[Patents] Re: [Patent] Pro Software Patents = Illegal

[orre@nada.kth.se (Roland Orre)]

On Thu, 10 May 2001, Martin Wahlén wrote:
> Jag tänker fortsätta leka Devils Advocate ett tag till....
> > Just think once more about the concept:
> > Start with a fundament, the trunk, then add small pieces of
> > fundamental designs, base forks, which can be combined into more
> > and more complex designs sprouting new forks from each fork.
> >
> > When you manufacture physical constructions they represent
> > leaves in this tree. Where you arrange your atoms and molecules
> > in some finite ways. A leave can lead no further to new designs
> > it's a "dead end".
>
> But is it really? Say that I come up with a new tire pattern for car tires
> and then multiple people elaborates on my design to make different
> patterns that have other characteristics but still use (parts of) my
> design?

Here you may answer a tricky question I've been dealing with. The
forks are always designs, but the leaves are physical realializations
of a specific design.

So the principle of e.g. having a rubber wheel armed with steel
and adding a surface with a patterned friction surface is the design.
This  design can the be realized physically into a set of leaves with
different patterns, different rubber mixtures, different sizes
etc.

As long as you are not implementing anything, a design could be
arbitrary complex, but... then you have the limitations of the
physical world, which makes most designs impossible to implement.

If you e.g. want to build a ladder to the moon, due to the
physical world being "crippled" so to say, you have no way you
could do that, even though the design as a concept is OK.

Considering software there are no limitations whasotever. What
I still haven't solved is if there are any "leaves" in the software
world. My first view was that an algorithm expressed in its purest
form, e.g. lambda calculus, is the design, then you could possibly
imagine the leaves as different specific implementations, but
these will not become leaves, as any specific program as such is
still an abstract description which can be infinitely extended.

A software tree thus has no leaves!
Leaves are something you get due to the limitations of the
physical world.

One question I got earlier was if one leave can be the result
of several designs? Yes, I think they can. Several designs can
accidentaly, but unlikely of course, result in the same
physical realization, even though the designs are completely
different. OK, now I see, the explicit bit pattern you get
in the computer's memory from a software design could be
seen as a "leave". That is, in a "leave" the abstraction done
in the design is lost, but several different designs could
e.g. accidentily be implemented by the same bit pattern.
"Leaves" can thus be copyrighted as being a pattern.

For software can only licensing work, like the GPL.
The GPL defines one tree or forest.
When e.g. Microsoft releases their sources soon (or have already)
that may be something like MRL (Microsoft Restricted Licence)
which would make another tree or forest.

I recently read in Computer Sweden about some MicroSoft's
attourney complaining about that GPL would be a threat
to all what copyright means. He had obviously not understood
anything.

When Microsoft realizes the real power of GPL and hopefully
make their own MPL later on, then there is fair competition
also in the computer world.

There may also be some "cross breeding" when e.g. FreeBSD trees
are joined into the MPL forest as well as into the GPL forest.

> True we are not likely to get as high in the tree as with software that
> lives in the center of the tree (World 3) in Hartmuts posting.
>
> > When you consider a software tree, there are no leaves, at the
> > end of each fork there is a design. Each design can be seen
> > as a fruit, a fruit have seeds which can grow into new trees.
>
> This I buy, but I think you need to put it in a format that can be
> understood by other people. A normal piece of semicomplex software (for
> instance bash), is made up of multiple modules (job control, parsing,
> interpreting). All of these modules are in turn made up of modules,
> lexical analysis & parsing makes up the parser.

Yes, I'll try to work out some good examples.

> > Each software design can generate a new forest of trees.
>
> > Wherever you cut a fork, you cut an inifinite number of new
> > designs.
>
> I agree, the answer that I have gotten multiple times when I come up with
> this type of argument is that "it is more like a graph", you can invent
> some other mechanism to circumvent the patented method/design and apply a
> new design.
>
> For instance say that recursive decent parsing was patented. You could
> still do operator precidence parsing and probably get by.
>
In many cases you have alternative solutions yes, but I also see
software solutions as more fundamental. If someone has the patent
on recursion and someone has the patent of iteration you are stuck.
If someone has the patent on simulating wavefunctions you could
not simulate a physical world for instance.

> I and Greg had a bit of a fight about linear programming and patented
> methods. My posision was that if Danzig had patented simplex, back in the
> 50's then there would be no computer industry today (ok, maybe that is
> beeing a bit to harsh).
>
> Gregs argument was that Kamarkar invented his method. I countered with the
> year of publication 1984. I think that that proved my point, and Greg
> disagreed.
>
> Martin

I'm not surprised, most of what Greg says can be proven false,
I sometimes get that feeling that Greg is another "devil's advocate"
as you express yourself :) because whatever he says encourages
us to get arguments to prove him wrong, which is usually easy.

My own theory (prove me to be wrong Greg!) is that Greg is trying
to prove that the whole patent system is wrong due to his acting.

        Greetings
        Roland