19 September 2005

Oracle Corporation's Application

Yesterday was the first day since I started this blog that I allowed a day to pass without at least one post. I am sorry about that but it was really not my fault. Sometime after 5pm on Saturday evening, the wi-fi signal of the Admiral House Hotel in Douglas went down and had still not come back on when I checked out at 10am this morning. Now that I am safely returned to Yorkshire, I can post the article that I had intended to upload yesterday.

Not long ago I commented on NMR's Application which was the first decision of the Comptroller following CFPH. Now there is another one and it is even more interesting than the last because of the route by which the hearing officer arrived at his decision. Oracle Corporation's appeal agent the examiner's refusal of a patent for a method of converting text from one mark-up language into another was rejected because the hearing officer, Stephen Probert, interpreted Peter Prescott QC's judgment in CFPH as prohibiting anything that foreclosed computer programs to the public:
"If computer programs are not to be foreclosed to the public, then it is clear to me that I cannot allow this application to proceed to grant. Not only would the present claims (if granted) foreclose computer programs to the public but, on my reading of the claims, there is little or nothing else that would be foreclosed by them."
This argument is attractive but it is hard to see how this reasoning can be upheld. If it is right, the no computer program would ever be patented. It will be interesting to see how the case law develops.

30 comments:

Anonymous said...

If it is right, the no computer program would ever be patented

Which is good! Applying patents to software shows a basic misunderstanding of both notions.

Computer programs are completely covered by copyright law - in fact they have slightly tighter rules than other texts covered by coptyright.

Patents can not be fairly applied to them because it is impossible to define exactly what is patented once program-language translation is brought into the mix, unless you say that only the specific program code in the application is covered. In which case you're back to what copyright already gives you.

The basic misunderstanding, fostered by those who stand to gain immense wealth from the mis-application of patents, is that a specific computer program is an invention, and it is the whole of the invention.

Just as a steam engine design is not allowed to cover all methods of moving a train using a boiler, only a specific method, so too should it not be possible to gain patents on, for example, decoding images that are stored in JPEG format. Any one original method could be, but then again we're back at what is already granted by copyright.

Oh, well. I know no one in the legal system cares about what's right or fair and I know big companies will keep throwing bribes at ministers around the world to get the law changed or overlooked but at the moment we in the UK have the superior system from the point of view of what patents were supposed to be for: fostering innovation, not stiffling it by hugely overblown sweeping monopoly powers that capture entire industries at a swoop.

T Worthington

Anonymous said...

"If it is right, the no computer program would ever be patented"

About time someone saw the light.

No computer program should be patented. Copyright isn't
perfect, but the patenting of computer programs defeats
the very ends of the patent program. In the computing field,
most of what is patented is simply an incremental advance on
previous work. Yet the patent offices still allow them.
Some of the stuff is so basic that it was never published,
or is published in areas that the companies taking out
the patents are careful not to bring to light. The patent
officers, who in Europe are rewarded for passing patents
not for rejecting them, and who in the USA are swamped
and unable to do the necessary background checking themselves,
are allowing insane "inventions" onto the books that will
eventually manage to make writing software the monopoly
of a few wealthy organisations.

Since a patent is supposed to be for a significant advance,
most of the computer patents that exist today should
be invalidated. Supporting that, most computer patents
that are challenged are eventually overturned, but only
at great cost to both litigant and defendent
(over $1 million each).

The patenting of computer programs is a rich-boys game,
meant to lock out the very people the patent idea
was meant to protect. Patents should not apply to
computer programs, and this ruling is a step in the
right direction.

Finally.

John Lambert said...

Thanks for taking the trouble to read this post and leave your comments. There is much truth in what you say. The argument on the other side is also compelling. If I incline more towards one side that the other it is because of the requirement in art 27 (1) TRIPS that "patents should be available, and patent rights enjoyable without discrimination as to ... the firled of technology..."

Anonymous said...

I think that last bit is '... field of technology'. The problem with that argument, though, is that software isn't a technology - at best, it's a set of rules describing a process.

It is quite easy to say, "if black box A - made of electronic gizmos - is patentable, then any black box B - made of software - doing the same task should also be patentable". It seems natural and right that we shouldn't care whether something inventive is made of metal or software. It's intuitive, but wrong.

