Subject: Related to: Examples needed against Soft Patents?
From: "Ruth A. Kramer" <>
Date: Thu, 30 Dec 2004 03:58:22 -0500

I found a comment by Max.Hyre over on LWN
( today (it's from a few weeks ago). 
You may want to read his entire comment, but I found this interesting. 
Can his suggestion serve as the legal bright line?

<partial quote>
What is needed is a legal ``bright line’’ making it easy to discern
which side of the SW/HW divide any given invention falls on. The obvious
one is a simple, clear, test: If the invention can be effectively
executed solely in software, that implementation isn’t patentable. 

That covers spreadsheets, graphics-editing algorithms, word processors,
computer controls for music, &c. (And must specifically exclude such
trivial hardware involvement as making pretty lights on a CRT’s face, or
wiggling bits on a communications line.) To apply it specifically to
instances mentioned above, the ``recently-built clockwork calculating
machine with a number of clever new tricks’’ may be patentable when
implemented in bronze, steel, and glass, but it doesn’t cover the
algorithm which displays the identical results on a CRT. Likewise, new
methods for running a robotic assembly line are covered insofar as they
implement novel machine tools or combinations of older mechanisms. But,
that patentability applies only from the machines back to, but not
including, the parallel port which sends the guiding impulses to those
machines. The software can be copyright, or kept a trade secret, but is
not patentable. Thus, your hardware-implemented ECC switches would be
eligable for a patent, but that would have nothing to do with ECC RAM. 

To continue this to one obvious limit, the production methods needed to
build a new FPGA are patentable, but simulations of that FPGA, or any
particular arrangement of internal bits to make it do something, are
not. Yes, this rather changes the rules for hardware engineers, but so
be it. (Spoken as a programmer. :-/ ) 
</partial quote>

Randy Kramer