Converting/Splitting Code - Open to Closed

Chris Sloan cds at
Tue Feb 13 23:51:23 UTC 2001

In the US, previously publishing something does not affect whether or
not it can be patented, AIUI, unless it can (for other reasons than
just being published) be considered prior art.

This is why RSA got the RSA patent in the US only.  They had already
published, so they were not able to get a patent in at least some
countries, but it didn't matter in the US.

I kind of like the idea that you have to keep your patentable ideas
secret until you patent them.  It means that if you find information
published, it either is already published or it can't be.  You don't
have to worry about someone patenting the information after you start
using it.  The downside is that our current process in the US can take
years to decide if a patent can be granted, and having to keep it
secret until they decide could be a real problem.

As for patenting something in the Linux kernel, or in open source in
general, here's how I think that it *should* work: If the algorithm
was created by the people who wrote the code, then the published code
should act as prior art, but if they are implementing an algorithm
which they did not invent, then the original author might be able to
get a patent (in the US, at least) at any time rendering the project
in jeopardy.  That being said, the main thing that I have learned
about the US patent office is that they constantly suprise me with
their actions, so if everything I said here was totally wrong, I'd
just shrug and figure they did it again.


On Tue, Feb 13, 2001 at 11:25:54AM -0000, Dave J Woolley wrote:
> > This is one of the dangers of basing Linux (or any other large,
> > multicontributor project) on the GPL; the threat that something embedded
> > deeply in the code could eventually have an external patent applied,
> > 
> 	[DJW:]  In the UK, one is always told never to reveal
> 	anything about something you intend to patent, except
> 	under a confidentiality agreement, because the patent 
> 	would be invalid if details were published first.
> 	Is this not true for US software patents?
> 	I'm pretty sure that no-one else could validly patent
> 	something that appeared in open source code, after the
> 	publication, as there would clearly be prior art.
> 	The real risk is that someone interprets an old patent
> 	of theirs as applying to code written after the patent
> 	was granted.
> 	I believe Netscape patentented SSL in order to reduce the
> 	risk of anyone else claiming that a patent applied.
> 	On the original question, I'm not sure about taking GPLed
> 	code out of GPL retrospectively, although I don't think
> 	the FSF intended this to be possible.  However, the
> 	copyright owner can licence the same code under a
> 	different licence and can licence a derivative that they
> 	themselves create under any licence they like.
> 	I Am Not A Lawyer.

More information about the License-discuss mailing list