Subject: Re: A patent solution that is no solution
From: Brian Behlendorf <>
Date: Fri, 17 Sep 1999 00:35:51 -0700 (PDT)

To repeat, I am not arguing for software patents; I am sorry that by
trying to alleviate the pain of patents I am seen as promoting it,
hopefully I can refine my analogy to medicine: is a doctor who proscribes
a transquilizer a promoter of the cause of that pain?  Anyways:

On Fri, 17 Sep 1999, Richard Stallman wrote, quoting me:
>     So what if there were a patent marketplace, where patent use was set as an
>     explicit cost per unit, and a unit could be a CPU, a user, 1000 ecommerce
>     transactions, etc.
> Standard terms like this might make patents a little less inconvenient
> for non-free software.  They preclude free software, though.  Any way
> of keeping track of the number of copies, users, computers, invocations,
> or anything else about the use of the program, is not compatible
> with making the program free.

Note that I stated that one condition would be that a lawsuit claiming
"contributory infringement" could not be applied to a distributor of free
software that implemented patents, so long as they explain how to use it
legally.  This would have to be established as case law in court.  If this
were the case, the only people responsible for paying would be the
end-user, who can keep track of copies, users, computers, invocations;
because if they can't, logically someone else is the "end user".

>       For those companies who wish to turbocharge GCC and are willing
>     to pay for those patents, being able to do so easily would be great.
> This would effectively make a proprietary modified version of GCC.
> People who do not believe in copyleft might think that is ok, but
> those of us who use copyleft do so specifically in order to prevent
> it.  That's why the GPL has provisions to block this.

Calling it "proprietary" suggests that, without specific approval from the
author, I can't turn it into patent-free free software.  However, I can:
the right to fork still exists, and the mere fact that patent-indicating
code exists in the code base can't prevent me from taking this
"proprietary" version, slicing out the patent-indicating code, and
resulting in patent-free free software, without requiring that I seek
permission from the author.  So all the fundamental freedoms appear
intact.  Unless of course the program simply won't function without that
algorithm, which I would agree should be labelled a sham.

> If some companies want such features added to GCC, and they want it
> badly enough to spend money to get it, I would tell them this: "Pool
> your money to buy a license that covers *all* users."

Such a license may not be purchasable.  You may be willing to live without
that patent; fine.  It would have meant living without any real public-key
encryption by commercial law-abiding entities on the Internet for a long
time, though, at least until Diffie-Hellmen expired and thus DH/DSA was
usable patent-free.

>     Two primary benefits from this model: it's better to have code that is
>     covered by a patent in the open source environment (open to peer review
>     and bugfixing, and even forking) than not, 
> If the Open Source Initiative accepts software like this, it would be
> a great difference between the categories of Open Source Software and
> Free Software.  But I do not think they accept this.  As I recall the
> OSD, this violates it very clearly.

I wasn't suggesting that OSI has made any such move to recognize that, or
was considering it, and in fact, they indeed probably would not.  However,
OSI only certifies licenses; this model may exist *outside* of the license
on the code.

Consider a (hardly theoretical) program written in Finland that implements
RSA and has a GPL license.  In Finland, there are no patent issues around
RSA's license, so there are no restrictions on its use there, so the
author says nothing about having to get patents cleared.  However, when
people in the US download it and use it, either they are told by the
author in the release notes, or realize on their own, "hey, to use this
legally in the US for commercial purposes, I have to contact RSA and get a
patent license."

Is this still free software?  It has the GPL, and only the GPL, on it; and
the author does not *mandate* that you get the patent license to use it.
However, not doing so could land you in a heap of trouble.  One reasonable
response would be, "then don't use it, use a commercial product where the
patent issues are handled already" - which then means I usually miss out
on all the other advantages of using, enhancing, and promoting free
software.  Which seems like a net loss to the free software community to

I'm done on this topic, it was an interesting thought exercise.  Down with
software patents, up with a revamped LPF, etc.