If Linus snubs new GPL, is that it for 'open source'?

Torvalds doesn't care. Trouble looms...

By Andrew Orlowski
Published Monday 6th February 2006 23:52 GMT

Analysis Linus Torvalds doesn't want to change the Linux kernel's software license, and he said so again last week. For good measure this time, he threw in some inflammatory remarks.

"I literally feel," wrote Torvalds, "that we do not, as software developers, have the moral right to enforce our rules on hardware manufacturers. We are not crusaders, trying to force people to bow to our superior God."

Since the crusades were a foreign adventure responsible for the deaths of tens of thousands, that's not the most diplomatic response, and FSF counsel Eben Moglen refused to be drawn into retaliation when we contacted him for comment.

Moglen did say that as part of the lengthy, worldwide consultation process for GPL v3.0 he'd be issuing further clarification on the two most controversial parts of the new license, Sections three and seven. We'll examine the particulars in a moment.

But stressing that he was speaking in general terms, Moglen told us this -

"Freedom is not about what works well. It's about what defends freedom when it can be given an intellectually rigorous and internally rigorous conception. We want to have a conversation on whether we are drafting it in a way to achieve this," he said.

"The question presented by DRM is not whether it can have good purposes, or whether it serves socially useful ends sometimes. It's whether user disempowerment - at a time when technology is moving to embrace the users' whole life - is a risk we can run to gain some particular benefit."

Don't let the means dictate the ends, he seems to be saying.

Torvalds remarks have uncomfortable echoes of last year's BitKeeper episode, when Torvalds dimissed the concerns of his kernel developers and mocked the ethical dimension of software development.

But if Linux isn't about ethics, then what is its purpose? And if open source simply means 'free' (as in beer) code at the end of the day, and it's not about changing the world, then why is it different to a BSD?

Let's examine how we got here.

How GPL got Linux out of the lab

The Linux kernel which Torvalds controls (Torvalds also owns the Linux trademark) is the best known and most popular piece of software libre in the world, and owes its popularity and respect in no small part to the freedoms guaranteed by the FSF General Public License. This license gives the recipient the right to modify and distribute the code, but more importantly, ensures a downstream recipient fulfills the same obligations.

The FSF doesn't update this GPL very often. It last did so with version 2.0 in 1991, with a minor addendum (the LGPL) appearing a few years later, and version 3.0 has been racing our way with all the speed of a continental plate stuck in a tectonic traffic jam for several years now. Its ratification looks some way away too. As it turns out, this is quite deliberate, as Linux is big business now, and the FSF is engaging on a massive consensus building project to make sure everyone's on board.

The FSF also has an additional issue to deal with that it didn't have in 1991, which is that the words "free" and "open" are today often used as broad-brush term, with the implication that they're synonymous and interchangeable. They're not, but when Linux looked set to conquer all before it, and was finding its way into computer systems ranging from phones to mainframes, and world domination was only a matter of time, the difference could be blamed on semantic nit-picking. Didn't open and freedom just mean the same thing?

Something else happened, too. The phrase "open source" became an invitation for any opportunistic wanker to hitch a free ride, hoping some of this magic would rub off, and turn into a lucrative pay day.

We saw the influential, Blair-ite think tank Demos team up with Douglas Rushkoff to suggest "open source democracy", which amounted to little more than a catchphrase. A rag bag, free-for-all trivia website morphed into "Wikipedia", which laid claim to be the world's greatest encyclopedia (that's turning out to be exactly what you'd expect it to be). Some sophists claim to have created "open source" cookies - the baked, not coded kind. And even the GPL has been disastrously misapplied, to things that can be, but primarily don't need to be "modified" to be successful, such as works of art.

But while all this opportunism and sloppy thinking took place in public, the gears were slowing. Something was halting the momentum of this great project.

Microsoft began to apply its deep pockets to buy off litigious rivals. And nervous corporate and public sector customers, who'd been looking at Linux with great interest, began to waver. Maybe they got nervous about the fall-out from the SCO suit. Maybe Linux advocates failed to prove the total cost of ownership case, which had looked a slam dunk at one time. Maybe the notorious factionalism of the technical community (eg GNOME vs KDE) proved to be a turn off. Maybe Linux, and software libre, failed to generate big ideas of its own. Big ideas, even if they're nebulous and entirely without substance - and Web 2.0 is a great example of a load of nothing going nowhere, as you so eloquently point out - seem to be necessary to attract the glaze-eyed attention of the corporate media, if only for a few weeks. Or maybe too many nutballs climbed on board, hoping to catch a bit of the "New Open Thing".

We don't know, but in the end it wasn't Microsoft that fomented today's dispute about GPL 3.0, but of all things, a small consumer electronics company.

Enter TiVo.

