How To Improve Patents

Some claim, for various reasons, that PatentsAreEvil; others respond that they're just a GoodIdeaBadlyImplemented. If the latter is true, a better implementation might be possible. What would it be like? Suggestions please ...

Early disclosure. A certain amount of time after a patent is filed, the invention is publicly disclosed so that interested parties (ie competitors) can either a) provide prior art, or other evidence that would invalidate or restrict the scope of the patent; and b) if they have products that would infringe, have sufficient warning and time to redesign. This would end the obnoxious practice of SubmarinePatents
Eliminate black box patents. Patent office is not supposed to grant patents on "black boxes"; however many patents of this nature have been granted. The most notorious seeker of these was Jerry Lemelson ( nicknamed BlackBoxJerry ). He was proficient in the art of writing black-box patents for things that nobody knew how to build, delaying them endlessly until a practical implementation appeared, then seeking royalties. BlackBoxJerry is now dead; however his estate is still collecting royalties on many of his patents.
Eliminate "application" patents. A patent should (IMHO) cover how, not what. If you think of a device/process as a procedure; then the only patentable aspects should be the steps in the procedure, not the function computed by the procedure itself.
Reduce what is granted inventors. Currently, a patent-holder is granted a 20-year monopoly on the production of his/her invention. Perhaps another scheme--statutory royalties (used successfully in the music industry, see IntellectualProperty), compulsive cross-licensing (found in the auto industry), etc. might be better.
Allow joint patents by independent inventors. Legend has it that AlexanderGrahamBell? got the telephone patent because he beat ElijahGray? (who independently invented the telephone) to the patent office by mere hours. In this situation; both parties should be granted patents--and both should be free to sublicense the invention as appropriate.

--ScottJohnson

Based on first-to-file (international style) not first-to-invent (US style); the former is simpler and cheaper (disputes can be settled by checking the dates on the patent office papers, rather than requiring a lengthy investigation), and doesn't have the philosophical baggage of trying to officially state who was the first inventor
Based on a pure registration system: anyone can cheaply and easily register a patent on anything (saving much of the time and expense of the current process, which includes official evaluation, searches for prior art, etc up front), and if someone else disputes it, then they can lodge an objection, citing the relevant prior art or other argument. This may seem silly, but it would save inventors a colossal amount of money, thus opening the system up to users other than the megacorporations, and it would save the patent office time, thus allowing them to do the work they were left with better.
Disputes should be handled (in the first instance) through patent office tribunals, not the courts (thus making the system cheaper for users, and preventing corporations from abusing smaller patent-holders who can't afford to go to court). Appeal to the courts would still be possible, but must be structured to protect the little guy.
Ensure that patents are judged by people expert in the field; much of the problem with software patents is that they have been judged by non-experts, who overestimate the novelty of the invention. And underestimate its obviousness. [I agree. Most patents are not rocket science, and the ideas are fairly common to experts in related fields. To make matters worse, many patents are written in vague terms without any concrete implementation. The "inventor" describes an idea for a solution, without enough technical detail to implement the solution. The thoroughness of the description can be properly evaluated only by experts in the field. In the current system, many patents are empty rhetoric that doesn't help foster anything -- they just creates a legal quagmire surrounding potentially cool technology.]
Enforce the rules about applicability, whereby something is not legally patentable unless it has an immediate application in commerce or industry. Many of the patents of genes were awarded in the absence of any demonstrated applicability.
Limit patents in highly innovatively volatile industries such as software and high-tech hardware. Say 3 years. Patents are supposed to encourage innovation, not stifle it. Patent lifetimes should be tuned to the rate of innovation in the industry. I can live with someone patenting an obvious algorithm if the patent expires in 3 years. (As someone who had a competitor release a clone product within a year, I would've gladly traded ten years on the life of the patent in exchange for having it granted in one year instead of three.) [I agree. A shorter term would speed up the rate at which ideas become public, and therefore would increase the rate overall rate of innovation. 5 to 7 years is probably a good start.]
Get rid of it. Nothing short of this will enable poor countries to develop. (Really? Couldn't these poor countries come up to the level of technology we had in 1983, without using anything but expired patents? I don't think we were exactly dirt poor in 1983.) [No, because although we were doing well in 1983, we didn't have to use our 1983 technology to compete with 2003 technology] (Most of them would still be far ahead of where they are now if they had 1983 technology. Since they don't even have 1983 technology, I can only assume that something besides patents is stopping their development, and that needs to be addressed first.) [Many countries are not bound by patent laws anyway. Many poor countries lack the education to implement such technology, even if they could read the patents. They also lack the manufacturing and farming facilities necessary to support cities.]
Establish a FixedPatentRate.

