Posted Mar 22, 2007 19:39 UTC (Thu) by lysse
In reply to: GNU/Busybox ?!?
Parent article: The road to freedom in the embedded world
OK, I've explained that (a) I'm a nobody, (b) I don't have a problem with you, and (c) I'm not connected with the FSF at all. Should I also have explained that I don't see your actions as being in violation of the principle with which I did associate myself, even though I don't see Stallman's actions as violating it either? If so - you got it; I'm sure you are as much on the side of the angels as he is; and attempting to play "you're either with me or against me" games is silly and wrong whoever does it.
As to the DRM issue: If the reasoning behind the use of DRM on a device is to only allow the distribution of authorised binaries signed with a universal key, then fine - provide a unique key with each machine that allows binaries to be signed for, and used on, only that machine. That way, I can recompile and replace the software, and distribute everything needed to do so - I just can't distribute a precompiled binary, which may be a technical annoyance but is not a limitation of my freedom to share my work (because anyone else can do what I did to generate a binary).
As I understand the FSF, their concern is that the freedom of a user to receive, alter and redistribute software is meaningless if you can't actually USE it; such a solution would solve that problem whilst allowing the use of signed binaries. Would such a scheme be acceptable under the proposed GPL3? If not, then it appears that I also have a problem with that proposal. But if it would, then for the life of me I can't see what all the fuss is about!
And that's just one suggestion made by a nobody (which is probably why for the couple of months I've been saying it, nobody's taken the blindest bit of notice of me). I can't believe there aren't other conceivable solutions that will satisfy both sides equally well.
You cite ROM distribution as a counter-example, but it seems to me that ROM is being used as just that - no more than a means of distribution of a binary. If you pull out the ROM and replace it with an EPROM you've burned yourself, it'll run quite happily; all you need is a means to write a new copy of the chosen distribution method. But if I were to blow that EPROM and find that the CPU refuses to accept it because the chip carrier bore the wrong manufacturer ID on its surface, I have an "authorised binaries" analogy. So I cannot accept your comparison as valid (and it saddens me that your pre-emptive response to such a disagreement is not that it's wrong, but that it's "just plain stupid").
Of course, with a machine-specific key, I can download an unsigned ROM image, sign it, blow it and use it; my freedom is preserved, yet so is the ability of the manufacturer to authorise binaries (my signed ROM is useless on any other device; the unsigned ROM image can't be used directly or by stealth; signing a ROM image becomes a deliberate step, a signifier of taking individual responsibility). I repeat that in my view the GPL3 should permit this scheme.
...That aside - my point was that the FSF are quite free to decline to dilute their licence to suit purposes with which they fundamentally disagree, just as you are free to decline to use their licence if you fundamentally disagree with their goals. I don't see their quest to ensure that the use of DRM does not impact the rights of an individual user to hack on their device as anything other than a reaffirmation of their founding principle - namely that a user of a piece of software should have the right to customise it, and to share their customisations. That's a principle with which I wholeheartedly agree. The problem I have with "authorised" binaries is that they take away my freedom to decline all the choices offered by the authoriser and make my own options.
But of course, the broader point is this - freedom isn't free. Every choice carries a cost; creating one's own choice carries one that most people would view as far too high. If you're determined to use DRM to prohibit me from customising my machine, I don't have any right to stop you - but I can insist that you don't use anything of mine in the attempt. That's what I believe the FSF are seeking to do with GPL3; it's a continuation of a philosophy they've always espoused; demanding that they not do it is akin to asking them to change their philosophy. You're not doing that, which is great; you're insisting that your software remain GPL2, which is fine, and will certainly permit someone to sell you a device containing your own work and then stop you from updating it. That's your choice, and I'm all in favour of you making it. Everyone who currently uses GPL2 will have the same choice to make; some of them will see things your way, whilst others will disagree; and I'm sure some GPL2+ projects will end up forking into GPL3 and GPL2 versions, which will be a shame. (And of course, GPL3 isn't even final yet, so the entire discussion is possibly premature.)
Moreover, licences can't be applied retroactively; so you can take anything GPL'd before the GPL3 is out and use and redistribute it under GPL2+ terms, even if it's immediately relicensed to GPL3 thereafter - you just can't stop anyone else relicensing it to GPL3 when they redistribute it. So anyone currently using GPL'd software in DRM'd situations can continue using what they have right now, even after GPL3; but they might be frozen out of future developments.
Likewise, a product developer who chooses to use DRM to lock out individual developers in the future will have to make that choice in the knowledge that they're also locking themselves out from GPL3'd software; and those who stay with GPL2, or omit the "or any later version" bit, will be granting implicit permission for those product developers to use their software. That's as it should be - everyone able to make the choice that works best for them, with a knowledge of the costs incurred in doing so.
to post comments)