Enter TiVo

The GPL always distinguished itself from other licenses by stressing a peculiar symmetry: the freedom to modify or distribute the source code would be passed to the end user in the form the upstream benefactor had intended. A perpetuity of sorts was established. You didn't have to tinker, but if you did, and made your tinkering available, you'd have to obey the terms on which you received the code.

This distinguished the GPL from BSD-style licensees, which were "open" in the sense you could look at the code, and "free" in the sense you didn't pay for it, but weren't, as in the now famous phrase, "free as freedom". And then a product was introduced that broke this social contract, while obeying the letter of the GPL version 2.0. This was TiVo.

When TiVo introduced its PVR time-shifting set top box, it did so using a Linux PC with a proprietary front-end. You could only tinker on the terms set by TiVo. This didn't deter a wave of enthusiasts, a small portion of the technical community (we'll unfairly, for convenience, call them the "O'Reilly crowd") who latch onto anything that demands your attention because it's "hackable", without quite seeing whether there are strings attached or where these strings might lead.

Linus Torvalds professed himself delighted, and naturally he's proud to see his kernel instantiated into real products. As you'd expect, he feels it's a validation of his adult life's work.

But GPL supporters who flocked to the cause because of "freedom" don't quite see it this way. What's the point of GPL, if it only turns out to be a rebranding of BSD? A sort of BSD with added, 21st century street cred? And a fat, drunken-looking Penguin as its mascot?

And doubly painfully, what's the point of a GPL product that ushers in a world of artificial technical restrictions on copyright material, DRM?

Linus actually had something to say on this, but we need to dive into the psychodrama that is Modern Copyright Discourse first, before we can understand why this debate looks so peculiarly lopsy, and so very heated.

Will the Matrix kill free software?

There are two sides to this argument, which we'll call "free" and "open", and both have good claims to make. But both sides like to throw off wild, metaphorical flares that light up the news pages, but are of no use to anyone. Let's separate the flames and see what lies behind their rationale.

If what I'm told by the GPL 3.0 advocates is true, then the world is about to end fairly shortly.

One proponent told me that the difference between now, 2006, and 2009, is that the value of your home in 2009 will be determined by the "freedom" your gadgets exhibit. This is a startling idea, one I'm sure today's real estate agents haven't yet pencilled in as a pre-printed tick-box on their forms. I'm paraphrasing, but the argument is that if the property owner didn't have "control" over all the technology in their home, then the home would have no value, or a lesser value than a comparable home on offer.

"I don't want to use the phrase 'Matrix'," said one, who went on to use the phrase Martrix - by saying, fairly emphatically, "it would be like living in the Matrix".

Utter nonsense, of course.

Your average property owner wants to get home, flick a switch, and find that "stuff" comes out - the stuff being, for example, light, heat or cooling if (s)he flicks a switch, or entertainment if (s)he flicks on the remote. Homes that don't fulfill these basic obligations have a tendancy not to get sold - they're probably car parks. In fact, you'd have to coral prospective home buyers in at gunpoint, and keep them there, to accept such a lousy proposition.

But Torvalds' solution is equally obtuse.

Faced with the moral problems posed by DRM, Torvalds opts for the 'stuckist' approach, of splendid isolation. Meaning he'd never watch The Sopranos, or anything worth watching except for a giggle, ever again.

Torvalds tells dissenters to go and build their own chips.

"Vote with your feet," he urges. "Join the OpenCores groups. Make your own FPGA's."

I'm right behind you with my soldering iron, Mr Torvalds.

Then he added the now notorious sign-off about the crusades.

" ... we do not - as software developers - have the moral right to enforce our rules on hardware manufacturers. We are not crusaders, trying to force people to bow to our superior God."

[Emphasis added, but hardly necessary].

As you can imagine, this accusation of moral laxity aimed at the developers who actually do the work on the Linux kernel, and many other projects across GPL land, has not been well received.

What appears to be moral, in Torvalds' own book, appears to be contingent on whatever Torvalds is feeling that day, and that's contingent on the market penetration of his kernel. Issues of morality are best left to genocidal "crusaders", who Torvalds feels are someone else entirely.

But if software libre isn't a moral crusade, what the heck is it? A 30 year old operating system, passing off as new one? A lifetime of dependency conflicts? A charity? The public can be cruel, and horribly judgmental, when it flicks the switch, and "stuff" doesn't come out. Torvalds may not imply that morality has no place in the Linux kernel, but the invitation to infer this doesn't really need our bold HTML markup. It's obvious.

A Linux without a moral element is a puzzling thing indeed. How would you or I begin to explain its value. Without "freedom" all one is left with is "free".

I don't think many Linux advocates would settle for this as the last line of defence. But in a swoop, Torvalds appears to have deprived "open source" advocates of arguing from a moral position.

Perhaps, by the time the long consultation process for GPL 3.0 reaches a conclusion, it will be clear that the word "open" was never really a substitute for "free". Until then, there's trouble ahead. ®