[SciPy-dev] Numerical Recipes (was tagging 0.7rc1 this weekend?)

[Bill Baxter]

On Sat, Dec 20, 2008 at 10:18 AM, Alan G Isaac <aisaac at american.edu> wrote:
> Quoting from http://www.iusmentis.com/copyright/software/protection/
>
>         An implementation of a standard algorithm may
>         require very little creativity and if so, is likely
>         not protected by copyright. The function of a work
>         can only be protected by a patent.
>
>         Even when providing a largely functional
>         implementation, creativity can still be found in
>         things like function and variable names, code
>         optimizations and even the comments accompanying the
>         code. Copying such an implementation wholesale then
>         still infringes on copyright.
>

The problem, though, is that terms like "requires little creativity"
are subjective.  You may think the code required no creativity, but
the publisher of NR may disagree.  And then it takes a court to decide
who is right.   And at that point even if SciPy is vindicated, the
SciPy community still loses because of all the lost time an money that
has to go into such a fight.  It's easier to just make sure you steer
well clear of any potential infringement from the beginning.

IANAL, but I think that especially once the issue is raised in a
public forum like this -- now you've got a record that the developers
were aware of a potential problem.  If nothing is done about it, it
might be regarded as "willful infringement".  But if an honest attempt
is made to replace the code, then you've at least got a public record
that you were attempting to do your "due dilligence" to eliminate
infringements.

So, silly as it may seem, I think that Jarrod's plan of action here is
correct.   The NR guys may be wrong about copyright law, but being
right is not a prerequisite for filing a lawsuit.  Merely thinking you
are is generally enough.  And sometimes not even that...

--bb