Thanks Billy for that link to Stallman's speech. To quote him on some important points...

On searching for patents:

Quote:

There's more than one way to describe the same mathematics, and as a result you can't reliably find all the patents that might apply to a given design. For instance, there was a patent in the US covering natural order recalculation in spreadsheets. This simply means that when it's time to recalculate everything it does each cell after the cells it depends on... and that way recalculating everything once, it gets consistent results. [...] [The patent] didn't mention the term "natural order recalculation", it didn't use the word "spreadsheet". In fact, what this patent covered was the algorithm known as "topological sort". This patent has dozens of different claims and each claim describes one way of setting up that program. I'm sure he tried to find every possible variation and make another claim about each possible variation so there would be no way anyone could implement topological sort without getting sued by him. But the patent did not mention the name "topological sort" even though that's the name that's been used for this algorithm since the 1960's. So, if you had been working on a spreadsheet and you had tried to search for patents relevant to spreadsheets you would not have found this one. The only way you would have known about this patent was if somebody happened to mention to you that people were getting sued... and they were. So you can't expect searching to find you all the patents that might in fact prohibit a given program.

On the purpose of patents:

Quote:

Patents are written in a twisted kind of legal language where the words don't mean what they usually mean. It's hard as a non-expert to understand what the patent prohibits. Sometimes it's hard for anyone to understand. [...] According to [an Australian study on patents], a tiny fraction of engineers -- a few percent -- ever read patents. One of the supposed purposes of the patent system is to disclose how to do things, but in fact patents provide such poor information about how to do anything and are written in such a way as to make it so hard to extract that poor information that it's generally not worth the trouble. And engineers know this and they don't. And they quoted an engineer saying "I can't recognise my own inventions in 'patentese'."

Clearly, a patent application is not an academic paper.


On pending patents:

Quote:

Patent applications now being considered are kept secret by the patent office. They may issue as patents tomorrow or a year from now, and there's no way you can find out. You can't check. So, there might be nothing that tells you a certain thing is forbidden or is going to be forbidden. You'll write it into a program, you'll release the program, people will start using it, and then you find out it's forbidden.

On this matter, the US PTO itself states:

Quote:

Most patent applications filed on or after November 29, 2000, will be published 18 months after the filing date of the application, or any earlier filing date relied upon under Title 35, United States Code. Otherwise, all patent applications are maintained in the strictest confidence until the patent is issued or the application is published. After the application has been published, however, a member of the public may request a copy of the application file.

Quote:

Original post by chronos Thanks Billy for that link to Stallman's speech. To quote him on some important points... On pending patents: Quote: Patent applications now being considered are kept secret by the patent office. They may issue as patents tomorrow or a year from now, and there's no way you can find out. You can't check. So, there might be nothing that tells you a certain thing is forbidden or is going to be forbidden. You'll write it into a program, you'll release the program, people will start using it, and then you find out it's forbidden. On this matter, the US PTO itself states: Quote: Most patent applications filed on or after November 29, 2000, will be published 18 months after the filing date of the application, or any earlier filing date relied upon under Title 35, United States Code. Otherwise, all patent applications are maintained in the strictest confidence until the patent is issued or the application is published. After the application has been published, however, a member of the public may request a copy of the application file.

United States Code TITLE 35 > PART II > CHAPTER 11 > § 122 provides further information (regarding requests to publish earlier, and non-foriegn applications not being published).

Those interested in how many applications are published before 18 months might want to search the PTO themselves using "2005" (enter in first box and hit search). You'll find ~317,286 hits (not all hits may be for application date: searching by the application date field did not appear to work).

If Stallman had stated, "some patent applications are not published before being issued...", he would have stated the facts correctly. The statement:

Quote:

Patent applications now being considered are kept secret by the patent office. They may issue as patents tomorrow or a year from now, and there's no way you can find out. You can't check. So, there might be nothing that tells you a certain thing is forbidden or is going to be forbidden. You'll write it into a program, you'll release the program, people will start using it, and then you find out it's forbidden.

is not accurate and is misleading. The following is true:

1. Some patent applications will be published shortly after the filing date (a few months in some cases).
2. Some patent applications will be published 18 months after the filing date.
3. Some patent applications will not be published until they are issued.

Additionally, some of the patent applications referenced in this thread were published well under 18 months after filing. 18 months for the life of a patent application can be a relatively short amount of time (patent applications can take many years before being issued).

The ~317,286 hits mentioned earlier may contain quite a few examples of < 18mo applications: an interesting example from search list position 221: Microsoft Geometric Database Search.

Nitpicking.

The point is that patent applications will often not be published until after infringing software has been published. This is not to suggest that it's impossible to obtain some protection by searching patent databases, but rather that a search of such a database is not sufficient protection against the many software patent landmines that software developers must avoid to survive.

Quote:

Original post by chronos Nitpicking.

Unfortunately Stallman is presenting himself as an expert on the software patent situation, and is trying to convince everyone that software patents (and apparently all Intellectual Property), should be abolished (recall his comments to the audience member asking about alternatives to patents: something to the effect of- anyone talking about the existence of intellectual property is (effectively) brainwashed). If Stallman's words regarding the unsearchability of pending software patents were used in a legal contract, they would be either (intentionally) fraudulent or (via omissions) negligent.

When one takes an extremist, fanatical position on one side of an issue, they can appear to be more interested in self-aggrandizement rather than the champion of the issue they profess to promote (this statement has been used against Ralph Nader).

