effectively been wasted because of the central question discussed during this time (the material criteria of patentability) is not suitable to solve any real-life problems. All is fatally flawed because of Mr. Stallman's ideologically conceived utopia was willingly adopted by Mr. Pilch, the founder of the FFII, and his followers. A five years discussion on excemptions might have lead us to conclusions concerning a balanced situation which might, as a compromise, be acceptable for the software guys as well as for those industries which need to make use of the patent system.
Note that software on a general purpose computer used to implement the "same" thing as what might have previously been done in hardware IS materially different, in ways that translate to vast cost savings for companies who choose software+computer instead of custom hardware. Even disregarding the severe civil liberty concerns over the patent system in general (why the hell should ANYBODY have any right to stop me building something with my own time and materials?), there are clear economic factors that argue against the "patents are needed to recoup cost of investment" argument from patent lawyers, EVEN IN THE EMBEDDED FIELD.
1. Software writing is significantly easier and cheaper to write than hardware design. It's still unbelievably manual and difficult, but far easier than hardware. Saving #1
2. Software can be upgraded in the field. Consumers have been shown to tolerate "firmware upgrades" of embedded devices (e.g. phones, satellite decoders) and endless patches to PC software (in my opinion wrongly, they forgive errors far too easily in software thanks to Microsoft convincing the world that software has to be buggy), and some devices automatically upgrade themselves. This is many orders of magnitude cheaper than product recalls. Saving #2
3. Advances encoded in software don't necessarily need new hardware, company doesn't have to redesign as often to bring "new" products to the market. Saving #3.
4. Software is more versatile, and the same hardware design can be use for multiple functionalities. Saving #4.
To argue that something should be patentable in software because it's patentable in hardware is misleading: Industry is moving AWAY from custom hardware. It should be the other way around - the computer-using industry is adopting more and more of computing industry's techniques, so protections for hardware should become more like software. The computing industry should not be forced to cripple itself with software patents because the dinosaurs can't compete and because patent lawyers like Axel Horns want to line their pockets at software developer's expense.
To Ted Appleby:
This is a feature, not a bug, of the patent system in all fields of technology.
Contrary to some allegations coming from anti-patent campaigners, enforcing patents on computer-implemented inventions is not at all something like a "dispossession" of the programmer, taking away his property or so.
Programming a (Europe: technical) solution on a computer (Europe: in a non-private or commercial context) without having checked the patents situation concerning the ideas and principles of the solution to be implemented beforehand is like building a house on a piece of real estate without having checked the land register and having talked to the property owner in order to get his consent.
To Ted Appleby, again:
- Also in the hardware world a patent is not awarded for the tedious, boring, and sometimes expensive work of bringing a chunk of metal into desired form by grinding or millcutting or so but for a bright idea conceived before doing that work. The same is true in the software sector: The invention which might be patentable is the entity from the realm of intangibles which the inventor (read: engineer, developer, programmer or who else) had before doing the actual implementation. A patent is not necessarily a reward for a huge subjective burden of work. A patent is an incentive to start thinking in unusual ways.
- Really, are YOU the guy who wrote such a statement?
"They should be taken out and shot as traitors"?? Really? Tsk. tsk.
Axel: "Programming ... without having checked the patents situation concerning the ideas and principles of the solution to be implemented beforehand ..."
Here, I think you illustrate exactly where there is a qualitatively difference extension of propertization. It *can* be done, but it is far-reaching - that the use of the word "process" to cover both physical manufacturing, and manipulation of abstractions, masks an extensive unbalancing.
The way I think of it, a patent is a levy on the cost of processing (not implementation, processing, since the patent covers every execution of the process). Physical processing has a significant cost, so the ratio might be an acceptable trade-off. Abstract processing has essentially *zero* cost, leading to a different outcome. This difference *should not* be ignored by saying they are both "processing", so what's the difference - a cost ratio of small vs. infinity is a very important practical difference.
