Joi Ito's Web

Joi Ito's conversation with the living web.

EU Council Plans to Scrap Parliamentary Vote without Discussion


For immediate Release

The EU Council of Ministers is demonstrating that the concept of democracy is alien to the EU. This Wednesday, the Irish Presidency managed to secure a qualified majority for a counter-proposal to the software patents directive, with only a few countries - including Belgium and Germany - showing resistance. The new text proposes to discard all the amendments from the European which would limite patentability. Instead the lax language of the original Commission proposal is to be reinstated in its entirety, with direct patentability of computer programs, data structures and process descriptions added as icing on the cake. The proposal is now scheduled to be confirmed without discussion at a meeting of ministers on 17-18 May, unless one of the Member States changes its vote. In a remarkable sign of unity in times of imminent elections, members of the European Parliament from all groups across the political spectrum are condemning this blatant disrespect for democracy in Europe.

I think software patents are a bad American idea. Japan has followed the US and now Europe is planning to pass one of the worst versions software patent law discarding many of the amendments to limit its power without discussion. Although I'm for many types of patents, I think that software patents are generally bad and in particular hurt small companies and innovation.

via xcasex on IRC (or should I say, I.R.C.)


You know, thinking back to the secretiveness of the formulation of the Bush Energy Plan, I can't help but to think that Dick Cheney could have had a great career in Brussels. The type of secretive, undemocratic maneuvering behind this patent law reeks of Cheney-and if he could find Brussels on a map I might place him under suspicion.

Apparently now software patents are also a bad Japanese and European idea too! ;->

This is one of the things we agree on Joi. Do you think customers can be made to care about this issue? If so, start with a market you care about and ask each of the vendors to sign a pledge that they have not and will not file for patents in that area. Craft the language so that if they break it, it's fraud. Only buy products and services from companies that sign the pledge. For extra credit require personal liability from the execs and board of directors.

Dave: Yes. ;-) Actually, someone on "I.R.C." said the idea didn't start in the US, but I think the US really embraced and promoted the idea.

I'm glad we agree on this. That's an interesting idea. I guess the trick would be how you would get all of the companies to sign it. Particularly since anyone can file patents and big companies with lots of money can move into a space. You can get dumbass patents like the Amazon One-Click patent and the legal fees to fight a BigCo coming after you with a bunch of dumbass software patents can break a LittleCo. I'm not sure whether vendors in a single space would be enough. I think we need to ban software patents and business method patents all together. Until then, I think everyone is going to file for defensive patents, but not waste time suing people if they have a real business.

I think groups like the IETF which have the ability to push for "encumbered technologies" have the ability to make a business case for people not patenting stuff if they want others to use it, but it doesn't always work.

I know that when software patents were being pushed in Japan, they were being pushed as something good for venture business. This it DEFINITELY is not considering the cost of filing, defense, skill required and the portfolios held by the big companies.

I certainly agree with the many observations on how poorly the patent system (mis)handles software patents. But if patents in general make sense, I don't see the logic for categorically excluding patents for inventions implemented as software versus hardware or designs (unless you conclude that patent examiners are genetically incapable of logical thinking when it comes to software).

I think a lot has to do with the cost of doing research on things like new materials, drugs, etc. You need to provide incentives for people to invest billions of dollars. You have to give them some sort of time to recoup the investment. The testing equipment necessary to even experiment with new memory materials is so much money that the patent process is trivial in comparison. On the other hand, the way software patents are used now, the legal costs often exceed the development costs.

In the case of many business method and software patents, they are silly or obvious ideas. Take the one click buy Amazon patent. The cost of coming up with that was quite low and all it required was a bunch of IP lawyers looking for stuff in the UI to patent. This favors hiring lots of IP attorneys, not hiring more engineers. (I know many people will disagree with me on this, and I realize I'm generalizing here.)

In particular, when you are trying to rally around open standards, it's best if the most important bits were not patented. Think what would have happened if Tim Berners-Lee had patented the web? Well, we wouldn't have the web.

I don't think the patent system is based on the assumption that people normally "invest billions of dollars" in each invention. The idea, as I understand it, has more to do with creating an incentive to be creative rather than to get a return on investment.

As I agreed in my previous comment, much of the processing of patents has been, to use understatement, deficient. An invention that is "silly or obvious" will likely not pass the approrpiate legal test - if this test is applied by people who understand the inventive technology (which is supposed to be the case).