The evaluation of patents should be based on objective methods, not subjective rules.
A business secret rule. To get a patent on an idea it most be a business secret. an idea is a business secret if it can be used in a business without giving it away. An contra example is amazons one click patent which can't be used in business without reveling it. This rule insure that the community get something in exchange for the patent monopoly.
A bounty system. The patent systems should be a public bounty system. Every patent application should have a bounty on it. The initial bounty is payed by the patent seeker but it should be possible for others to add to a bounty.
Internet system: The patents examination should be made public on the Internet where it should be accessible to all without cost.
Non-obvious tests: First the technical problem should be posted and the should be a period for the public to post solutions to the problem, if a solution is posted. Then the poster get the bounty and the patent is rejected as obvious. But the posted solution remains available to the public. Patents are supposed to be on particular solutions, not on any solution to a problem. How can this proposal be adapted to that?
Novelty test: Then the patent solutions is posted and the should be a period to find prior art. If prior art is found the patent fails the novelty test and the finder get the bounty.

-- KnudSoerensen?

Clever ideas. Things like these would indeed make the standards more objective, and should remove the problem when the patent examiners don't understand novelty or obviousness in the field. The bounties would provide an incentive for random people to find prior art or prove obviousness, counterbalancing the tendency of government to issue as many patents as possible.

In Australia in 1980, a group of people was commissioned by the government to interview people who filed patents and people who had been obstructed by patents. (Their conclusion was that the Australian patent system was never any good to anyone whatsoever and that the whole thing should be scrapped, but the government didn't take their advice). [Can I get a copy of their report? Anyone have a link to it?]

Exactly what to do about the patent system is a very highly complex question, depending on a heap of details. So what we should be pushing for, rather than any specific change in the patent system, is the realisation that information cannot be "owned". Any legislation concerning patents is fine, so long as it springs ultimately from the premise of unownable information. Unfortunately, recently there has been movement towards thinking that "Intellectual Property" can be owned - this is what we need to resist, not any particular implementation of the patent system.

You don't need a "large group of people" in order to arrive at an idea on how to fix the patent system. Ideas can only be originated by individuals. A large group of people can produce consensus, but first someone has to come up with an idea that everyone else can rally around. Anyone might come up with a good idea, so there is no harm in asking for such an idea or in accepting it.

Of course, if you come up with such an idea, maybe you shouldn't put it here. Maybe you should patent it and try licensing it to the government for millions of dollars. System and Method for the Legal Protection of Intellectual Property. Sounds good to me. [Too much prior art. Not that prior art stops anyone these days]

You do need a large group of people to gather information on all the problems of the patent system. That is what we need, and what those at c2.com can't do. Ideas are more likely to work if you've got some information about the problem to ponder while formulating them.

A link about some opinions on patents:

http://www.geocities.com/tablizer/patents.htm

It suggests a kind of "flat tax" to pay for patents. This allows producers and sellers to get started without being bogged down in courts because patent holders fight among each other for a fix portion of the pie instead of involving producers and sellers.

A law should be passed that explicitly bans patents on "business processes" and "goals". It seems these slip thru because the patent office focuses so much on details that they miss really vague claims in the same patent. We cannot rely on their vigilance, so we need explicit laws.

A patent may not have protected the right person in this case. True? http://www.amasci.com/maglev/lev/expose.html#update (link broken as of 2004-03-29; is http://roddriver.com/index_science.html an adequate substitute ?)