The whole natural of how inventive something is depends on the ingenuity of the creation, the new knowledge that it gives us. However, there is little new knowledge that can be applied to software: there are terrible patents, such as ARM's patent (also examined by S. Probert), which patents well-known computer science applied to a specific task. This is caused entirely by not viewing software as what it is: algorithms, rules, methods. The only improvements to this art are mathematical ones: algorithmic improvements in time and space (speed of processing & the memory required) and discovery of new algorithms.

I don't think software should be at all patentable, in much the same way I don't think business methods or any other algorithm-based subject should be patentable. That's not to say that the presence of software should be fatal to a patent claim; obviously there are many innovative processes - particularly industrial - that require computers to work. But that is a different kettle of fish entirely, and is not an excuse for allowing patents on software itself.

Anonymous said...

The public disclosure aspect of patents is more than out-weighed by the dramatically larger control a software patent gives. Physical inventions also have the public disclosure requirement. What is the additional balancing factor to compensate for the broader rights a software patent grants?

The reality is that software patents are patents on ideas, not implementations of those ideas. As I said in the first comment, a program only expresses one way of performing a process but the patenting system would cover doing the process by software regardless of the method. This is clearly contrary to the nature of patents, which are supposed to protect (and thereby encourage) new ways of doing things. One does not get a patent on, for example, using clips to hold plastic tubes onto the inside of an engine compartment, you get a patent on your particular clip. But in software this is meaningless: my Forth program to look up customers in a database using a cookie sent by their browser is unrecognisable compared to your C++ program, even if I copied the algorithm. But so too is a totally new Forth program to perform the task without any reference to your original code.

How does a judge differentiate the two cases? S/he can not - many programmers could not. So the answer, according to the pro-patent lobby is to cover the general idea of looking up customers in a database using a cookie - that way you are bound to catch the plagiarist. Problem solved, and tough luck to the guy that did the "clean room" version.

Yet, a Forth or Lisp, or Assembly translation of a C++ program is never a true copy. Finding a way to do something in a different language IS, in fact, a creative process no matter how closely the algorithm is followed and I would certainly stand by a claim that such a translation is a new method, distinct from the original. There will be more border-line cases (C++->Java, C->C++, even C->Perl sometimes) where the question of copyright infringement can be sensibly brought into play, but once again, there is no justification for the massive power of a patent to be introduced.

T Worthington

Anonymous said...

I was more impressed by the little man argument. I believe that this argument from Oracle should have been accepted - and the patent denied because of it. Little, large and average-sized people have been converting between document standards using various mental procedures for a very long time, and very likely using mental procedures similar (though far less formalised) to those covered by the patent - ie there is almost certainly a huge amount of prior art.

Anonymous said...

Yes, it's hard to see what the little man argument has to do with anything since all computer programs are lists of instructions that a human could follow, albeit more slowly, in the same way as a computer. Introducing it simply showed that what Oracle was asking for a patent on is something that people have been doing for years. I've hand-translated TeX to HTML and vice versa myself.

However, the little man argument arose from a High Court ruling, so it's probably just a spurious load of nonsense from a 90 year old judge who may have understood every fiftieth word he heard while he sat for a whole week picking his nose and eating the result in full view of the rest of the court, which is exactly what the judge did in a case where I was a witness. After that performance, and the resulting miscarriage of justice, I have little interest in the drivel that emanates from the High Court or the incompetent asses who sit there.

Anonymous said...

I agree with all of the other posters here - a computer program should not be patentable; my comparison would be the automobile itself: A particular class of program is like a particular type of vehicle - would it have been advantageous if the first person to build a car got a patent on it, leaving everyone else to build busses, taxis and trucks?

Anonymous said...

According to the spirit of the (US) constitutional foundations of patent law (as I understand it), a valid patent on software would have to be a patent on the algorithm, not the result. However, this is not possible in the US, because it's fairly trivial to show that any algorithm is a mathematical formula - which isn't patentable.

I think it's a travesty of justice that, owing to the desire to not block development too greatly via patents on mathematics, computer patents that are even *more* broad have been allowed.

Btw - if valid prior art had been filed on this particular concept, the patent officer might have died laughing at Oracle - this translator, at its basic level, is the same as any computer language compiler - except that it's a MUCH simpler thing, as SGML and HTML are far less complicated than any "Turing Complete" computer language.