I personally agree that the one-click invention appears to be (a) non-novel, and (b) obvious. But as I recall, some parties put up quite a contest to collect evidence to support this opposing logic, but failed to do so.

I believe the patent process has too many lawyers involved, whether we're talking business methods, software or other kinds of inventions.

But again, while I agree with most of your observations about deficiencies, I fail to see the logic in your solution (to simply outlaw these kinds of inventions).

Don Park gives a personal example. I think every decision should be balanced with pros and cons. There are certain cases where I could imagine a the incentive of a patent being required for an individual or a company to put in enough effort to cause innovation, but I believe that the cost to society of having software and business method patents is higher.

I think the cost of coming up with possibly patentable "ideas", the speed at which innovation is occuring, and the inherently low cost of production and manufacture in comparison to the legal costs for the industry make the patent process something that I do not think makes sense.

I agree that "investing billions" is not the original intent. I was trying to point out that inventions that cost billions are, to more patent worthy. But I still think that "innovation" that patents should incent are something like. "Wouldn't it be great if we could record sounds?" or "wouldn't it be great if we could convert sunlight into electricity?" Then toiling for years trying to figure it out. The, "hmm... it would be pretty useful if we could... hey, I know a way to do that!" To me, those are "ideas" that come easily, should be spread and widely used. Cited with attribution would be great, but As Dali says, we all have more "ideas" then we can use. I guess maybe I'm arguing about "ideas" vs "invention" and arguing that there are very few "inventions" in business models and software, and those worthy of being called "inventions" are not common.

Whatever the Council adopts will still need a majority in Parliament.

While the EU is somewhat lacking in democracy because most of lawmaking happens behind closed doors in the Council, in this case there will be no Directive against the Parliament.

And, with the extremist proposal the Council is discussing now, securing that majority would require a large shift in opinion there.


You seem to believe that coming up with a patentable invention is simple - a quick, easy and low-cost "flash of intuition.

My view is that coming up with ideas may sometimes happen that way. But then you must sift through all those "flashes" to identify those that are not just new to you, but which haven't occurred to others. That takes a lot of work, examining what's called "prior art."

Then you need to determine (and prove) that the idea wouldn't be obvious to anyone else (even if they haven't thought of it yet).

Then you have to come up with a concrete way to implement the idea.

Then you have to come up with a way to generalize your implementation (the claims) so you don't end up with a patent that's so narrow it's worthless.

Trust me - that's a *lot* of work (even though it costs in the hundreds or thousands of dollars, not billions).

Now you're very right that for inventions implemented in software, the cost of producing the implementation may be low. But many people don't, I think, realize that you don't have to actually build the invention (whether it is implemented in hardware, software, or whatever) to patent it. All you have to do is design it. Using your logic, you could say that the cost of implementing an invention in that sense is the cost of the drawing paper involved, which is certainly not expensive.

The patent office has developed many areas of skill over the years; these have served it reasonably well. Then, not too long ago, the courts determined that, just because the implementation was in software or business process, patent protection should not be refused (as it had previously).

The problem, IMHO, is *not* that there's anything intrinsically wrong with patenting such inventions. Rather, the patent office skills base wasn't in place (and largely still is not) to properly evaluate such patent applications. As a result, there have been some pretty dumb decisions coming out.

My whole point is that the problem is dumb decisions about the inventions, not bad inventions per se.

Joichi -- I have to say you're just absolutely wrong on this in my opinion.

Software patenting is essential.

Why are we entitled to *patent* the idea of a program?

Should'nt it be that I too am then able to patent my litterary work?
no, is the answer to both of the questions posed here.

first off, software patents are defined as litterary works in the TRIPS and Bern convention, section 12 I believe.

The reason why software houses want stronger patents is most likely because of the widespread practice of "illegal" software, piracy which earlier was used to spread applications to users who couldnt afford it and then cash in on it when that person started to work or had the resources to buy it.

for a good example of why software patents are silly look here:

Why can't we respect the TRIPS proposal, which, as it happens, is being violated by the new directive, the epo and the uspto.

