Re: Unisys and LZW

[mikebw@bilow.bilow.uu.ids.net (Mike Bilow)]

kz1f wrote in a message to Mike Bilow:

> We have had the same hazzles with the RSA encryption algorithm.  In the
> U.S. you have to pay royalties to RSA, Inc. when you use it.  In Canada
> we do not.  Different laws.

 k> This is surprising.  Canada does not recognize US patents? I
 k> know that alot  of Asia doesnt, hence the bootlegging of
 k> Microsoft products etc but I would  think that particually
 k> given the trade agreements with Canada that they  would have
 k> to honor our patent laws and rights etc.

No one is entirely certain whether U.S. law allows algorithms to be patented. 
Until about 15 years ago, every patent lawyer in the U.S. would have told you
that algorithms absolutely could not be patented.  Nowhere except in the U.S.
does anyone think algorithms are subject to patent even now.  The Patent Office
itself is silent on the issue, taking the position, as I understand it, that it
is not their place to make that decision by denying patents, and they therefore
allow the holders of the patents to try and defend them in court.

The Patent Office also admits that they are incapable of accurately determining
claims of "prior art" on algorithms, since many algorithms make it into the
public domain and are widely known before someone thinks to write them up. 
This accounts for some of the most notorious patents issued for algorithms,
such as the patents for XOR and for network bit order, and for the Compton's
patent covering multimedia and hypertext that was recently revoked after
review.

The patent law is such a complete mess that most businesses decide that they
are better off to violate a patent claim and see what happens, attacking the
validity of the patent claim if ever challenged.  This has become so much the
usual practice that the patent system itself has been rendered almost useless.

 k> Free Software
 k> Foundation doesnt view algorithms as patentable either but
 k> that doesnt mean they dont  have to respect the patents from
 k> those that do think they are. 

I think you are confusing FSF with the League for Programming Freedom.  FSF is
the organization which also opposes copyright of software, and in furtherance
of that goal administers the GNU General Public License.  LPF supports
copyright of software, but opposes patent of software.  Unlike copyright, a
patent covers all implementations of an idea, even if performed or discovered
independently by someone in complete ignorance of the work of the patent
holder.  Worse, patent protection begins with the creation of the idea and the
patent process is pursued in secret, so many indpendent rediscoveries may occur
and make it into production before it is possible for the rediscovers to know
that they have committed an infringement and will be liable for it.
 
-- Mike