I think the discussions around the various GR proposals miss some important
Hardware manufacturers are in the habit of keeping their firmware
closed-source. Some firmware contains algorithmic code and should be
categorised as programs while others are plain data that control settings for
registers, selectors for hardware subsystems etc.
Consequence: It is pointless to discuss if a specific piece of firmware is
DFSG-free or not. The manufacturer won't tell you if it falls in the realm of
DFSG-free data or non-DFSG program binaries. The interesting discussion is if
we should err on the cautious side and consider unknown entities to be
non-free or err on the permissive side and consider the unknown ones to be
free until proven non-free.
The reason why hardware manufacturers keep their firmware as closed source is
that they believe it gives them a competitive advantage. It prevents the
competition from copying what they have produced and it may prevent the
competitors from detecting (intentional as well as unintentional) patent
infringements. In the current world these are pretty compelling arguments and
as long as the bulk of the customers do not demand hardware where the
firmware is free, the manufacturers will continue to keep their firmware
In a possible future dominated by free software, we can make hardware
manufacturers provide source code for firmware, because we have swayed enough
people to our ideas, or possibly because enough people rely on Debian that
the manufacturers will feel obliged to listen to what we say.
In another possible future, we have by sticking to our principles made such a
strong impression on the hardware manufacturers that some of them decide that
the benefits (end user flexibility and control, third party improvement,
pre-training of prospective employees etc) of opening up the source to the
firmware outweigh the drawbacks.
Consequence: Since Debian is about providing its users with free software, the
goal (concerning firmware) should be to be able to provide the firmware that
users need to run their systems that is known to be DFSG-free. The debate
should focus on which path is most likely to accomplish the goal.
My personal experience is that the larger the company, the smaller the
interest in change will be and they will only change when outside pressure
forces them to. This leads me to believe that the quickest way to a future
where we can distribute free firmware is by getting many users and that is
best accomplished by for the time being erring on the lax side - considering
unknown firmware to be data.