Zeth said...

The average man has too short arms to fly if he jumps off a building.

"This argument is attractive but it is hard to see how this reasoning can be upheld. If it is right, the[n] no man would ever fly unpowered by using his arms."

You are assuming the solution here. Just because you have a hammer, it does not mean that everything is a nail. The correct legal protection for computer programs, which are literary creations, ritten in plain text converted to binary format for a computer to read, is copyright.

Computer programs are recipes that the computer follows, the ingredients are no more than a list of mathematical algorithms. A programmer sits at a desk and types the programs into a text editor (these days s/he uses cut-and-paste) in a subset of English, s/he then compiles them into the language that the relevant computer can read.

Some novels are better than others because the writer has more artistic skill, inspiration or time; likewise some computer programs are better than others because the writer has more artistic skill, inspiration or time.

You cannot patent a novel, a recipe or a painting. Why should you automatically assume that computer programs are patentable? Just because America has them? As my mother said, if your sister jumped off a cliff, would you follow?

If you want to patent a computer program then I will patent legal arguments and after you have made an argument in a case, I will sue you for all the money that your company made. Of course, it makes no sense to a lawyer to patent a legal argument, in the same way it makes little sense to many programmers that patents are silly there too.

Anonymous said...

First, thank you for your comments, and for providing a forum for such discussion.

I cannot see your argument for software patents. Your falling back on 'any field of technology' misses the mark because the field of technology here is not 'software' but the binary logic that comprises the processor that is executing the software.

All processors do essentially the same thing, the only difference is the native instruction set of a particular processor. Therefore, software is merely an interpreter, translating human readable pseudo-code to that native language. As any processor has a very limited set of actual operations, 95% of all software is simply repetition of a very small set of instructions. The other 5% is the actual implementation or algorithm, and if the algorithm is indeed unique and new, that is very adequately protected by copyright.

It should also be noted that software does not function at all without the processor to execute it.

Should someone invent a new way of processing these instructions, such as Transmeta's Crusoe chip, then by all means, that should be patentable, but merely organizing a specific order of instructions being fed to that processor, which is what software does, should not.

A software program is ONE implementation of a function. This is exactly what copyright protection is designed for. Permitting a patent on it however precludes ALL implementations. The decision is correct. Allowing a patent on software forecloses all software from the public. It would be like saying the automobile infringes on the rickshaw because both implement the same function of expediting travel on a roadway.

Anonymous said...

"If I incline more towards one side that the other it is because of the requirement in art 27 (1) TRIPS that "patents should be available, and patent rights enjoyable without discrimination as to ... the firled of technology...""

Does Software belong to a "field of technology" in terms of patent law or more specific Trips. The answer is no. It would be useful to codify that common Trips interpretation to stop intentional misinterpretation of trips 27 which is very catchy in a lobbying process. In fact Trips is not even an executive treaty.

Found this on the Net:
http://swpat.ffii.org/papers/europarl0309/amends05/lenz05/

Anonymous said...

"Should someone invent a new way of processing these instructions, such as Transmeta's Crusoe chip, then by all means, that should be patentable"

Actually, in my opinion Crusoe is a bad example and it should not be patentable. It is very difficult to draw a line between hardware, software or a hybrid system. As far as I know, in Europe, at least in some countries, computer chips "as such" are also not patentable. Only the technology to represent information should be patentable - that includes transistors, quantum bits and not much anything else. So the patent has to cover a technology, and not anything else.

John Lambert said...

Many thanks to everyone who has already contributed to this debate.

First, on the status of TRIPS. The agreement is an Annexe to the WTO agreement which all member states of the WTO have ratified. The WTO agreement itself has fairly well developed dispute resolution and enforcement machinery. It would be good to reach consensus on issues where there is divergence. There is machinery for that too. It has in fact, actually, happened. The provisions for implementing the annexe by lesser developed countries was modified recently at Doha.

Secondly, I do not think anyone (even in the USA and certainly not here) argues for patents for all computer programs. The question is whether an invention that satisfies all the other requirements of patentability - namely, novelty, inventiveness and utility - that is actuated by digital code rather than by purely mechanical means should be denied protection simply because it is implemented by software.

