Closing the GPL loophole: it doesn't define 'free software'. [LWN.net]

Closing the GPL loophole: it doesn't define 'free software'.

Posted Feb 18, 2006 2:26 UTC (Sat) by pimlott (guest, #1535)
In reply to: Closing the GPL loophole: it doesn't define 'free software'. by xoddam
Parent article: FSF: GPLv3 Update #2

I find Karim more convincing on what the FSF represents to accomplish with the anti-DRM clause. The FSF are not impractical idealists who take merely symbolic stands. For many years, when asked why the FSF wasn't concerned about embedded software, RMS would answer: It's practically impossible to replace the software, so it's not a real freedom issue. Today, of course, it's possible to replace the software in many devices. So when the FSF puts their effort into drafting an anti-DRM clause, I have to think they believe it will provide an effective freedom.

Karim has (rather brilliantly, IMO) shown that device makers can comply with the anti-DRM clause while making modified versions of the software completely useless on the device. This is no effective freedom at all for the purchaser of the device. The anti-DRM clause strains to prevent the device from retaliating against the modified software. Karim has show that it can retaliate forcefully. You claim that the anti-DRM clause is a symbolic victory, because at least one can run a modified version of the GPLv3 software--never mind that it's useless. I say that the FSF isn't about symbolic victories, it's about real freedom.

So Karim has convinced me that the anti-DRM clause is a mistake. Don't forget that the GPLv3 should be around for a decade. If embedded device makers (and other suppliers of closed systems) implement Karim's treacherous scheme in year one, that clause will be pretty embarrassing for the next nine. I think the FSF should decide whether to go a little further with the anti-DRM clause and outlaw Karim's scheme (no opinion here on whether that is legally possible), or give up.

(Log in to post comments)

What Karim has and has not shown

Posted Feb 19, 2006 23:48 UTC (Sun) by xoddam (subscriber, #2322) [Link]

> Karim has (rather brilliantly, IMO) shown that device makers can
> comply with the anti-DRM clause while making modified versions of
> the software completely useless on the device.

He has done no such thing.

Karim has (rather brilliantly, I agree) shown that device makers can
comply with the anti-digital-restrictions clause of GPL'd software on a
device which implements digital restrictions, *providing* that the user
is not restricted thereby from using, modifying and copying the GPL'd
software.

What Karim has has *not* shown is that a distributor may comply with the
GPLv3 whilst rendering modified versions of the covered work useless.
The licence text "such that its functioning in all circumstances is
identical" explicitly disallows such disabling of a GPL'd work.

(As sepreece has pointed out, the wording here could be better and it is
therefore unlikely to remain in the present draft form. Anything we
speculate on here is obviously contingent on the final licence text)

Since the GPLv3'd software would be user-modifiable (and the user would
be protected against DMCA-type penalties), any digital restrictions
scheme implemented with GPLv3'd software may be removed by the user -- so
a device maker would be best advised to use non-GPL'd software for this
purpose.

> If embedded device makers (and other suppliers of closed systems)
> implement Karim's treacherous scheme in year one, that clause will
> be pretty embarrassing for the next nine.

The purpose of the GPL, including the digital restrictions clause, is to
preserve the freedom of users to share and improve GPL'd software. All a
copyright licence can do is grant permission to copy and modify something
*where this would otherwise be prohibited by law*.

The general fight against DRM must proceed elsewhere. In the meantime
it's now the year *before* year one, the licence is still being drafted,
and all these considerations are already being pointed out. Early days
yet.

What Karim has and has not shown

Posted Feb 20, 2006 22:17 UTC (Mon) by pimlott (guest, #1535) [Link]

What Karim has has *not* shown is that a distributor may comply with the GPLv3 whilst rendering modified versions of the covered work useless. The licence text "such that its functioning in all circumstances is identical" explicitly disallows such disabling of a GPL'd work.

Well, this question centers around the interpretation of "useless" and "identical functioning". How far can the device degrade the usefulness of the modified GPLv3 software without triggering the anti-DRM clause? Karim proposed a device that "wouldn't allow much of anything but booting your kernel and xoring bits in RAM with no access to display or networking or anything useful". That makes the kernel (or any other bit of software) useless to me--the only way you know it's even running is it gets warm and the battery eventually dies. :-) But it's hard to see how this violates the "identical functioning" requirement (do you think so?), and if it does, I expect there is a more subtle ploy that doesn't yet has the same effect.

What Karim has and has not shown

Posted Feb 21, 2006 0:36 UTC (Tue) by xoddam (subscriber, #2322) [Link]

> Karim proposed a device that "wouldn't allow much of anything but
> booting your kernel and xoring bits in RAM with no access to display
> or networking or anything useful"

IMO that would absolutely violate the licence, if the same program
unmodified would be able to use the I/O devices in question. That's
certainly the intention: Moglen said (in the talk, Ciaran O'Riordan's
transcript is linked below), 'I would translate that for vernacular use,
into "plays all the same movies".' If the legalese doesn't already mean
what Moglen says it does, I'm sure he'll fix it before we reach
GPLv3-rc1.

Karim also suggested, less extremely, a device where GPLv3 software is
present, usable and free, but is prevented from reading the cleartext of
digitally restricted media (regardless of whether, and by what key, it is
signed). That would not violate the licence.

The point I'm trying to make is that the licence is intended to protect
the freedom of the user with respect to the licenced software, not to
prevent the use of DRM altogether.

What Karim has and has not shown

Posted Feb 21, 2006 3:05 UTC (Tue) by pimlott (guest, #1535) [Link]

IMO that would absolutely violate the licence

So this is what it comes down to: The device maker might argue that the kernel itself is behaving identically, only the other components are declining to talk to it. Or, more subtly, say, the display could continue to talk to the kernel as if operating normally, but actually remain blanked. It is clear here that the device as a whole is functioning differently, but is the kernel? I and I think Karim are afraid that the answer is no, and that this loophole cannot be closed.