🚀 go-pugleaf

First, my knowledge of case law extends up to Diehr, so later cases
are unknown to me. I do not believe this will effect the main thrust
of my argument.

My personal opinion on patents for _algorithms themselves_, and anything,
which would in practical effect give protection to an algorithm, is that
they are a fit of lunacy that should be banished from the planet.

Why do I take this view?

First of all, what exactly is a mathematical Algorithm? Let us assume
that it is possible to give some half way decent definition of the
term. I then ask - what is protected. Let us assume that the Arhennius
equation of Diehr was in reality something a little easier:
	say f(x) = x + (x + x)
Would the patent be infringed if someone else came up with a similar
patent but for a formula:
	g(x) = (x + x) + x ?
	That is to say, does the protection of the algorithm extend
to equivalent formulations of the function? Note that they are
*different functions*. (here by equivalent I mean that for all x,
f(x) = g(x) )

	Proponents of patent protection for algorithms may want to
reply immediately 'But of course! This is obviously going to be an
infringement, and something that we *should protect*.'

	Why should we protect them (setting up straw person...)
: in order to promote the growth of industry.

Why: because:
*If we do not protect them then the patent can be easily circumvented
by use of an equivalent form of algorithm where appropriate*. 

	Fine, the only problem is that this means that we are
cutting the throat out of any innovation in the field immediately
and irreconcilably.

How?
	Lets say that I come along with a process for curing rubber
which uses the equation:
	h(x) = 3x
	The formula is, again, clearly equivalent, but this new
function requires only one operation, where the other needed three!
I have made an innovation in the field, and yet, in the view of those
who would patent algorithms, my innovation should not be protected.

Main Result:

	Thus patent protection of algorithms is distinctly
*counter-innovative*.


Side Issue:
	Further, let's assume for a second that all common sense
has departed those in favour of algorithm patents. How far will they
go to protect an algorithm? Will they also strike down an algorithm
of the form:
	j(x) = (x + (x+x)) * gamma
	where gamma = 1.0000000000001 or similar? Here now we have
that j(x) = f(x) *only* when x = 0, and yet the practical effect
again is that j(x) approximates f to a degree of accuracy which is
sufficient for all practical purposes.


In short, those who would support patents on algorithms per se
[...] haven't bothered to think about what they are
arguing for.

Brendan

#include <standard.disclaimer.h>
#include <Usual.copyright.rubbish.h>

	

[mod- note most patent systems require at least 3 elements for the
grant of a valid patent:
	1. Novelty- the invention must be novel over what has gone before
	2. Unobvious- the invention must be unobvious in light of the
		general knowledge in the field.
	3. Utility- the invention must be useful.

It is doubtful that the examples cited meet these criteria. However,
if you were to cite an algorithm that did meet these criteria then it
would be clear as to what we should be discussing. For example, take
the RSA algorithm and try to fit it to the above criteia.
]