Someone argued against following the US lead. It is no longer just an American lead. The Japanese statute has never contained an equivalent to art 52 (2) (c) EPC. Neither do the laws of many other countries including those of some developing ones such as Nigeria, as I noted yesterday.

Finally, an interesting issue has been raised up by a paper published by the American Enterprise Institute and Brookings on the nature of software patents. I mentioned a paper published by those institutes in a post yesterday. Some of you might like to read that paper. There is a link to it from the post on this blog.

Anonymous said...

The question is whether an invention that satisfies all the other requirements of patentability - namely, novelty, inventiveness and utility - that is actuated by digital code rather than by purely mechanical means should be denied protection simply because it is implemented by software.

That is indeed the false question that we are being asked by the pro-patent lobby. It is false because it simply is not true that such a computer program is denied protection. It is, by copyright law. The problem is that there are very very few real programmers in the legal profession or in government so the technicality of why this is as good as a patent on a physical invention is lost. Naturally, the people who do understand this and who stand to gain by being granted massive powers by software patents are not spending a lot of time on educating either lawyers or politicans on this point, but their efforts to secure this power should be a hint that there's something amiss somewhere.

After all, hardly any of the major software producers have actually made much of their money on the back of even the free-for-all American patent system. They haven't needed patents up to now, because they have had the protection of the copyright system. Patents are just a greed-fest grab for protectionist rights to stave off competition.

Every new implementation of a solution in software IS a new invention, by its nature. This is why software patents inevitably lead to patenting ideas, because simply patenting the software gives no more protection than the law already allows, and the law as it stands gives enough protection to allow Microsoft to have become the richest company i the world.

So the answer to the question at the top of this post is "They shouldn't, and they aren't".

A better question is "Why do software inventions deserve more protection than other types of inventions?"

T Worthington

Anonymous said...

Of course, copyright is actually already better than a patent in one very important way: it lasts much longer. Adding patents to this would make software far more strongly protected than other inventions.

Where are the balances for these extra protections?

John Lambert said...

Copyrights and patents do different things:
- copyrights prevent copying and other restricted acts but they confer no monopoly as such;
- patents prevent such acts as making, using, selling or importing anything that falls within the scope of one or more of the claims of the patent specification regardless of whether the infringement is intentional or not.

In fact, computer programs are not the only things that enjoy simultaneously two forms of protection. In this country, for example, aspects of shape or configuration of the whole or part of an article can be protected automatically from reproduction by unregistered design right for up to 10 years from first marketing of a product made to the design.

Design right protects purely functional aspects of shape and configuration as well as aesthetic aspects. Quite different from the position in the US. We have no Baker v Selden.

For that reason, I question whether there is an imbalance in the case of computer implemented inventions as opposed to purely mechanical inventions.

Anonymous said...

Well, we're clearly not getting anywhere here. As a programmer of 25 years I understand the issues and the technicalities of why copyright, when applied to an invention which is entirely written, is at least as good as a patent.

I also understand very clearly the extra power - far beyond what is granted in the physical world - that a patent gives.

And, I also understand that a lawyer is bound to see expanding the patent system as a way of increasing the available work for lawyers.

Unfortunately, it also reduces the available work for programmers since it is effectively a law against thinking ("intentional or not", which means that, for example, there are certain types of multiplication which are illegal in America, even if done with a pencil and paper), which is what we do for a living.

So, either you don't have the experience to understand the technical issues or you don't have the motivation to try, but either way I think it's a bad idea for you to be involved in this debate or industry. The world would be a better place if you concentrated on other areas of law.

I'm not trying to be insulting, I just don't think people that do not understand an industry should be involved in regulating. I wouldn't practice law and I wouldn't be terribly insulted if a lawyer with 25 years of experience told me not to.

Patents are a threat to the whole industry, with only a few companies standing to gain, and even then only in the short-term. No software company has needed patents up to now and yet our industry is one of the greatest success stories of all time, beyond even the original industrial revolution. The push for patent extension into this field is driven purely by a few very wealthy companies who are, bluntly, bribing their way into governmental and international think-tanks in order to preserve their positions in markets. Positions gained, of course, by taking advantage of the freedoms we have enjoyed by not having patents. They now want to pull the ladder up behind them. I am reminded of Disney here, who push for infinite copyright duration despite the fact that they built their company by exploiting copyright-expired work.

