[sf-lug] more re Novell/MSFT and GPL

[Rick Moen]

Quoting jim stockford (jim at well.com):

> As to patenting software, I once worked as sys adm
> and de facto DBA for the US Court of Appeals for the
> Ninth Circuit--i.e. techy for the federal government.
> My boss told me that all software written for the feds
> (and i presume for states and counties, etc.) is in the
> public domain. that's a lot of code in the public domain.

1.  That's not exactly right.
2.  And, besides, that would have no bearing on patents.

1-details:  Inside the USA Copyright Act, 17 USC 105 (i.e., title 17 of
the United States Code, section 105), you'll find the following:  

   Copyright protection under this title is not available for any work
   of the United States Government, but the United States Government is
   not precluded from receiving and holding copyrights transferred to it
   by assignment, bequest, or otherwise.

Like most US Federal statutes, that's really quite clear:  If a Federal
agency directly _creates_ a piece of software (or other creative work
normally covered by copyright), then it passes directly into the public
domain.  On the other hand, if it _hires_ some outside party to create
the code, then it can accept copyright title pursuant to contract.

So, the Feds _can_ arrange to own copyright over custom-written software
if they feel like it, just by putting the job out to bid.
Alternatively, if they don't care about ownership and produce the
software in-house, then it immediately becomes public domain.


2-details:  Public domain doesn't mean "free from patent encumbrances".
Public domain merely means "has no copyright ownership title".  Patents
and copyrights are _entirely_ different things.

o  A patent is a limited monopoly over _all_ implementations of a
   practical method or algorithm, that is new and innovative.  
   Ownership requires filing an application and fee with the US
   Commerce Dept.'s Patent and Trademark Office (USPTO), 
   surviving a review by a USPTO patent examiner, to determine 
   that the method/algorithm is properly described and truly 
   innovative, and eventual issuance of the patent.  Patent runtime
   is 20 years from the earliest filing date (of a patent application),
   if you filed on or after 1995-06-08.  (A different 17-year rule
   applies for earlier patents, the last of which will expire in 2012.)

o  A copyright is a limited monopoly over copying, basing works on, 
   and (in some categories) performing a particular expression of
   a specific creative work that has been set down in "fixed form"
   (i.e., not just something you're thinking of, but never wrote down),
   inside about seven or eight specified areas of creative endeavour 
   that include writing software.  You do not need to apply anywhere 
   or pay anyone to gain ownership of a copyright:  You gain ownership 
   over it automatically by creating it and putting it in "fixed form".

   (Notice that copyright law poses no restrictions on _using_ the 
   creative work, just on copying it and creating derivative works.)

   If an individual creates the work, copyright expires 70 years after 
   the author dies (per latest term extension for Disney's benefit).
   For corporate authors, the term is 95 years from the date of
   first publication or 120 years from the date of creation, whichever
   expires first.

Copyright covers a _creative expression_; patent covers an _idea_.
http://en.wikipedia.org/wiki/Idea-expression_divide

If I want to copy some software over which someone owns copyright, and
he/she doesn't let me, I'm absolutely free to write (or commission) my
own own equivalent software.  So, you can always work around a copyright
by homebrewing.

You cannot do that with patents.  If you think Microsoft's Active
Directory architecture is kinda cool and decided to write your own
from-scratch implementation to interoperate with theirs, they can
stop you on patent-violation grounds _if_ they happen to own a patent
on, say, "A method for authenticating to a shared directory service
[blah blah]".

The trick, of course, is to know what patents to avoid when creating
things.  Patents that are kept carefully obscure, to mug people with
them after they're already in trouble, are called "submerged patents".


> finding existing code wrt some patent trial could be
> difficult, of course. seems a significant consideration.

Sorry, not relevant.

What _is_ relevant, if you want to challenge a patent, is to show "prior
art" -- independent use of the method by others prior to applicant's
filing date.