RE: math patents

[jim bell <jimbell@pacifier.com>]

At 11:21 PM 4/15/96 -0700, Robin Felix wrote:
>At 04/14/96 1457, jim bell may have written:
>>At 09:08 AM 4/14/96 -0800, Lee Tien wrote:
>>>My recollection from law
>>>school is that the law was friendly to math patents in the period before
>>>the Supreme Court weighed in.  There were some PTO denials, which courts
>>>reversed (I think the Court of Claims heard these back then).  So I think
>>>the trend was toward patenting processes even if mathematical until
>>>Gottschalk v. Benson
>
>>I seem to recall reading that one of the breakthrough "algorithm" patents
>>was from the 1970's, in which a rubber-curing/molding process's cure time 
>>was determined by a mathematical formula based on heat, pressure, mold 
>>shape, and a number of other variables.
>
>You're referring to Diamond v. Diehr, 450 U.S. 175, 195 (1981).  I have a 
>half-finished article I wrote in 1994 on software algorithm patents, about 
>32K, available at 
<http://www.delfinsd.delfin.com/felix/Algorithm_Patents.htm.  It's the good 
>part, the background material minus footnotes.  Although it's a bit dated, 
>the description of foundational cases is still accurate.

Thanks for the reference, and yes, the article was very interesting.  As 
usual, it sounds like the legal system has gotten the whole thing screwed up.  

I am still mystified, however!  If I understand the thrust of the legal 
cases you cited, purely mathematical algorithms are still not patentable, 
yet the patents on public-key cryptography are about the most purely 
mathematical ones that could be imagined.  They are not an element in the 
process, they ARE the process.

To recap, I've asserted (with no definitive proof, obviously) that when 
public-key cryptography was invented, in about 1976, the US government 
decided that it wanted to restrict it as much as possible from ordinary US 
citizens.  Due to the 1st amendment, legal restrictions on speech would not 
fly, and copyright was out because that would only have protected one 
particular program, if even that much.  The final alternative, patent 
protection, was essentially unavailable (or thought to be so) because of the 
traditional non-patentability of software and mathematics.

Patents would not have prevented the Russians from using RSA, nor any other 
foreigners, so as far as I can see the only group of people impaired by the 
RSA patent were American citizens as a group.  

To me, there are at least two mysteries that need to be solved here.  The 
first is why the cryptography patents were issued in the first place.  The 
second, and perhaps even more incriminating, has to do with why the patents 
were applied for.  Because the patent application has to be filed within a 
year of disclosure, the RSA patent would have to have been filed at latest 
by April of 1977.  Yet, that predates some of the earliest cases in your 
article by a year or more.  I've never heard a cogent explanation as to how 
R, S, and A decided that a "doomed" patent application was worthwhile, 
unless they had some insider information that the Gottschalk v. Benson case 
would be essentially ignored and the patent granted anyway.

This isn't unrealistic paranoia or conspiracy theory, either.  It is 
reasonable to assume that since the government would be the first, largest 
beneficiary of keeping RSA out of the hands of the public, and since the 
government made up and controlled the patent office and the court system as 
well, it could easily have made a decision that RSA was going to be 
patented, and Messr's Rivest, Shamir, and Adleman told of their luck in 
plenty of time to apply for a patent.  The other cases from 1978 on might 
simply have been window-dressing, to make it look like the courts had had a 
change of heart unrelated to the subject of public-key cryptography.

My question is this:  "Is there anything you're aware of that contradicts 
this impression?"  Or, is there a way to confirm this?