Protectionism is regarded as "so 19th century" in trade and seen by WIPO as a stifling trend which only works in the short term. Yet we are told, again by WIPO, the exact opposite in this debate: that huge multinational companies need to protect themselves with vast patent portfolios from some strange unspecified threat. Well, I'll specify the threat: small, new companies with fresh ideas. The breadth and power of patents combined with the ease with which they are granted by badly-trained staff, means that almost no new software of any sophistication could possibly hope to be free from one of these patents. Thus, a company that won't "play ball" with the existing dominant players can simply be stamped out of existence.

This is not a good model for industry or society. Patents for software must be fought every step of the way in order to prevent the wealth that the computer revolution has brought us being vacuumed into the coffers of a handful of companies.

The current attempt in the states to scrap the concept of prior-art is further reason to resist following their lead into the stagnation which beckons at the end of this road.

T Worthington

John Lambert said...

Once again, thank you for expressing your views.

Whether you like it or not the law allows patents to be granted for some computer implemented inventions and has done in this continent at least for many years. My job is to advise and represent people who are affected by that law.

If you don't like the law as it stands, you have a perfect right to use all the political processes available to you in your country to get it changed. I suggest you complain to your Member of Parliament, write to your newspapers, join one of the many campaign groups against software patents and so on. It will probably not be enough for you to persuade your own government because the chances are that it is party to TRIPs which appears to incline towards patents for computer implemented inventions. Your government would have to persuade other governments (including those of countries like the USA and Japan whose patent acts have never precluded software patents) to change their laws. In my capacity as a lawyer, I wish you the best of luck because lawyers benefit most when laws change and what you propose would be a change of the law.

Speaking as a consumer and citizen of the UK, however, I should prefer the law to stay exactly as it is. The purpose of providing legal protection for new products and processes is to encourage research and development in the development of such inventions. Many of the inventions that have improved the quality of our lives in the last few years have been software driven. Often the only component of an invention that is not already in the public domain is the software. Copyright provides protection against copying, but it does not protect against writing a similar program witout reference to the earlier work whereas a patent might. To give an example, I fear that there would be much less (if any) incentive for a motor manufacturer to invest the millions needed to develop a computer controlled system for switching from petrol to electric power on my hybrid motor car if it could not look forward to a monopoly of that system.

One last point, I am not a programmer but I am not completely unfamiliar with the technology. I studied the mathematics - binary airthmetic, Boolean algebra etc - long before I actually saw a computer. I made and programmed a simple machine when I was a schoolboy in the 1960s - years before Bill Gates started work. I used the internet before it was the internet as a graduate student at UCLA in the early 1970s. I am, and have been for many years, a member of the British Computer Society and the National Computing Centre. Can you say the same?

Anonymous said...

For an authoritative view I suggest reading what Donald Knuth has to say about algorithms and patenting:
http://lpf.ai.mit.edu/Patents/knuth-to-pto.txt

John Lambert said...

Thanks. It is an interesting letter. I hope other visitors will read it.

I am not sure that the poor old Commissioner could do much. His job is to apply the law as made by the US federal legislature and interpreted by the courts.

If the law is to be rolled back to what ti was before Diamond v Diehrin the USA or VIACOM here it will require political action.

In view of this letter and perhaps some of the comments that others from the USA have left, I have to ask what has taken so many Americans so long to object to the direction of their courts and Patent Office.

Anonymous said...

"I fear that there would be much less (if any) incentive for a motor manufacturer to invest the millions needed to develop a computer controlled system for switching from petrol to electric power on my hybrid motor car if it could not look forward to a monopoly of that system."

Indeed, and that is why patents on hardware are generally a good and fair thing. A patent on such a device covers only that device, or rather the original parts of its design. A comparable software patent would, by necessity of the fact that there is no objective way to measure similarity between compiled code, cover all such devices, whether or not they actually worked in the same way. That would stifle innovation in the car industry just as quickly as it will in the computer industry.