You can't just patent an idea, no matter how original it might be. To even apply for a patent (which by no means suggests you'll get it), you must first "reduce the idea to practice", which is to say, develop a concrete implementation (actual or design) of an original idea. This is usually very hard work, much different from the "lightning flash of intuition" that may have accompanied the original idea.

A literary work per se isn't an invention, even if it is quite clever and original.

The creative aspect of a literary work which can be given protection (copyright not patent) relates to the way you express some idea. Unlike an inventor, the creator of a literary work automatically gets copyright protection.

I hate to be repetitive, but please don't interpret the poor job the patent office has done (though for somewhat understandable reasons) in applying proper patent discipline to software inventions, with any underlying incompatibility between patents and software inventions.

Terry Steichen says 'much of the processing of patents has been, to use understatement, deficient. An invention that is "silly or obvious" will likely not pass the approrpiate legal test - if this test is applied by people who understand the inventive technology .... while I agree with most of your observations about deficiencies, I fail to see the logic in your solution (to simply outlaw these kinds of inventions).'

This is a popular fallacy; that the patenting of software and business methods would be acceptable, if only the 'inventive bar' was raised so that trivial patents were not granted.

The problem with this is that

1. it ignores the fundamental problem with these kinds of patents, which is that they patent _ideas_ instead of physical inventions.

A parallel would be to allow the patenting of plot-lines in fiction, meter in poetry, or combinations of ingredients and cooking methods in recipes.

These are all ideas, transformed into output 'products' by performing them as input on a set of hardware (books, cooking equipment), in the same way as software patents and business method patents are abstract ideas that operate on input, generating output, when implemented on a CPU. So, should they be patentable, too?

Patenting of physical designs is fundamentally different from patenting of abstract ideas in one key way. Physical designs must function correctly under real-world physics, and this requires extensive up-front design and prototyping, before they can be turned into mass-produced products.

Abstract ideas can be developed mentally, and the up-front work required before the idea can be put down on paper (at least in current patent practice) is trivial by comparison.

Consider these EPO patents: EP0807891 (Sun's 'shopping cart' patent) EP0689133 (Adobe's 'tabbed palette window' patent), or the P2P patent referred to by Don Park. The up-front work required to devise these applications is trivial to anyone with a rudimentary knowledge of UI design; the hard part appears to be writing the legalese, and I understand the patent lawyers take care of that part. ;)

Compare with US patent D0450164, a design patent for a Dyson washing machine. The level of detail, and extensive specifications, is massive, and it's clear a lot of work had gone into the process before the patent application was filed.

2. In addition, Terry assumes that extensive prior art searches really do take place. From what I've heard from patent applicants, and from what I've observed in the range of granted software patents, this is cursory at best, and generally performed by the patent lawyer and the examiner, not the applicant themselves.

I've even observed quite a few patents where prior art, cited in the patent, implemented exactly what was claimed!


In spite of your statements to the contrary, you simply can't patent an abstract idea (in the form of a software, business process or other type of invention). What you patent is an invention, which implements the supposedly original, useful, non-obvious (and probably creative) idea. This is not my opinion, it's my understanding of reality.

When you define your "claims" what you're doing is, in effect, attempting to generalize your *implementation* as much as possible. In effect, you are striving to achieve as much abstractness in the *implementation* as the examiner (and prior art) will allow. But you still have to have to include a concrete, non-abstract, specific implementation in your application.

Despite your impassioned arguments, patentable implementations are no longer limited to hardware, but include software and even more general systems. Again, this isn't my opinion or preference - it's (as I understand them) the facts.

There're no rigid stipulations regarding the number and quality of prior art searches. The law simply says that to be patentable, the invention can't already exist, period.

If the pre-patent search of existing stuff (prior art) is deficient, the patent may sooner or later be challenged and removed. The examiner certainly does a prior art search; the wise inventor (with or without an attorney) will usually do the same. [One major problem here, however, is that there's not only a dearth of software expertise in examiners, the databases of prior software-related art are very limited, compared to other types of inventions.]

Again, there are certainly very serious problems in software and business process patents. I'm simply urging that we focus on the precise, actual causes, rather than attacking the concepts.

'Despite your impassioned arguments, patentable implementations are no longer limited to hardware, but include software and even more general systems. Again, this isn't my opinion or preference - it's (as I understand them) the facts.'

Not in Europe, which is kinda the point of this discussion ;)

"mathematical methods, schemes and rules for mental activity, methods of doing business and programs for computers are not patentable inventions", to quote Article 52 of the European Patent Convention according to .

I assumed since we were using US inventors as examples, we were speaking of the U.S. Patent process. My comments obviously refer to the U.S. and not necessarily to the E.U.

Well, you can't in theory patent an idea per se, true.

However, with well-written software patents (e.g., Amazon, LZW and many others), there is simply no way to implement the idea without violating the patent.

In practice, that means that you can patent a software idea. That's the problem.

It's not theory - you simply can't patent an idea. What you patent is an invention, which is an implementation of the idea. When you write your claims, however, you attempt to generalize the implementation as much as possible (so it approaches the breadth and abstractness of the underlying idea itself).

However, it is a rare idea that hasn't been implemented in some form by someone else. The trick is to generalize (via your claims) as much as possible without becoming so general that your claim would exclude these other implementations.

That being said, I think there's a big difference between a core idea, and a process. The process can be defined pretty precisely (which makes it hard to "patent around" it). But the process implements some idea, and it's a lot harder to (legally, via a patent) monopolize the idea.

Maybe if we could figure out how to distinguish a process from an idea, we could sharpen our ideas on effective patent reform (by making it harder to patent a process)?

I think that's the point here. In software, an implementation looks a lot like an idea. Business method patents do to.

I think the other discussion here tracks the open source discussion which is whether these patents provide more value than they destroy. Even if I agree that maybe there are some software methods that might be worthy of patents, I believe that the difficulty in assessing these patents, the incredible value of open source and having software in the public domain, and the small-company "we can't afford patent lawyers" nature of much of software innovation makes software patents a net negative to society.

I think the core problem is less an intrinsic characteristic of sofware and more that the patent examiners aren't sufficiently knowledgeable in the field of software.

Take the (infamous) One-Click patent awarded to Amazon. The underlying idea is to provide a streamlined way for an online user to enter billing information with an online order. The invention is based on a server-side lookup of previously entered and stored billing information.

Simple? You bet. Obvious? Of course - except that the patent examiners didn't realize how obvious it was. If they did, the patent wouldn't - couldn't - have been awarded.

That, IMHO, is the true nub of the problem. (And it's probably the same problem bedeviling the topic of business process patents as well.)

Terry's comment about obviousness doesn't hold up. The 'obviousness' of one-click was tested in court, not just in USPTO. (see , section 1b.)

One of the problems with prior art is that the patent-holder has a much greater incentive to fight denying the existence of prior art than the alleged infringer or original creator has to assert it. If the patent-holder wins, he literally hits the jackpot in terms of royalties and damages.

On the other hand, if the alleged infringer wins, all he gets is the right not to pay the royalty (which was probably small anyway, relative to the total investment in the business) and a warm feeling in his heart that he has done the right thing.

He will probably still have a big legal fee to pay. Added to that, everyone else will now be able to pile into his marketplace by developing a copy of his product.

So it's much easier to do a deal out-of-court in the first place, or just stop using the invention.

To avoid this scenario, many big companies with the cash are going around patenting anything that might just be novel or inventive. They may never get any royalities, but at least they know that if a related issue ever comes to court, they will have a sporting chance at making a killing if the decision goes their way.

This is why you end up with Software Patent Soup, incorporating undoubtedly dubious ingredients such as the double-click patent.

Of course, this is a big boy's game, and you need serious capital to engage in it. And therein lies the rub. The whole mess ends up not protecting anyone and makes it impossible (especially for startups) to innovate.

It is true that if we could somehow separate software ideas from the processes that specifically implement them, then we would be able to solve this mess.

Unfortunately we can't do that. If you want to get Kantian about it, the underlying problem is that software innovation is characterised by the discovery of synthetic, a priori truths, while physical innovation is characterised as the application of synthetic a posteriori, knowledge about the world around us. It is worth having a look at to understand this.

Business method patents are slightly different and are generally not synthetic a priori knowledge.

I didn't say (or mean to suggest) that a court hadn't found the "one-click" patent unobvious. I simply said that the invention was, IMHO, very obvious (and thus not justifiably patented). In other words, while the court sided with the patent holder (Amazon) and the patent examiner, I feel it was clearly wrong.

Why the court came out (IMHO) wrong is at the very core of this issue we're discussing.

In the court's view of the case Antoin cited, the simple fact that Barnes and Noble implemented the same feature, somehow "proved" that the feature was unobvious, which is very strange logic.

It is precisely this kind of (silly) legal logic that allows these kinds of patents to prevail. (I presume Antoin agrees, given his reference to the one-click patent's "undoubtably dubious ingredients.")

Antoin's arguments about the financial incentives of the patent-holder and (alleged) patent infringer are certainly true. But they aren't unique to software patents.

No, it's the Microsoft double-click patent I think is silly. It is silly because there is obvious prior art.

The thing about financial incentives particularly applies to software because a lot of innovation comes from small companies (as opposed to domestic appliances and computer chips, which are more of a big boy's game).

You are saying that in the Amazon case, the court was 'wrong'. But the court applied a similar standard to this patent as it would have for other types of patents.

You are incorrect to say that the court ruled the patent was non-obvious because it had been copied. The main reason cited in the article is that "the expert specifically testified that it never occurred to him to actually make this modification. Further, the court looked to other testimony indicating that at the time the invention was made, one skilled in the art would not have considered the modification." The reason you mention was anciliary.

The court set a precedent. I can't see how you can say the law is right but that the court is 'wrong'. Higher courts don't just enforce the law. They also set precedents that effectively make the law. And Amazon is an important precedent, which other judges will undoubtedly follow in similar cases.

Obviously B+N didn't appeal it, because they didn't feel they would win on appeal. They certainly didn't give up because they'd run out of cash.

If you don't agree with the decision, tell us how you think the law could be clarified to ensure that this mistake would never be made again. (If B+N had been a startup, this case would have almost certainly pushed the respondent into bankruptcy.)

I believe that, in the one-click case, the court erred in applying the legal standard of non-obviousness.

IMHO, that invention was very clearly an obvious extension of prior art (despite the court's ruling to the contrary).

As you probably know, "non-obviousness" is probably the most subjective of all the basic criteria for patentability. Therefore, conclusions of courts (including this one) on this criteria are far from settled law.

In other words, there's still hope for "good logic" pertaining to the existing basic law to prevail in the future.

First, innovation in software is not about some sort of 'pure truth' or anything. It's about concrete results, most of the time. It's about allowing the user to do something, or creating a virtual machine to solve some sort of problem.

As long as there's a real-world application involved, I call that an implementation. A mathematical equation is not an implementation because it's usually perfectly general. But what about a mathematical equation that only deals with rubber temperature? One that contains variables that apply to a real-world situation? Well, I think that could be an implementation of _part_ of a machine to solve a problem if you coded it in software.

So there is a difference. E=MC^2 is not a machine. But if you incorporated it into a program to calculate fuel consumption in an antimatter spaceship engine, it would be.

Second, regarding obvious patents, you guys are talking about an ex post view of obviousness. Many simple inventions are not obvious ex ante. What about the intermittent windshield wiper? Noone had done this for years and years. It was very simple. But it wasn't obvious for some reason... I tend to think the one-click patent is the same. No one had thought of it because they didn't care so much about getting the user's experience down to one click. It's certainly arguable both ways, but my point is that a lot of things that seem obvious after the fact are not so obvious beforehand.

You are right that software innovation is not about discovering 'pure truths', generally. But sometimes it is, and that's when we have the really big problems. Business method patents and UI-type patents can often be worked around. But LZW and RSA are examples of 'pure truths' or synthetic a priori knowledge. It is often impossible to work around these patents.

It's important to accept that there are several different strands within the patent knot. There isn't a single solution to unravelling the whole thing.

I agree with most of what you say, but I disagree with respect to LZW or RSA. I think these are specifically geared towards a real-world result, which is compression of a file, or encryption of a file, on a computer. They are not really that useful, IMHO, outside that realm. And although it may seem that a computer file is simply data, I would argue that it's not the same as "Pi" for instance. It's not a priori knowledge, but an applied algorithm.

1 TrackBacks

Listed below are links to blogs that reference this entry: EU planning to pass stupid software patent law without discussion.

TrackBack URL for this entry:

According to an email I received today from Gay Mitchell, the Fine Gael candidate in the European elections, he supports the government's position on advocating software patents. So that's another person you shouldn't bother voting for tomorrow.... Read More