On Fri, Jan 30, 2004 at 11:23:29AM -0600, Steve Langasek wrote:
> > If, tomorrow, we get an email from Bill Gates saying "hey, if you want
> > to include all the Windows 95 and 98 stuff in non-free, that'd be great;
> This seems like a pretty good example of how sometimes, distributing
> non-free software ultimately benefits... no one except for non-free
> software authors.
Well, that's utterly wrong. It'd benefit me, personally, because I'd not
have to spend a couple hundred dollars on Windows 98 licenses in order
to be able to reliably run certain Windows software that don't work
under Wine and aren't available for Linux (and don't have replacements).
Now, sure, maybe Bill's doing that because he thinks it'd benefit him. Good
for him, that certainly doesn't bother me at all.
> While there are certainly those who would derive short-term benefit from
> having genuine MS DLLs available to apt-get, there is a larger question
> of whether the long-term impact would negate the short-term benefits.
*shrug*
Maybe having a good GPLed C compiler makes it less likely that someone'll
write a BSD-licensed C compiler. Does this harm the BSD projects? Maybe.
Odds on it doesn't though: if the "less free" version does everything they
need, and they can spend their time on things that have more value than a
free-er compiler, that's great.
The question people have to ask is whether:
V(GPLed C compiler) + V(x hours' work on other BSD stuff)
<=>
V(x hours spent writing a new C compiler, under a BSD license)
Writing a new compiler that's as effective as gcc presumably requires a
lot of hours, which means that V(x hours' work on other BSD stuff) can
be pretty large. But the difference between a BSD licensed compiler and
a GPLed compiler isn't very much -- so you can redistribute it without
source, big deal. Some people might like that a little, but, well, so
what.
Compare the same thing for Wine and Win98:
V(freely redistributable Win98)
<=>
V(Win98 rewrite under a free license)
Now there's a few benefits from having source to Win98, but personally,
I don't care about any of them: I want more-or-less bug-for-bug
compatability so I can run Windows apps more natively under Linux. Nothing
more. So I'd rate the above comparison as an equality, to within a very
close approximation. If Bill was offering the first one for nothing,
and the second required a bunch of hackers to spend their time rewriting
Win98 instead of hacking on OpenOffice or the kernel or KDE or whatever,
then there's no comparison:
V(freely redistributable Win98) + V(hundreds of hours spent on
KDE, the Linux kernel, gcc, OpenOffice, libgtk, etc)
>>
V(Win98 rewrite under a free license)
> This is one way in which the "our users and free software" duality is
> useful: it reminds us that our decisions need to be evaluated for their
> effect on *both* of our intertwined priorities, because what seems to be
> obviously advantageous for our users may prove not-so-advantageous
> because of the detrimental effect it has on free software with a
> concomitant impact on the user's experience (or vice versa).
Personally, I think more free is almost always better than less
free. Having stuff widely distributed, easily available, and more readily
redistributable than proprietary software generally is is a win.
> Not everything that helps a user solve a perceived immediate problem is
> a clear win.
I dunno, I think it'd have to have clear, unambiguous and very very
severe problems for it not to be. In the end, every improvement we
make is about solving problems for users that are either immediate now,
or that we're worried will become immediate later.
I don't really see how putting Win98 in non-free on Debian (having
been given permission) could be any worse than it being available for
free download on microsoft.com as far as the free software community
is concerned, though. And having it be on Debian can help the community
since we can integrate it with our systems, or avoid some obnoxious things
that upstream might like to do (like collect email addresses, or require
you to download everything at once rather than in smaller package-sized
components, or not make it readily available outside the US, or whatever).
Cheers,
aj
--
Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.
Linux.conf.au 2004 -- Because we could.
http://conf.linux.org.au/ -- Jan 12-17, 2004
Attachment:
signature.asc
Description: Digital signature