patent failure

(From my response to this post on groklaw.net.)

We need to understand that the patent office has been failing it's Constitutional mandate from quite early on.

First, patents should not even issue as often as they did before the 1960s, and things have just gotten worse, especially since software patents were allowed.

Second, the duty of the patent office, when issuing a patent, should be to be sure the patent *teaches* the art. Without that, no one knows what is patented, what was invented, what the patent builds on in the existing art, etc.

Third, no one should ever be allowed to claim things they did not invent. Innovation has lost its meaning in the present economic context of rampant entrepreneurialism. But that is not an excuse. There has to be some real invention. Simply moving ahead in a path others are too busy to pursue is not enough. Simply checking a research path others abandoned as less interesting is not, by itself, enough. Simply getting a common or even less common meme down onto paper is not enough. There has to be real invention for a patent to be issued.

Implementation has to be required. Without implementation, ideas are just ideas, dreams are just dreams. Without implementation, the problems always met in implementation remain unsolved. Without the solutions, there is no invention.

Descriptions of dreams and ideas are worthy of copyright, but not patent.

In fact, most of the patents which have issued would have been well worth copyright in some technical journal. Perhaps there is a literary market going begging, maybe journals for exchanging blue-sky ideas without bothering to adorn them with some sort of science fiction plot.

Independent invention has to be recognized. The US Constitution does not say anything about an inventor being granted a monopoly just because he/she/it registered the patent first. If a second application is made for an invention, and it is clear that the claims overlap, and there is no clear evidence that one inventor is riding on the coattails of the other, the common claims should be co-owned, and both should be allowed to pursue their inventions independently.

Otherwise, there will be some inventors whose Constitutional rights to their inventions have not been secured. There will be inventors who are less equal than other inventors.

And that breaks the foundation clause of the Constitution upon which patent law is built.

The ownership of the patent should not be alienable. It is reasonable for the law to support the assignment of rights to practice some technology under a patent, but the Constitution says absolutely nothing about securing the rights to an invention to anyone but the inventor. There has been much said about the necessity of allowing inventors to monetize their patents, but selling a patent outright has been shown to almost always fail in that purpose, as well.

No corporation or other entity in the industry should be supported in demanding exclusive rights as part of a deal. If assignment of exclusive rights is allowed at all, there should be fairly tight limits and judicial review of the reasons. The way some industries develop at certain times, even a year would be too long. Five years would be the absolute limit, especially in modern industry.

Inventions are never, absolutely never invented in a vacuum. No inventor is ever a sole inventor. History is replete with evidence of this.

Thus, the external society has a completely justifiable claim to some rights to the invention. In order for society to access the invention, particularly after the patent has expired, there should should have been, not just an indexed, searchable catalogue of inventions, but a full encyclopaedic history of inventions and their implementations, maintained by the patent office, from the outset.

Otherwise, there is no promotion of progress. This is a fourth way in which the patent office has failed.

There is an index of patents, but it is just a catalogue of registrations, without even the tags and auxiliary indexes that would make it searchable for someone who doesn't know what to search for.

In general, nobody really knows more than half of the terms to search for.

Patents are supposed to be for inventions. That means that the terminology has to be inventive, as well. Under the present system, independent inventors using independent terminology will file independent patent registrations, and the patent examiners will often not recognize that the separate registrations claim the same inventions.

In its present state, even with the changes that the patent office has (belatedly) instituted to try to deal with the software patent mess, the patent database is not really searchable. That means that a registration can't successfully consider and reference even a significant body of the prior art. And it means that the examiner can't do his or her job in determining that the invention really is an invention.

And lawyers have to be able to understand what is covered and what is not. The boundaries should almost never be open to discussion in the courtroom. Any attempt to define the boundaries of a patent in the courtroom should require consultation with the patent office, and narrowing rulings from the courtroom should be attached to the patent registration.

Rulings which expand claims should never be allowed when a patentee is trying to enforce the patent. That's post-facto -- too late, un-Constitutionally so.

Patents are often explained as a trade. Society protects the patent in exchange for getting access to the technology, particularly after the patent expires.
Without such a catalogue and encyclopaedic history, there is very little way for society to recover the value supposedly traded.

There have been private efforts in this direction. (See, for example, the CRC handbooks.) But they are woefully incomplete.

We talk about underfunding the patent office, but that's not the only thing where Congress has failed. Congress should have mandated the encyclopaedic history from the outset, the tagged, searchable catalogue, and the staff to construct and maintain it. The technology to build and maintain such a database was not as advanced as it is now, but Congress and the patent office didn't even try.

That was a failure then, it must be corrected now.

The costs can and should be born in part by the industries. The construction of fee and tax schedules is an unsolved problem, but it is a problem that must be solved if we intend to maintain the patent system.

The concept that the lack of strict protection was necessary to foster the environment in which technology could be developed was suspect from the outset. Examining the protectionist rhetoric generally reveals underlying assumptions against the concepts of freedom upon which the US Constitution was constructed.
And, looking at history, one does not have to be paranoid to wonder whether the protectionist arguments were not part of plot to turn patents and copyright into an extra-Constitutional Intellectual Property right which would almost certainly become an effective back-door for breaking down the Constitution's checks and balances.

