[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Taking a position on anti-patent licenses

Nick Phillips wrote:
> So. The proposition to discuss would appear to be along the lines of:
>     Debian accepts that it may in certain circumstances be desirable
>     (or at least acceptable) for software licenses to limit certain
>     freedoms in order better to protect Free Software as a whole.
We may not accept this.... see below.

>     A 
>     current example of this would be the GPL, and the conditions it
>     places on distribution of derived works -- basically that they
>     must be distributed with source, under the terms of the GPL.
Actually, we don't necessarily accept that this is a limitation on 
freedoms.  :-)  

According to what I will dub the "first, do no harm" test, a free license 
should not restrict activities which would be unrestricted had the recipient 
not accepted the license.  The GPL and all other free licenses pass this 
test, because their restrictions are all on activities normally prohibited by 
copyright law; many non-free licenses, such as the GFDL, pass this test as 

>     In the light of the threat that software patents pose to Free
>     Software, we believe that it is likewise acceptable for software
>     licenses to place conditions on the use of software patents
>     against Free Software authors and distributors.
Well, sometimes.  :-)

> If it turns out that we accept that, then we would need to go on to
> determine what kinds of restrictions we might regard as being "OK".

I believe that most of us have come to the conclusion that self-protection 
clauses are free.  These are of the form:
"If you make a legal claim stating that use (/distribution/etc.) of this 
software infringes a patent, then you may not use (/distribute/etc.) this 

I believe that most of us have come to the conclusion that clauses which may 
impact legitimate non-software patents are non-free.  These include those of 
the form: "If you claim that the author has infringed a patent, then you may 
not use (/distribute/etc.) this software."  They also include those which use 
the phrase "a patent applicable to software", since this is so broad as to be 
a meaningless limitation.  (A "patent actually applied to software" would be 
a different matter.)

There is argument over those clauses which do not impact legitimate 
non-software patents, but which do impact uses of software patents against 
unrelated software.  I think many of us consider these to "contaminate other 
software", or to fail the "first, do no harm" test (in the case of "defensive 
patent lawsuits" unrelated to the software at hand).  However, we have 
actually seen very few, if any, clauses which are really written carefully 
enough to fall into this category, so I suspect we haven't thought about them 
as carefully as we have thought about the other two categories.

Josh Triplett wrote:
Requiring that distributors of a piece of software refrain from making
accusations of patent infringement regarding the software itself is
consistent with the goal of upholding the freedoms of users over that
software.  As such, we consider license condititions acceptable that
terminate a licensee's rights to the software if that licensee raises a
patent lawsuit claiming that the software in question infringes their

However, many software licenses choose to go further than that,
requiring that distributors refrain entirely from engaging in patent
lawsuits against any authors of the software, regardless of whether
those lawsuits are related to the software or not.  We do not support
the practice of patenting software, but we find it unacceptable for
licenses to place requirements which pertain to other, independent
works.  We believe this policy is consistent with the principles behing
in Debian Free Software Guideline 9, "License Must Not Contaminate Other

Indeed, I think this is my position.  :-)

Reply to: