[ppc-mobo] GPL-like hardware design license?
graham at collector.hscs.wmin.ac.uk
Tue Oct 5 11:00:10 UTC 1999
> I would consider whether a GPL-like license is even appropriate for a
> non-software project. Many of the motivations for creating the GPL (and most
> other software licenses) simply do not exist for hardware design. 1) I can test
> a bug fix in software for free, but the equivalent in hardware may cost several
> hundred dollars. 2) Software can be reproduced near infinitely for
> insignificant costs, the same can not be said of hardware. 3) Free software
> licenses are based upon the actual code, whereas I *think* that hardware
> design is more geared along the lines of circuitry (algorithms) as opposed to
> layouts (code). And 4) end users can download open software for free but cannot
> do the same with hardware, even if they were proficient enough to build their
> own boards it would still be much cheaper to buy them from "closed" vendors.
I think all the above points are arguable (tho' given recent comments
about philosophical discussions being 'noise' for this group maybe
the argument should move. I've just set up a list in a (probably
hopeless) attempt to centralise this rather scattered discussion;
to subscribe mail majordomo at collector.hscs.wmin.ac.uk with message body
'subscribe hardlicense-discuss'; otherwise, both pleb at cse.unsw.edu.au and
openip at egroups.com are lists currently discussing the same topic).
Back to the points above:
1) Hardware can't be gpl-ed; people are talking about hardware DESIGNS.
Designs may be debugged using simulators (ie. for free). This ambiguity
seesm to come up in every discussion of the topic I've seen: it is
the designs not the physical product that are relevant here.
2) Hardware designs can be reproduced near infinitely for insignificant
3) Yes, there are significant issues with patents in hardware design
that may not arise in software - this is a real issue, I believe. But
saying 'complicated' doesn't have to mean saying 'give up'.
4) The answer to this point depends on the kind of design you're talking
a) Designs for FPGAs or other programmable logic may be downloaded into
the hardware. The situation here is similar to that of software and will
become more so. There is already quite a large market for 'closed'
designs of this type. For this type of design I believe rms' position is
that the gpl can be applied directly, tho' it does seem to me that there
may be issues specific to hardware design not explicitly covered by the
b) For other types of hardware, the argument shifts to cost. I have
gnu/linux for 'free' because CDs are cheap to manufacture in bulk (I would
have more trouble affording the time and money to download it). Other
types of hardware are more expensive unless manufactured in the same kind
of quantity. But they are still manufacturable; there is no reason why
a free design can't be manufactured by a conventional manufacturing company
but has to be hand-made. This is already happening, though on a very small
scale so far.
In general, the cost of buying a manufactured free hardware design
is more closely related to the cost of buying a computer than to the
cost of buying a CD, but the cost of buying computers isn't an argument
against free software. Why should it be for hardware design?
So I don't believe that cost is the main issue here;
there are (at least) two other more important ones:
1) is it desireable to gpl (or similar) hardware designs?
2) What happens if no equivalent to the gpl is worked out?
Obviously, I think the answer to 1) is 'yes' or I wouldn't be bothering
to argue (one of the main, but certainly not the only) reasons being that
without such licenses there has been no way to combat the trends for
manufacturers to release ever less information with their products,
preventing both interfacing and repair (this is the problem Open Hardware
tries to solve from 'the other end')).
2) is more urgent: whether or not free software groups believe
design licensing is possible, companies like Sun and IBM certainly
do. Their licenses for their nominally open designs are more restrictive
(IMO) than their equivalent software licenses. For example, it appears
to me that the IBM license actually makes it impossible for manufacturers
of their nominally 'open' boards to distribute circuit diagrams or
layouts of the boards to purchasers of the boards (this is my
Without the opposite pole of a gpl-like license this type of license
has been accepted with little argument.
At the other extreme, people who do see themselves as creating open hardware
designs are inventing their own licenses in the lack of a generally accepted
one, each one with conflicting clauses
(eg. 'no military use allowed') which may well not be legally valid, but
if their wishes are respected the end result is that combination of
different designs becomes impossible.
More information about the License-discuss