On Friday 13 Nov 2009 14:58, raymond.meyer@rambler.ru wrote:

> I came across the following patent:


> http://www.freepatentsonline.com/7363626.pdf


> I didn't read the whole thing in depth, but it seems to describe a set

> of algorithms and data structures for running multithreaded

> applications.


> Can you actually patent algorithms and data structures? I think most

> software patents revolve around those two entities.


> Can I just go and invent a "new" way of partitioning my application

> into multiple threads and then claim a patent on it. So anyone who even

> dares to do data/function partitioning and multithreading has to pay me

> royalties.

The test of "obviousness" seems to be defined differently between patent lawyers and the rest of humanity. I am constantly amazed at what is permitted to be patented.

Lots of patentable (i.e. "non-obvious, not prior art") things, especially software features, are of the sort that if the requirement was on your spec sheet and you did not come up with a similar/better solution by coffee time you should be looking at a P45.

Real code probably has lots of features that may be patentable, just that most of the time it never gets discovered. It is also impractical to check unless you have very big pockets.

Some large companies have also claimed patentable features within other people's code when it wasn't there.


"The impossible is easy"