If even a giant like Knuth's words about the dangers of patents (which strangely echo those of Bill Gates in his classic work of Newspeak preserved at
http://www.bralyn.net/etext/literature/bill.gates/challenges-strategy.txt) fall on deaf ears, then there is no hope in the political arena for the programmer-on-the-ground who sees vast chunks of his area of endeavour being swept into locked cabinets by the miss-application of patents.

Our only hope, really, is to educate industry leaders as to the damage they will do to themselves in the long run by this course. It is a hard struggle, but ultimately it holds much more hope than trying to talk to the sort of politicians-for-hire that are behind this sort of legislation and act as the catspaws for the monopolists.

I, and many others, are fighting that fight for the economic and intellectual good of the people of the world, including the very leaders of industry who are leading it into a blind alley. We are trying to keep software a viable area of new growth for the future, so that its fruits may be enjoyed by the up-and-coming economies of the world as well as those who already have them, in the teeth of those who want to stave off such competition for as long as possible.

"Can you say the same?"

T Worthington

John Lambert said...

I commend your commitment and wish you well.

As I said yesterday, my job is to advise and assist people on the law as it stands.

In my private capacity, I am an active member of a political party that combines social democracy with liberalism. I have published one article on an intellectual property issue from one my party's websites, but it was not specifically on software patents.

I try to keep my political views out of this blog. Not always easy - I slipped yesterday when I was staggered to learn that the majority of the most affluent sections of US society has no provision for long term care. The purpose of this blog, however, is to discuss what I consider to be interesting developments in the law - particularly apparent departures like Mr Probert's decision in Oracle - and nothing more.

Anonymous said...

Sorry for dragging this out even more, and possibly abusing your blog, but I still felt the need to answer this...

"I fear that there would be much less (if any) incentive for a motor manufacturer to invest the millions needed to develop a computer controlled system for switching from petrol to electric power on my hybrid motor car if it could not look forward to a monopoly of that system."

1. I am an experienced programmer and have worked on embedded control systems in the past. I seriously doubt that the software to handle that job would cost millions, though of course the system as a whole certainly would.

2. If you give patent protection to the software required, one single company will have a stranglehold on hybrid vehicles. The softwares design will fall logically out of its intended function, which is the same for every hybrid vehicle designed, and therefore the same basic design ideas will be required in every hybrid vehicle.

If you want to promote competition, apply copyright protection to that software. If you want to stifle competition, grant patents. The copyright grants a monopoly on that system, including derivative works (ie not just straight copies). The patent can easily end up granting a monopoly on all systems capable of performing the same basic function.

What really makes me scared of software patents is not the patenting of genuinely novel ideas. What scares me is that the successful patenting of blatantly obvious, trivial and derivative ideas is already pretty much routine. For example, a successful patent was granted on the idea of putting media control buttons on a laptop. Such buttons had been present on desktop PC keyboards for years before the patent (not to mention on stereos etc) but laptops weren't used much for media playback then. Hardly non-obvious, though.

Anyway, the truth is that there are very few genuinely original, non-trivial, non-obvious ideas in software. The key ideas tend to be the basic building-block algorithms and data-structures that were pretty thoroughly understood decades ago. The new stuff in any application mostly results logically from functional decomposition.

I would guess that there are key heuristics in certain types of systems that might genuinely deserve patent protection, but a lot of the advances in what computerised systems can do have much more to do with faster processors and increased memory capacity than original ideas.

Why might heuristics in particular deserve patent protection? Because when you have to give up mathematical perfection in order to handle practical issues (speed and needed resources), the design process actually has to be much more creative. While all software development is much more creative than generally recognised, the basic ideas still generally derive from the functional decomposition of requirements. Finding suitable heuristics often requires much more than that.

Also, it is generally possible to find distinct heuristics for the same basic task - there is more room for others to create unique alternatives for similar applications.

But even then, it would have to be a non-obvious heuristic. For example, there is likely no perfect rule for deciding when to switch between electric and combustion power in a hybrid vehicle. I *might* concede that a heuristic to make this decision could be patentable, provided it was non-obvious and specific (ie leaving room for others to find alternative heuristics). But I suspect that the heuristics in use for this are pretty obvious, and that the only real variation is in the vehicle-specific constants.

The (now expired) patent on RSA, though, was to some extent justifiable.

1. It is a non-obvious heuristic to solve the problem of asymmetric encryption, as demonstrated by the fact that the idea of asymmetric encryption was around for some time before anyone came up with a way to do it.