Seth:
I can't buy that. Take for example - a number of years ago - a team of expert engineers working on that what today is known as GSM mobile telephony specification. Assume they are faced with the problem of creating a method of making a "handover", i.e. that a phone call is not interrupted whenever a cellphone is moved from one cell to another. Several person-months or even person-years of R&D later they write down the specification of a "handover" method which is relatively easy to implement even in a hardware environment with restricted capabilities and which nevertheless really works.
Their employer files for a patent which, of course, falls into the category of patents on computer-implemented inventions.
I don't see a zero cost situation.
Even if the main inventor has invented the workable resulting solution (which then later on was cast into the final version of the GSM standard) during the end phase of the R&D project within five minutes when having a likewarm shower bath at home - the entire R&D process for finding a workable "handover" solution might have taken months, if not years, and might also have cost a lot of money.
Patents can be a tool to re-claim this money.
Ever if a technology turns out to be a commercial success (like GSM), there is an awful lot of wannabe free-riders out there on the markets which might consider (of course only hypothetically, ha, ha) that it could be cheaper to give tax-deductible donations to FFII (for efforts to destruct the patent system) than to play the patent game and, in particular, to pay patent royalties for a period of twenty years to the patent holder (which might well be a direct competitor).
Sigh. Processing, not implemention. Processing, not research. Processing, not customer-service. A patent is a levy on the *process* - not implementation, not research, not customer-service, etc. As you outline, there is a generic propertization argument, which runs that any development has costs overall, thus society should create a granted monopoly of some sort to support these costs. This generic argument should not be regarded as sufficient in itself and overriding all other considerations of scope and implication.
The challenge to those arguing against software patents is to distinguish factors affecting software from hardware. Articulating these differences can be difficult, since there are subtleties involving standards, which have been examined at length in this discussion.
But one key aspect that appears repeatedly, in the League For Programming Freedom publications, in other commentators above, in my own point - is that an abstraction has different *processing* cost characteristics from physical manufacturing, and so there is a fundamentally different outcome in monopolizing it (different, as in very expansive, not can't-be-done). The generic propertization argument does not rebut this, because the generic propertization argument formally works for anything and everyhing.
Re Axel Horns: Really? Tsk. tsk.
"X should be taken out and shot" is a standard colloquialism in British English, try googling it - e.g. "Any software manager who has not read this book should be taken out and shot". I realise you are not a native english speaker and may not know that. Hey, then again, some slave traders were shot, and I view the restrictions of patent law similarly (note that I AM in the tiny minority of people who would like to see an end to all patent law, not just software patents).
But you miss the point: Me having to check patents (presumably actually by employing one of your disgusting brethren IP lawyers to run a check in reality) before developing software is NOT like me checking the property register. Real property is scarce. If I build a house on a piece of land, you can't also build there. It makes straightforward sense that society needs to arbitrate this. But if I build a machine similar to yours, you can still build your machines, If I write software similar to yours, you can still write your software. Economists speak of "free markets", and free markets are what patents are antithetical to. All the USA's talk about "free trade" is flagrantly hypocritical while corporatists are doing their best to _destroy_ global free markets via organisations such as the WTO and WIPO.
One thing I have noticed lately that the people arguing the hardest for software patents seems to mostly be lawyers. Which of course makes sense as without software patents the need for their services would be less as there would be no need for engineers to hire them to help with search for prior art, patent registrations and patent litigation. Maybe these laywers you should be honest enough with themselves to admit that its their own pot of gold they are protecting not engineers or the european economy.
I have done as much arguing on this matter as I can deal with. Many of my arguments have gone uncontested, which is fair enough, not everyone has as much time as me for this. I'm just going to try to say a few more things, and then it's done.
Axel: The European Union is not yet a sovereign entity, although the new constitutional treaty will move in that direction and the practical realities are that the EU, and by extention its legislative body, has substantia