[Patents] Assumed patentability

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

Mr Aharonian,
It would be interresting to hear your view upon this.

Assume that software was patentable.

1) Assume that I use a certain algorithm commercially.

  a) Should I a pay a license fee based upon the number of times I
     use this algorithm in my design?

  b) Should I pay a license fee per computer system, per CPU, the
     time a specific CPU is spending within this algorithm?
     Should the fee be proportional to the processor speed?

  c) In case I sell an information service where this algorithm is
     used, should I pay for the number of times the algorithm is
     invoced or a proportional fee based upon what I earn from the
     usage of this algorithm.

  d) If my computer programs have invented this algorithm on their
     own, do I need to implement some kind of detection system
     within my programs that could detect any patented algorithm
     being invented and then automatically start paying according
     to a-c above?

  e) Should I even pay a license fee based upon the number of times
     I'm myself thinking, using this algorithm.

2) Assume that I use a certain algorithm non commercially.

   Even though I'm rather clever I don't even have one single clue
   about any way I should be able to pay a license fee for its usage.
   As I have understood, the US patent system doesn't even allow
   non commercial usages of patented designs.
---------------

In 1a I would usually have no problem hiding the algorithm in my
design, for anyone, even my computer science friends, so it would
be trivial to hide it for anyone supervising my design, and I
can guarantee you, that I would not check my programs where I
have accidentaly reinvented any "patented" algorithm, I don't
have time for that. Are you going to check my programs or are you
going to pay someone checking my programs?

In 1b, assuming that I e.g. pay per CPU and then run this algorithm
on two different virtual CPUs within this CPU, should I pay double
then?

In 1c, I can just set up the information service which is utilizing
the algorithm in a country where this algorithm is not patented and
call the information service remotely, but most likely I would guess
that my software would use this algorithm so little so it's not
measurable anyway...

In 1d, I'm probably clever enough to invent a detection system which
would be able to detect any automatically invented instances of the
algorithm. Who should I charge for the development cost? You?
Of course, I could patent this detection system... But I can assure
you, if someone else has patented that detection system, I would
not install it on my computers, because I don't want to pay that
license fee. Who would force me to do that? You?

In 1e, who should detect when I'm thinking commercially using this
algorithm? You?
---------------

You see Mr Aharonian, even though I would certainly want to be honest
about paying the license fee, I can see no way how I possibly could.

I can imagine that someone who is stupid enough could possibly
believe that the usage of a specific algorithm is detectable/payable?

I'm rather clever with computers and I have no way how this could
possibly be done realistically.

For once, give me an honest answer, on the above!

Software is simply not patentable Mr Aharonian.

        Best regards
        Roland