2. There are other possible heuristics to solve the same problem, as demonstrated by the alternative asymmetric encryption methods (elliptic curve etc).

I think the term heuristic applies here because there is no perfect solution to the problem 'I want to encrypt some data'. The idea of how to encrypt is the key thing, not any specific implementation of that idea.

Similarly, the solutions to the problem 'I want to compress some data' are imperfect heuristics, and there are distinct ones even for similar types of data (e.g. MP3 and OGG for audio data).

But even in cases like this, there are potential problems related to standardisation - even defacto standards - resulting in stifled competition. How do you sell OGG-encoded audio files to people whos players only support MP3s? In reality, you have little choice but to pay for software to create MP3s (or whatever iPods use) and sell those instead. Where there is no choice, there is no competition. And of course standards are of fundamental importance in computing. Without them, you couldn't even read a web page - necessary standards include ASCII/Unicode, HTTP, HTML, CSS, jpeg, png and so on.

So anyway, have oracle come up with a non-obvious heuristic for 'how to parse and translate ...'? Unlikely. Parsing and translation theory has been well developed for a long time. Most known parsing algorithms (LL, LR, etc) are specialisations of left corner parsing. Even regular expressions operate in a rather similar way to LR parsing. There are interesting ideas in this field - LALR, for example, probably counts as a non-obvious heuristic optimisation of LR - but they have mostly been around for 40 years or more. In practice, most if not all parsing and translation software uses fairly obvious simplifications and specialisations of old ideas.

Anonymous said...

Oops - I think I need to clarify...

"I think the term heuristic applies here because there is no perfect solution to the problem 'I want to encrypt some data'. The idea of how to encrypt is the key thing, not any specific implementation of that idea."

The implementation of RSA does not involve heuristics other than some mathematical heuristics for classifying numbers as prime/non-prime which are besides the point. The idea of RSA can be seen as a heuristic, though, because there is no perfect rule or solution to the problem "how can I encrypt some data" and there is no way to functionally decompose that requirement that leads directly to RSA or any other specific encryption method.

John Lambert said...

No need to apologize. You are not abusing my blog at all. I am grateful to you for your insight.While I may know more about computing than some lawyers I certainly don't claim greater technical knowledge or experience of the ICT industry than you.

If you re-read my original post you will see that my crictism of the hearing officer's decision was not so much the rejection of the application but the route by which he reached his decision. On the facts of the particular case before him, I suspect that Mr Probert was probably right. My concern was that if he is right, it would be impossible for any computer implemented invention to be patented. That would be a considerable change in the law as most people perceive it to be. But it might also have unfortunate consequences for our economy.

Let me take the first point first. Since you post your comments in the middle of the night our time I surmise that you may live somewhere in the Pacific or East Asia which means that your patent legislation is probably different from ours.

Our patent law, which is harmonized across Europe, precludes patents for inventions that are computer programs "as such." That is usually taken to mean that an invention that is nothing more than a new type of computer program cannot be a patentable invention.

However, the legislation goes on to say "the foregoing provision shall prevent anything from being treated as an invention for the purposes of of this Act only to the extent that a patent or application for a patent relates to that [computer program] as such." The problem for patent offices and the courts is to reconcile those almost contradictory provisions.

The solution that has been reached is to allow patents for inventions that are more than computer programs - for example, a flight simulator or a digitally controlled machine - so long as all the other requirements for patentability have been met.

This is a long way short of the American and some other countries' positions whose laws place no restrictions on the scope of inventions that qualify for patents.

When the European Patent Convention was negotiated, computers were big and clunky and the only use for computer software was running a computer. The situation that we have reached today when every motor car and mobile phone has a computer was not not in most legislators' contemplation.

The problem of excluding software patents is that you exclude an awful lot of otherwise patentable inventions such as the ones mentioned above.

That takes me to the second point, namely whether it is in our interests to deny patents for all types of computer implemented inventions. We trade in an increasingly competitive world. If businesses in this continent are to compete on equal terms with those in the USA, Japan and other advanced countries, the protection for new technologies here cannot be less favourable than in the territories of our competitors. If it is less favourable it would be a reason against investing here.

Anonymous said...

