At 05:33 PM 1/16/97 -0800, Bill Stewart wrote:

At 08:13 PM 1/14/97 -0800, jim bell wrote:

...

US. Alone, that would have denied non-US patents to RSA. However, such an explanation grandly ignores the fact that computer software (let alone mathematics in general) was not considered patentable ANYWHERE (?) before public-key systems made their appearance in 1976. ... I'm still waiting for an "innocent" explanation for the US patent office beginning to issue software patents. I don't think there is one.

The first software patent was Dennis Ritchie's patent on setUID.

So what year was that?

Computer security is one area that, if you're going to allow patents on software at all (which I think are a bad idea, but that's politics)

No, it's not JUST politics. It's also (im)practicality. Even if, arguably, software OUGHT TO be patentable, going from a time in which software wasn't patentable to software patentability is essentially impossible. First, it requires (or, at least, ought to require, if they were to do it right...which they didn't) a searchable compilation of all previously existing software ideas and constructs to ensure that obvious or previously-existing ideas were not given patents. Since prior to about 1976 there was no motivation on anyone's part to collect such a database (secrecy being the most useful protection) then it was practically guaranteed that the Patent Office would be unable (or unwilling???) to refuse to grant patents on old ideas. Further, unlike a lot of big-ticket engineering, programming can be done by individuals with a relatively low-cost computer. Many if not most of the software constructs that were later called "patentable" were probably actually invented by people who had no interest in patenting them...and the few greedy ones who did were given the rewards. That's not fair, particularly when they may have been invented during a time in which software patents were known to be impossible to get. Also, since unlike other areas of engineering current programmers didn't have to pay for previous software constructs ("do-loops were free") there should have been distinct limits on the royalties that anybody could ask for in software-patent cases. Another problem is that much of the costs of most software companies are personnel and advertising. Any realistic royalty system (even assuming all other problems were solved) would have to determine what portion of a program's functionality was attributable to a patented concept, which I assert is essentially impossible. Do you charge for execution time? (the guy who patents the "wait loop" would make a fortune!) Length of code? Number of times called? Royalties must, presumably, be limited to a reasonable and practical level, and also (obviously) they may also have to be divided among a number of potential claimants, in the same way that (for instance) a car manufacturer may have to pay royalties on potentially hundreds or even thousands of patents to sell a single car. If no mechanism currently exists to arrive at a fair division of such royalties (about 20 years after the patents for the various encryption systems were filed) , I think it may be presumed that nobody seriously believes that this is possible. In short, I think it's absolutely foolish to believe that any sort of workable system could ever have been arranged to patent software, even if all of the issues above had been addressed fairly. And the reason I asked for an "innocent" explanation is to give people the benefit of the doubt so that they could show that the system was working. It isn't, and you know it. It's not surprising that no REAL LAWYERS (tm) around here were willing to tackle this issue. They know it's all a fraud.

...

It also ignores the strong likelihood that the reason for the Patent-Office policy change (done, apparently, without benefit of a corresponding law change) was because with public-key/RSA there was finally an example of software the government wished to deny to the average citizen, and the only mechanism (short of secrecy, which was broken) to do so was to patent it.

D,H,R,S, and A didn't _have_ to apply for patents.... With hardware security products, the NSA has the power to seize and classify systems which are applying for patents, for no particularly good reason, and they used to do it often.

Actually, I don't think they REALLY "have the power." What I mean by this is simple: Since American patents are supposedly grantable for applications filed up to one year prior, there is no guarantee that they will even see a patent application to grant until long after the cat is out of the bag. For instance, there was

an analog scrambler for CB radios that got its patent applications seized in about 1980. And most software patent applications for crypto have involved machinations to avoid getting trapped by this, like publishing in foreign journals before submitting applications.

Why should they have to publish in FOREIGN journals? Like I said before, since up to one year can elapse after disclosure (and, apparently, even longer in the case of the DHM patent, submission to ANY journal, not merely domestic ones, should do to prevent NSA's tricks. (and today, publishing on the Internet should achieve exactly the same results.) Richard Schoeppel commented this way: "Roger Schlafly has made a reasonable case that the DHM patent is invalid because the invention was "published" more than a year before the patent was filed. (He dug up a copy of a preprint they were circulating with a stamped received-date of ?May 1976?.)" What are we to conclude? These sleazebags can't even follow their own rules! Look, I asked for an INNOCENT explanation of software patents. INNOCENT explanations don't include the obvious kind of scams pulled like this. Jim Bell jimbell@pacifier.com