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

Re: Barriers to an ASP loophole closure (was Re: OSD && DFSG - different purposes - constructive suggestion!)

On Tue, 2003-03-11 at 11:58, Steve Langasek wrote:

> I find this an acceptable compromise.  The GPL already implements
> something very close to this:  if you give someone a copy, they're able
> to pass it on to a third party who in some cases then has grounds for
> demanding source from the author.  Extending that to letting a PCH demand
> access to changes if someone tells him about them doesn't seem too much
> of a stretch.

Close; the case in question for the GPL only arises in the case where
someone actually makes a written offer valid for at least three years.
Have you *ever* seen such an offer?  I haven't.  Was the entity making
that offer still around three years later?  

Further, it would seem that anyone making such an offer would know about
it, since they do have to write it down, and that explicitly states that
the offer is valid to "any third party".  In other words, I can only be
screwed over by that clause if I agree to it in writing.  Extending to
the case where that's a mandatory part of the license is a rather
different story.

I've been trying to write down my objections to re-classifying public
performance (a.k.a. "ordinary use" w.r.t. server type software) as
distribution, and I'm beginning to conclude that GPL'd software is
inappropriate for casual copying, even without the added burden imposed
by closing the ASP loophole.

3a) requires that I give the complete copy of the source along with the
binary; if public performance is classified as distribution, I'd be
forced to ship the complete source code for the Linux kernel, Apache,
PHP, and who knows what all else with every web page so as to be clearly
in compliance with this section.  Adding a link to the source on someone
else's site would be very hard to defend in court as falling under this
section; hosting the source code myself means I've just agreed to become
a kernel.org mirror.  As an aside, this might mean that binary-only
mirrors are in technical violation of the GPL, because the source code
must be distributed with the binaries, and distribution of the source
code by a different entity probably doesn't count ("in the same place",
from the comments after section ).

3b) requires that I make a multi-year commitment to being reachable
*and* agreeable to working without profit during that time period.  I
have no control over how far my written offer will be propagated without
my knowledge, yet I must honor that written offer, as any weakening of
this clause will lead to even larger loopholes than the ASP loophole. 
If the "at cost" clause is interpreted strictly, then I'm unwilling to
make such an offer, since it means that I have to do charity work while
the bank comes in to foreclose on my house.  If it is interpreted
loosely, then someone from the music or movie industry comes in and says
that their costs are very high, and that you'll have to pay a million
dollars for the source code.  Per file.  Either way, I've never seen
such an offer, and there's no way that I'll ever agree to make such an
offer without having enough cash in hand to ensure that I can survive
those three years on "at cost" work.

3c) requires that some other sucker has agreed to 3b) *and* that that
offer has been either made to me or relayed to me.  Furthermore, even if
I do have such an offer, I may not take advantage of it if I happen to
be engaged in commercial activity related to the distribution.  Right
now, I don't think that's much of a problem, but if public performance
is classified as distribution, then I think that it would be far too
easy to be engaged in "commercial distribution" -- e.g., any business
web site would be engaged in commercial distribution of Apache under
that definition.

I haven't yet gotten everything together in my head yet, but I'm afraid
that closing the ASP loophole will result in the possibility of being in
violation of the GPL-v3 simply by making a machine with too much
software on it accessible from the net.

Furthermore, if such a change is sloppily written, "too much software"
might mean "a bare-bones bootable system", as responding to pings is a
service, too, and is hard to distinguish from software-that-cures-cancer
as an ASP (technically, that is; both receive network packets and
respond to them with other network packets; I am sure that there are
other protocols that aren't quite as trivial as ping but still far less
than something we'd actually call an ASP that could be used as an
example as well).  

Stephen Ryan                                        Debian Linux 3.0
Technology Coordinator
Center for Educational Outcomes
at Dartmouth College

Reply to: