[patrick.pelissier@gmail.com: Fwd: Optimizing 1/x]

[Paul Leyland]

On Mon, 2005-09-12 at 21:20, Ashod Nakashian wrote:

> pay-off. The next question would be how slower is the next fastest 
> non-patented algorithm compared to HP's? Also, although I'm anything but 
> a lawyer, I'd be surprised if law forbids using a patented algorithm for 
> educational/non-profit purposes. So I assume that one could argue that 
> making patch (with the said HP algorithm) available would benefit many 
> and the burden of following the patent laws would be on the user, not 
> the provider of the code.

I am not a lawyer either, but I have studied the UK's Copyright, Designs
and Patents Act, 1978, in fine detail and I've sat through numerous
presentations, seminars, workshops, etc.

The first thing to note is that the position varies by jurisdiction. 
However, most countries have signed up to the Geneva Convention and so
the legal regimes are broadly similar in much of the world.

Secondly, people are most certainly allowed to study the patent itself
for the purposes of self-education.  The justification of patent law is
that it brings out into the open view inventions that would otherwise be
kept as trade secrets.  In return for public disclosure, the inventor is
given a strictly limited monopoly on production of the invention.

What you are not allowed to do without the permission of the patent
holder is to make copies of the invention, irrespective of the purpose
to which the copies are to be put.  You are allowed, indeed encouraged,
to learn from the design and invent things which solve the underlying
problem in a better way than that provided by the patented invention.

Summary: anyone distributing HP's algorithm with the disclaimer "for
educational use only" is very probably in violation of the law in many
countries.  However, I am not a lawyer and this does not constitute
legal advice.  If you want legal advice, hire a lawyer.


Paul