🚀 go-pugleaf

In article <7081@ns-mx.uiowa.edu> jones@pyrite.cs.uiowa.edu (Douglas W. Jones,201H MLH,3193350740,3193382879) writes:
>Personally, I consider the UNIX set uid bit (Patent 4,135,240) to be an
>invention of fundamental importance.  This invention is perhaps the
>simplest solution to the mutual distrust problem anyone has found.

Possibly so, but a 17-year monopoly, which might have been appropriate
to the slow pace of technological change that prevailed in the 1700s,
is excessive.

>The patents on data compression algorithms (4,464,650 4,558,302
>4,814,746 4,797,729), on sorting algorithms (3,380,029), and on
>cryptographic methods (4,200,770 4,218,582 4,405,829 4,424,414)
>also appear to be cover genuine inventions, although I'm not familiar
>with the details, and I understand that the LZW family of compression
>algorithms are covered by conflicting or overlapping patents.

None of these are patents on algorithms.  All are patents on CLASSES
of algorithms, and the public-key encryption patents are so broadly
drawn as to cover ANY possible public-key encryption system (they claim
the idea of public-key encryption itself).

I don't object to software patents merely because in some cases, the
patent office screwed up and gave the wrong person the patent on
something that's obvious.  The public-key work was all done by professors
that had plenty of incentive to publish their work; the patent system
did not serve in this case to benefit the public by "promoting the
progress of science and useful arts".

>There are two directions that the Patent Office could move.  One
>direction would involve abandoning software patents, either because
>they can't get staff who can tell the difference between good patents
>and bad, or because they determine that even if the silly software
>patents can be eliminated, software patents would be detremental to
>the advance of our art.

Software patents are detrimental to the advance of our art, so much
so that if there is no change, only large companies with many patents
to cross-license will be able to remain in the software business.

>The second direction would be to shore up the patent office staff with
>a corps of experienced programmers, or at least programmers with
>advanced degrees to assure that they have people who really understand
>the state of the art.  They may also need to require some software
>training on the part of lawyers who want to practice software patent law.
>This might work to cure their habit of granting embarrassing patents.

Embarrassing patents aren't the only problem.  Consider this: suppose
you had gigabytes of fast memory for low cost.  How would you restructure
the way you program?  What new algorithms would you use that make no
sense now?  I'll bet that if asked hundreds of good programmers that,
I'd get similar answers.  But only those programmers in companies with
huge budgets will have access to those types of machines at first (the
IBMs, Fujitsus, and Cray Researches of the world); they will solve the
new problems in the most logical way and then they will patent their solutions
(this is LPF's argument).  These techniques will be non-obvious to people
who are used to thinking that a megabyte is a big chunk of memory, but
obvious to those who are used to programming the leading-edge machines.

Result: when such a machine becomes as common as a PC, the giants will
own all the patents and all programmers for these machines must either
license or violate the patents.

--
--
Joe Buck
jbuck@galileo.berkeley.edu	 {uunet,ucbvax}!galileo.berkeley.edu!jbuck