[Haskell-cafe] Re: US Patent for the idea ...

[Steve Schafer]

On Sat, 17 Apr 2010 23:33:48 +0100, you wrote:

>I think in all fairness to examiners that in a way they have an 
>impossible job due to the fact that what is a clever idea to one 
>programmer will be a trivial idea to another: the field is so huge and 
>people have such different experiences.

In US patent law, algorithms themselves were deemed unpatentable quite
some time ago (I believe that European patent law is more liberal in
that regard, but I don't know all of the details). So a lot of the
discussion concerning software patents in this country has been on
whether or not software can be considered to be an "invention" separate
from the underlying algorithms used in its construction. Since those of
us who work with software realize that software is often little more
than a restatement of an algorithm in a way that is suitable for a
computing device to "understand," it's very difficult to draw a clear
line between the two.

People do occasionally come up with truly novel ideas about how to
perform some software task, but it seems to me that unless the novelty
involves some aspect that can be separated from the algorithmic approach
used, it shouldn't be patentable. For example, quicksort, though
certainly novel, is purely an algorithm, so it shouldn't be
patentable--it is completely independent of any "tangible"
implementation. But a sorting technique that is optimized for large
datasets that can't be held entirely in volatile memory, and explicitly
takes advantage of known characteristics of disk latency, etc., could
very well be patentable.

-Steve Schafer