Re: [Patents] some practical ideas

[Peter Vandenabeele <peter.vandenabeele@mind.be>]

On Fri, Dec 20, 2002 at 12:30:15PM -0500, Gregory Aharonian wrote:

Dear Greg,

> Peter is obviously ignorant of much outside software.

To give you a clue, I have a M.Sc. and Ph.D. in EE and worked in 
Silicon Valley and Belgium in the field of electronics and micro-electronics
for a number of years. I worked with patent lawyers in Belgium and San 
Francisco.

When employed in California, in 1992, we used a patent on HW to barrage 
a new entrant German competitor for importing his tools into the USA and 
selling them to major USA electronics companies for a few months (some 
USA customers favored the German design over our California based design).
Of course, as soon as a judge looked at the patents it was clear they were 
not relevant to the specific German technology and the import ban was lifted, 
but it still served its purpose, to hold back the competitor from selling in 
the USA for a few months.

> and I change his words:

For clarity, these are not my words, but a text you modified:

> >A fundamental problem with hardware patents in comparison with the
> >kinds of patent of which I am an author myself, is that Hardware
> >Development is advancing at such speeds, is developed on such a
> >massive world-wide scale, uses such intrinsic complexity in bringing
> >together thousands of simple "Circuits" in a single VHDL program, that
> >there is no way anyone could correctly judge if a certain "Invented
> >Circuit" is new or already used by some-one, somewhere on this planet.
> >Nobody can claim he can check correctly, realisticially that no
> >one else used the basic methods of e.g. "Instant Signal Processing"
> >somewhere before in the world .......

In the way it is proposed here, this would hold very similarly, since it 
is essentially a SW method to reprogram a chip. Another analogy with the SW
world, is that the vendors of FPGA are incorporating 3DES encryption on the 
"HW designs" (actually a bitstream) that are loaded into the FPGA's. This is 
because, analog to the choice made by most Software vendors to use Secrecy to 
protect the source code of their programs, also the vendors of electronic 
systems use Secrecy on the bitfiles loaded into the FPGA to protect their 
work. Again, Secrecy is an acceptable method to use, since it does not suffer 
from the scalability problem I refer to. However, currently FPGA vendors in 
the USA are also starting to patent even the most basic pieces of IP (under 
VHDL format) as a defense against others that may do the same and attack them 
at a later date. An increasing amount of budget is spent on legal matters 
instead of actual development. 

Using copyright does not cost money upfront and is clear cut. If I don't 
steal code from any one, then I know I am honest and don't have to worry 
about being attacked at a later date by anyone else. And if i would steal
code, it would be easy to prove that (either from analysing the executable
or asking a copy of the soruce code under injunction from a judge). I 
believe the potential advantages of the broader protection by patents (which 
I assume are fair comensation for the R&D done by large and small companies 
or individuals) are more than offset by the practical disadvantages of 
software patents:

- completely unscalable to check the anteriority research
  (as the amount of previously published and patented SW grows 
   dramatically the cost of a correct anteriority research becomes 
   prohibitive; in practice, the anteriority research is not done 
   to cover _all_ SW that is that was written somewhere world-wide 
   and incorrect patents are produced by large numbers; in contrast
   to hardware, where the design is visible from the outside, the
   interior design of SW is often hidden by Secrecy.)

- makes it impossible to guarantee "Freedom of software"
  (unless only techniques are used that are 20 years old, which
   is an infinity in SW development; reducing the lifetime of
   a SW patent to a typical cycle of 3-5 years might be a
   resonable balance if a government really wanted to enforce
   software patents in its territory; and even if only older or 
   published techniques are used, the mere threat of the legal 
   costs of patents are deadly for a small company).

- legal costs are more or less flat fee and strongly favor the
  large companies relative to small companies. They promote the
  "status quo" where the large companies can suppress new entrants.

> hardware.  Many hardware patents are as bad as many software patents.

This is a very strange argument to hear from your side. Taking it literally, 
you would indicate here that a reasonable amount of SW patents are bad
(which they actually are, but it is interesting to hear this from your side).
This exactly proves my point that the correct implementation of the concept 
of SW patents is not practical (and it becomes increasingly worse) and we 
would rather not enforce the measure (due to its side effects), than continue 
to use it under these circumstances.

> They increasingly are designed using the same tools by the same people
> with the same amount of future use and obsolescence.  It is just as
> hard and just as easy to search software.  

Not true. Software development does not require the level of "scaling"
that you need to preform HW development. It is entirely possible for
a few people, e.g. at some university in India to make very advanced
SW and protect it under Copyright (virtually automatic) and use Secrecy
or not (depending on the purpose). How would you ever know that a new
SW patent of a company in USA or EU is not reusing a Method that they
already used.

Just a very practical yes/no question. Does a search for a SW patent
include the code trees and mailing lists of the most important Free Software 
projects, that are in many respect at the forefront of SW development 
(e.g. the linux tree, the FreeBSD tree, the OpenBSD tree, the linux powerpc
tree, the linux ARM tree, the eCos tree, the Apache tree, the sendmail tree, 
and 100's of other important Free SW projects, ....) ? Who is the person
that claims to know and track all the developments and arguments in all
these Free Software projects ? I would hire this genious if he was 
available. And this is still only the public information, not taking 
into account the multitude of software that is not published.

> that of selling hardware designs in software form.  If anything,
> hardware companies experience more patent problems and have paid
> more in patent license and infringement fees than software patents,
> and probably have more lawsuits going on.  And for the most part,
> Peter is clueless about the difficulty of searching for prior art
> to invalidate software patents - he should take to those of us in
> the US who do it regularly and get some facts instead of fantasies.
> And if he would talk to such people, he will find out that searching
> for software is comparable in difficulty to searching for hardware,
> even when you have to go back far in time (for example, British
> Telecom's hyperlink patent).

Please educate me in how you can prove with reasonable accuracy that
no one in the world has ever used the Method/Apparatus for which a patent 
is claimed. Since most people do not believe in SW patents they do
not enter every piece of a program they ever wrote or every development 
into the patent database. You would need to look outside the patents 
database, but to my current understanding, this seems completely 
unrealistic (and even then, you would miss most of the code which is 
covered by Secrecy).

> Software is as much a technical invention as hardware. 

This is exactly the whole question and is a arbitrary statement. I have some
arguments to claim that it is _not_ an invention, but it is "development". It 
is the write-up of some idea I have on how to present certain information on 
a 
screen, how to optimally execute some calculation or search in a database. 
This comes very close to just one implementation of an _idea_, not an 
invention. And yes, this effort of doing the write-up deserves IPR
protection, just as the way one author writes up a murder mistery and
another author writes up another murder mystery. Should we patent the 
concept of a murder mistery with 7 people in a desolate castle, being killed
one after the other ? (years after Agatha Christy, the "Name of the Rose"
still got a Nobel Price for literature). The idea should not be patented, but 
the individual stories should be Copyrighted to protect the work of that 
author. We really have to chose, either Copyright _or_ Patents.

> light of the physics of the universe I live in.  Maybe spacetime is
> different in Europe.

On the level of SW patents, the universe is indeed different, but there is 
a proposal towards world-wide harmonization, to which I am not favorable.

Peter

> Greg Aharonian