For Approval: CeCILL - (now patent laundering)

[Stephane Dalmas]

>Not sure you understood my original point, but then "patent laundering"
>is
>an English expression that may not translate well into French.
>  
>
Maybe "blanchiment de brevet" ?

>What it means is the practice of expanding a patent license
>or covenant not to sue beyond the context of the initial grant
>to argue that additional products are also licensed.
>
>The "locus classicus" of patent laundering, in the chip world
>anyway, is where company A gives company B a site license to use
>A's patents to make chips at B's silicon wafer fab.  B then turns 
>around and argues that *devices manufactured for 3rd parties at B's
>wafer fab* 
>are also implicitly licensed under all of A's device patents.
>Got the picture?
>
>  
>
Of course. I understood your point well.

>As written, CeCILL is an open invitation to try to launder software
>patents: "Licensor undertakes not to enforce the rights granted by these
>
>patents against successive Licensees using, exploiting or 
>modifying the Software."  Note the CeCILL Licensor's covenant not to
>enforce applicable
>patents is *not* limited to use of the CeCILL licensed software itself,
>but is arguably applicable to all Licensees who use, exploit, or 
>modify the Software but who also might be infringing the patents
>in other applications.
>
>So, the following scenario can ensue.  Licensor (let's say France
>Telecom
>for purposes of illustration) has patented software which they want
>to make available gratis to the open source community, but for which
>they want to reserve the commercial rights for proprietary SW.  
>FT licenses the SW out under CeCILL.  Crafty company B comes along, 
>uses a copy of the CeCILL code somewhere, and then argues that FT cannot
>sue them for use of the patent in company B's proprietary SW because
>of the CeCILL non-assertion provision.  Voila, patent laundering.
>
>This is either a subtle drafting flaw in CeCILL, or a somewhat sneaky
>way to
>attempt to wage guerilla war against SW patents.
>  
>
As I mentionned earlier explicitely, this is not a subtle drafting flaw 
and neither a sneaky way to undermine software patents. It reflects what 
we thought was natural in an Open Source license such as CeCILL. We knew 
that some people may disagree on that.

However for "real" software patents (patents that are entirely about 
software and not about a mixture of soft + hard) it is not clear at all 
to me how patent grants in licenses such as EPL, CDDL, and Apache 2.0 
are in fact significantly more protective (for the patent owner) than 
CeCILL. If you have an Open Source license granting patent licenses and 
that allows people to make derivative works that are also immune to 
patent claims, it is easy to "launder". For example if FT makes 
available under such a license a software with a patented video encoder, 
company B can just extract the encoder from the source code, it can even 
translate it to another programming language and obtain a derivative 
work immune to patent claims from France Telecom. The only constraints 
on B's software will come from the software license itself.

Stéphane.