"Since you post your comments in the middle of the night..."

No - English, but with sleep disorders.

"... not so much the rejection of the application ..."

Yes, but I wanted to reply to a specific point and was suffering from awake-all-night-with-nothing-much-to-do syndrome.

My concerns about software patents include the fact that international companies just get the patent in America anyway and trade agreements mean patents that could not be obtained in Europe can be enforced here anyway. This leaves European companies and individuals at a competitive disadvantage, and this disadvantage (at least for companies) is one reason why European patent law is likely to be changed to match American patent law more closely at some point.

At the same time, I think it is extremely difficult, maybe impossible, to find any software invention for which the benefits of allowing patenting clearly outweigh the problems.

Valid fears, I think, though somewhat off topic.

Anonymous said...

Oops again - I guess I should have read your *whole* last post before replying.

Mind you, should we really have bad laws to avoid particular competitive disadvantages, despite those laws acting to reduce competition generally, or should we try to get other nations to adopt good laws and encourage fair competition generally?

And now I think I've missed my plane back to the real world ;-)

Simon Hart said...

The decision of Mr Justice Mann in the Macrossan case [2006] EWHC 705 (Ch) handed down on 3rd April 2006, appears to have significant implications for software patenting in the UK. The case concerned the patentability of an automated method of producing the documents necessary to incorporate a company. The system has been deployed on the internet since May 2003 at www.ukcorporator.co.uk.

A Patent Office Hearing Officer had rejected the patent application in March 2005, relying on all three of the commonly raised exclusions in the context of software patents, namely, the mental act exclusion, the computer program exclusion and the business method exclusion. Mr Macrossan sought to counter all three exclusions and also raised an allegation of ‘apprehended bias’ against the Patent Office Hearing Officer. The bias allegation was based upon an alleged secret submission made to the Hearing Officer by the Patent Office examiner and was also based upon a Patent Office internet publication (still appearing at http://www.patent.gov.uk/about/ippd/issues/softpat.pdf ). The internet publication refers to the Patent Office’s “ … strong tradition of rejecting patent applications for software … “ and goes on to say that “… this tough approach has ensured that only patents with a ‘high presumption of validity’ are granted.”.

The Court ruled against Mr Macrossan on two of the three exclusions and also on the bias point. Interestingly however, the Court ruled in Mr Macrossan’s favour on the ‘business method’ exclusion point.

The Patent Office has, for years, been relying on the business method exclusion to assist it in denying all manner of patent applications, for example –



· a method for facilitating online payment transactions between participants in a network based transaction facility (eBay case BL O/314/04),



· an automated food ordering system for cafeterias (Fujitsu case Applic. No. 9912296.2),



· a web-based on-line user interface for enabling a customer to custom configure a computer system (Dell USA’s case),



· an automated reminder and actioning system for use in a patent attorney’s office (Venner case BL O/106/04), and



· an automated system for machine translation of a foreign language communication (Wordlingo case BL O/110/04).



But the Court in Macrossan’s case accepted Mr Macrossan’s argument that the business method exclusion does not apply to ‘partial methods’ of doing business. Mr Macrossan argued that the business method exclusion was directed to entire methods of doing business, such as a method of running a particular franchise business. He drew a parallel with the telephone, submitting that a telephone, assuming for the argument that it was a new invention, would hardly be excluded from patentability merely because it may be used in business. It is merely a ‘tool’ for use in business, rather than an entire business method. The Court accepted these arguments and ruled against the Patent Office on this point.


The bias point has rarely, if ever, been considered previously in the context of a Patent Office Hearing and unfortunately the judgment only partly clarifies the law on this point. Mr Justice Mann at one point said “ … I do not consider that the full bias rule, as such, applies to a hearing of this nature.”. It is apparent from this remark that His Honour thought the rule has at least some application, but unfortunately His Honour did not elaborate. In the final analysis, the Court found that the rule against apprehended bias did not assist Mr Macrossan on the facts of the case. One is left wondering as to the precise scope of rule’s operation, as the case for apprehended bias in the Macrossan case seemed quite strong.

Simon Hart
B Com LLB (Uni of Qld)

Term Papers said...

I have been visiting various blogs for my persuasive essay research. I have found your blog to be quite useful. Keep updating your blog with valuable information... Regards