At the time the US Constitution was written, the internet was far in the future. Books were not as commonly available, and there were not as many of them written as we have now, by a long shot. Guilds and other organizations of craftsmen had various traditions upon which people of an inventive turn of mind shared their inventions with the rest of the community of skilled workers. These traditions generally did not involve externally enforced patents.

The patent was an artifact of feudal society. Permanent patents were known to be evil, and the Constitution attempts to deal with those concepts in the prohibition of titles of nobility, etc.

We who no longer have the recent context of feudal society to help us understand this may find this difficult to understand, but we can see that the Constitution mentions a limited time for securing the rights of inventors and authors. The time limit was seen as the only feasible way to prevent what would otherwise become an equivalent of a title of nobility.

One final point of failure, but this is Congress's failure -- the limits should vary from industry to industry, and should be adjusted regularly, with an absolute upward limit of the present 20 years. Most modern industries change too fast, and unexpired, undeveloped patents are nothing but legal land-mines, bait for the non-practicing entities. There should probably be some requirement of continued efforts to develop the patent, with default early expiring when patents have remained undeveloped for several years.

For example, if software patents must be allowed (I join with those who argue against it, software is mathematics in far too many real senses.), they must be limited to between one and five years from the date of initial submission. (The problem being how to determine the appropriate duration, since most would only be worth a year, and most companies trying to maintain exclusion would try to opt for the maximum. Of course, if strict exclusion is disallowed, as I argue it should be, there would be less financial incentive for abusively extending limits.)

Going a little to what some will say is the extreme, I would note that the GPL provides a potential model for an alternative to patents, one which I would argue actually has a far greater potential to "secure the rights of inventors to their discoveries".

Here are the reforms that the patent system needs:

(1) More examiners, and qualified examiners. Qualified staff to construct and maintain the databases necessary to administer patents.

(2) Construction of the database, a full, searchable, encyclopaedic, historic catalogue of patents. All patents, back to as far as records exist, and the mandate to never discard old patent records.

All records, including rejected applications, including 250 year old tech, have to be available, or examiners will miss prior art.

(3) Incorporation into the database of extant industry journal articles. (The Library of Congress should be able to provide a lot of these.)

(4) Review of all patents, both active and expired, as part of the process of building the database. The fact that an expired patent should never have issued is definitely relevant when considering a new patent registration.

Members of the industry would be invited to send qualified personnel to participate in the review. I'd suggest drafting, but that would require setting up a framework to legally define the structure of all industries, something the government has to keep its hands out of.

(5) Change the law back to requiring an implementation for registration. Physically submitting a device would not be required, but evidence of the implementation would be. In addition to full sets of diagrams, specifications, and other documentations recording prototypes or production runs, photographs (as appropriate), production run records, and testing records would also be required.

Testing records must show how closely the prototype or production devices or products meet the claims. Otherwise, the context required to discuss the claims in court is just not there.

(6) Enforcing patents must never be allowed to extend to doing things that would break the checks and balances of the Constitution.

It seems obvious, but in the heat of an argument people tend to forget common sense.

(7) Abstract methods and processes must be entirely removed from the body of patentable subject matter. There simply is no way to enforce such things without walking all over the Constitution's checks and balances.

(8) The assembly of a prototype for testing and other non-market purposes cannot be proscribed. Otherwise, there is no way for a third party to test a patent's claims, to try to find a work-around, test safety, or even to check that a different patent does not infringe.

(9) Enforcement which extends beyond the external market must require the same sort of bureaucratic procedures that searches and seizures require in criminal cases. Allowing corporations to abuse citizens' rights when the government is not allowed makes a mockery of the checks and balances of the Constitution.

(10) Software should be disallowed with abstract methods and processes, but, if the software industry really is insisting on shooting itself in the foot with software patents, abstract software patents must not be allowed. The patent must include the source code, specifications must include the run-time system context, and full testing results showing that the claims are met must be submitted.

Software is inherently abstract, and thus really shouldn't be allowed. Abstract patents are too tempting as a back-door around the checks and balances. It's too tempting to try to expand abstract claims in court after the patent was granted on narrowed claims. Thus, either software should be disallowed as patentable matter, or it should have extra requirements, to make sure a patentee doesn't deliberately or accidentally sneak in abstract claims.

(11) Expiration should become variable, with the maximum set at 20 years. I'm not sure how to establish criteria for shortening patent duration. But when almost all the parts of new desktop computers were not manufactured five years ago, it's too tempting to view patents more than five years old expansively.

(12) A registry system modelled after free software licenses such as the GPL and the MIT templates, etc. should be provided as an alternative to exclusive patents. (And with that, I will sometime have to rant about fixing copyright law.)

The purpose of such a registry is two-fold:

The current patent system, even after reform, weighs too heavily in favor of large corporations. A system modelled after the open source/libre software licensing model could be constructed to be far more accessible to individual inventors.

On the other end of the spectrum, industrial consortia could register the technology they propose for standards under such a system, and most of the arguments that we see now clogging the standards processes would become moot.

Popular Posts