It would be to Stallman's advantage to tell the truth (the whole truth, even if at first it may appear to weaken his position), including discussing positive changes to the patent system (such as the change that brought patent application publication in 2000) as well as positive acts by commercial ventures (such as the world's largest patent holder, IBM working with the PTO to improve the software patent system). Talking about good things that have been happening will help people get involved that may otherwise have believed that change was impossible.

Quote:

Original post by chronos The point is that patent applications will often not be published until after infringing software has been published. This is not to suggest that it's impossible to obtain some protection by searching patent databases, but rather that a search of such a database is not sufficient protection against the many software patent landmines that software developers must avoid to survive.

This point is true for any patented artifact (not just software).

Again, I agree the current software patent system could be improved:

How would you improve the software patent system?
Do you believe that intellectual property should exist?

John.. Here's my idea on how the patent system should work. It might sound crazy at first (like half the things I say), but if you think about it a bit more it starts to make a whole lot of sense. Before I go into details I just post the basics of the idea and see what you think.

Make software patents opt-in.

You opt-in via the act of possession of a software patent.. If you, your company (or parent companies), or maybe even employees own any software patent on anything, then you opt-in. If you are a member of the system you have access to the patent database, if you are not, you have no access.

ie.. If you want the "protection" of the system, you can have it. If you think the "protection" is not worth the cost, you are not forced into the system.

The only other change that would be required is to fix problem with patent trolling (which is already part of the new reforms in progress).

Quote:

Original post by billy_zelsnack Make software patents opt-in.

It would appear that such a measure would only be useful for (some) smaller entities/companies. I've done more research into the "Ghost Car" patents, and it appears that Atari (with some versions assigned to Midway) did a pretty good job of protecting many novel ideas from Hard Drivin'. I must admit, I first learned about the Ghost Car idea from playing their game. They have protected it from imitation, and I believe that is fair. If my customers request Ghost Cars as previously patented, I'll tell them the feature is owned by someone else (saving me extra work [wink]). This will also force me to come up new gameplay elements.

Now let's say that my small company applies for patent protection on a novel invention. The bigger companies won't be able to compete, as they likely all have many patents (forcing them to opt-in). However, the bigger companies may not chose to compete in the same game-space/genre as my game anyway because it is too risky. All of the other small, patent-free developers would then be able to flood the market with copies of any element of my game they choose. This would seriously hurt the uniqueness of my product, with a direct effect on sales (especially if many hobby developers were giving away their clones for free). There would be no way to stop such dissolution of uniqueness, and any market advantage I would have had would be lost.

Perhaps the software patent issues can also help to explain why the bigger software developers don't innovate as much as consumers would hope: they're playing it safe, using IP that is not likely to generate software patent issues.

In summary, I don't believe that any loophole that would allow a competitor to get around IP protection would work. Additionally, some companies would start subsidiary non-opt-in entities, hiding any ties, allowing free dissolution of protected IP, thus hurting their competitors. When it comes to IP and money, some companies will do anything to get ahead (case in point: the well-known Rambus debacle).

What if two companies develop similar ideas and end up with two products on the market, independently developed, with neither copying the other? In this case, there should be a provision to allow, effectively, joint ownership of any patented element(s), provided both companies can reasonably prove they didn't copy the other company (I believe "patent notebooks", logs, and revision control systems can be used in helping with patent disputes and prior art. Check with a patent attorney or search online for more info).

Suggestions to improve the software patent search issue:

Let Google and Microsoft compete for a government contract to provide "nearly perfect" search technology for the PTO (IBM would also be involved in the design of the system). This technology would be used to strongly enforce new patent laws:

1. A patent applicant submits their application in digital form (say, PDF).
2. The novel search engine automatically searches for possible prior art matches.
3. A list is returned with all possible, relevent patent applications and existing patents.
4. If a patent infringement suit is filed in the future, and the patent owner's patent was not provided in the automated list, they cannot sue for damages.
5. Companies not wishing to patent their software could also use the system, prior to publishing their software, to get protection via this patent search/list system.
6. The system could be extended to index into the existing, prior art databases around the world.
7. In cases where human review shows the search system to be broken, the system will be updated to prevent the case from reoccuring, and only the first entity involved will be exempt (future matches will not be exempt).
8. This system would also greatly help in invalidating bogus patents with prior art (including multiple patents covering the same inventions).

As interesting as this is from a physics point-of-view, this is not a physics or math thread. It belongs in the Business* forum, and so I am moving it!

Way to kill the thread! Nobody reads the business section. haha! :)

Quote:

Original post by billy_zelsnack Way to kill the thread! Nobody reads the business section. haha! :)

Well, on a positive note, since there are very few posts in this section, threads won't be scrolled off into oblivion (as quickly happens in the lounge).

Patent System in Disarray article on Information Week.

One of the ideas for improvement in the article is to open up third-party examination and prior art submission before the patent is issued (sounds like a good idea to me [wink]).

This article (and previous posts on this forum) sparked an interesting idea: since patent claims are based on boolean algebra (and effectively a specialized "patent" grammar/language), I wonder if it's possible for someone to exploit the system so that the patent claims become generative, broadly extending the scope of the claims (perhaps via using nested and complex Markush claims). This is something a clever programmer might exploit (in the same way the C/C++ preprocessor was exploited to implement Generative Programming).

NTP vs. RIM, latest on Tom's Hardware.