The ultimate point of the articles at that URL is "A patent only protects its owner as long as the owner enforces it.", which is obvious. The rest of the story told there is largely crap. Consider, for instance, this quote from a page about Earnshaw's theorem (that the patent is claimed to violate -- which it does not):

"if we don't require a static configuration, then it is possible to achieve quasi-stable levitation with permanent magnets by spinning the levitated object and using the gyroscopic moments to offset the instability. A number of interesting devices of this type have been constructed. This form of levitation is called quasi-stable (rather than stable) because the rotation of the levitating object results in the emission of energy in the form of electromagnetic waves, so eventually the rotation will be brought to a stop, and then the system will go unstable." http://www.mathpages.com/home/kmath240/kmath240.htm

Toys with levitating permanent magnets are much older than 1980, patent or no patent (the patent office is largely incompetent, I point out for those who have been living in a cave)

Patenting is a method for releasing information into the Public Domain while at the same time retaining monopoly power over that information (invention, process, design, etc.) for a limited time. In exchange for this monopoly power, the inventor does not have the option of withholding the information from the public domain but he does retain the right to exclude anyone else from profiting from, making, using or selling the invention.

It occurs to me that there are 2 problems with the current patent system as implemented:

1) The "inventor" must be forced to disclose and release the highest and best use of the invention - and this disclosure should be reviewed to ensure that it is, in fact, a complete disclosure. Subsequent minor changes would not result in re-issuance of the patent. (This should help to eliminate the "perpetual patents" that some companies receive as a result of dribbling out minor improvements or tweaks to an invention every 17 years or so.) [The disclosure must also be reviewed by experts in the field to ensure that it is complete enough to reproduce the invention.]

2) The term of the patent should be reduced to a duration more appropriate to the type of invention. Maybe mechanical device patents should have a life of 17 years, process patents should have a life of 10 years, and software patents (* if this should be allowed at all) should have a life of 5 years.

This doesn't address the problem with patents on things which are obvious or unoriginal.

Most of the above seems to assume that life is too good for holders of patents rather than the other way round. Except for a particularly obvious software monopoly, is this generally held true? Yes. Are people motivated enough to do research in things? No. Most patents provide no motivation, unfortunately. The problem in my view is that there is too much money in some patents and too little in others. Now that makes sense... People might build more for the third world if there was a ratchet with a higher percentage of small revenue and a negligible percentage of large royalties. --AndrewCates

It should be illegal to use patents for suppression. In other words, if you have a patent on something, you should be required to do your best to manufacture it, rather than waiting for someone else to, and then suing them.

One of the very basic problems with modern patents, at least in the US, is economic. Specifically, the patent office gets its money from fees when it issues patents, so it has a conflict of interests. One interest is to maintain a good patent system, and the other is to make money. The very first thing that must be done to fix the patent system is to remove this incentive to issue patents. Not much headway will happen before such a change is done.

Also, I don't know how to accomplish it, but we need to have the patent office step back and raise the bar on novelty and lower the bar on obviousness. The system has just too much bad precedent due to greed and the new-ness of software. A simple heuristic for an old-school patent is "Let's take an expert in the industry and ask him 'Can you make a device to transmit sound from place to place over large distances in a week?'" If that expert can do it and says it was obvious and not-novel, then it's not patentable. If you asked the experts in the day of Alexander Gram Bell to make a voice transmitting machine, or asked experts in the day of Edison to make a machine which gives off light from electricity with a certain efficiency (e.g. a light bulb), the experts would not be able to do it. The same standard needs to be applied today to software. For example, give the following requirement to any competent programmer: "I want people shopping online to be able to use past credit card info and shipping info when making a new purchase". Any competent programmer would say implementing that is obvious and not-novel.

I don't want to quibble over process vs machine vs algorithm, etc. I think the focus needs to go back to experts deciding "Wow. Is this new? Could I have made such a thing given the requirements easily, aka is it not obvious and is novel? Does this have a practical business application now?"

Ex: I could maybe support a patent of "GUI", the very first instance of having windows with items in them (but I hope it would have a short patent life). It may have been novel back in the day. (I don't know. I wasn't alive.) But after that first patent of GUIs, I can't imagine any invention involving a GUI which isn't blatantly obvious, like "one click", "double click", "click and drag", "resizing", etc.

Ex: Amazon's One Click patent. Presumably the internet, GUIs, and basic operating systems concepts are not covered by patents. It is (blatantly) obvious to any programmer to use a configuration file (cookie) to store information to not force the user to enter it again. It does not pass the "not obvious and is novel" test.

CategoryLegal, CategoryIdeaSpace