Subject: Re: keeping patentable algorithm free
From: Ian Lance Taylor <>
Date: 30 Jul 1999 11:46:20 -0400

   Date: Fri, 30 Jul 1999 09:43:04 -0400
   From: John Cowan <>

   Ian Lance Taylor wrote:

   > One easy and relatively inexpensive way to publish an algorithm with a
   > legally verifiable date in the U.S. is to register it with the
   > U.S. copyright office.  You can send them a program listing, and they
   > will basically file it with a timestamp.

   Sorry, not enough.  For patent purposes, the invention must be
   described in openly available literature.  Registration with a
   government agency doesn't cut it, as nobody (in practice) can obtain
   the listing.  Publication on Usenet or the Web serves the necessary
   purposes: the algorithm must be *available* to persons learned in
   the art.

Yes, you do need to do more than merely register it with the copyright
office, and you are correct to point that out.  I assumed that that
had already been taken care of in this case since the algorithm in
question is part of a free software package.

However, as far as I know, publication on Usenet or the Web rarely
provides a legally verifiable date.  It's not like an appearance in a
published magazine.  I suppose you could try subpoenaing the records
of or