Document Licenses and the Future of Free Culture
|Document Licenses and the Future of Free Culture (2006)
Talk at Wikimania 2006, August 4, 2006
So welcome to the 2:30 session on licences and interoperability.
We are honored not just to have Larry sticking through to continue the discussion that he essentially began at the end of his keynote, but also Eben Moglen, law professor at Columbia University, counsel from its inception to the Free Software Foundation and the founder of the… How does it go again? The F…
The Software Freedom Law Center. So, it’s just great to have both of you in one place, and I think we should just begin.
Eben, Larry laid it down at the end of his talk and said: “Why can’t we all just get along?” and had a concrete proposal for doing so. Would love to hear your thoughts, and any other way in which you’d like to get us started on attacking this issue.
Well, thank you. Look, it’s…
I was with Larry 100% all the way through the talk until I discovered I was the one who was going to do the work, at which point I began to have misgivings for the first time. The point that Larry is making this afternoon with his accustomed grace and drama, he has also been making for about a year and a half now with his equally customary farsightedness. As he pointed out, he has been working within his community to produce the platform for interoperable free culture in several directions; he said in his talk, in an undertone, that he had attempted to get the Free Software Foundation’s technical enthusiasm behind the platform for free culture, and that’s right.
The progress of the free Flash viewer and foundry called Gnash is entirely owing to Larry’s effort to instigate the Free Software Foundation to fund and sponsor Gnash development, which it has been doing, and which is going to pay off very large very soon, in offering a free platform for content creation of a kind which lots of people now do in unfreedom simply because they have no comparable free tool.
It is also true that this question of license interoperation has come up and been discussed because Larry forced it onto the agenda, and I think it is a powerful and important plea that he’s making. There are a couple of things to say about licenses that he didn’t say, though I think I ought to start by endorsing his proposition that the best license in this area is a largely invisible license. That is to say that it is the job of the legal technology to get out of the way and to allow creation to occur. But in the, now not very long but, as he would say, “getting less weird by the day”, history of the creation of the free licenses, that’s half of the job that licenses do.
The other half of the job that licenses do, and this was also touched on in Larry’s remarks, the other half of the job that licenses do is to protect the freedom of what has been created. To prevent appropriation in ways which are destructive of the underlying political economy of free creation. That proposition, that licenses must both facilitate creation and defend the freedom of what has been created, had, as an outcome with respect to program code, an emphasis on the protection side of the ledger.
Stallman’s worry from the beginning was that facilitating the creation of free code could be done in a lot of different ways. And indeed, if you think about it, there are a fairly large number of free software licenses, and they have a certain diversity of body plan. There’s the BSD plan, which basically corresponds, I think, the desire for transparency: say as little as possible, permit as much as possible, and get out of the way as soon as you can. The MIT X11 license is even more demonstrative of the impulse to facilitate: do what you will, end of sentence, end of license. The problem, as Stallman saw it, in the mid‐1980’s, was the facilitation was the easy part of the racket, protection was the hard part. And building a device which was tolerably simple and which was adequately protective against all the various likely means of attack on freedom was not so simple.
GPL2, which I had nothing to do with, achieved that outcome rather well. GPL3, which I have a lot to do with, seems to be attempting to achieve that outcome with a great profusion of additional words, the only excuse for which is: There’s a lot more to do these days to protect freedom, because it is spread more far, and there is more worry about it. All of this may, however, be more true about executable code than about works of other forms of authorship. That is to say, protecting the freedom of free software and protecting the freedom of free literature and protecting the freedom of free photographic images may be different jobs with a different quantum level of intensity to deal with.
In general, the proposition seems, at the outset, rather similar. The goal is to prevent people from taking free material and incorporating it in unfree contexts in such a way as to reproprietize what has been freely chosen. And accordingly, when Stallman set himself to the free culture problem – defining culture fairly narrowly as technical reference manuals at the first go round – we got a license in the GNU Free Document License which was equally intensively committed to protecting as to facilitating.
I am here, I should say, expressing one person’s opinions, I’m not speaking on the behalf of the Free Software Foundation, and therefore I am able to say a thing, which when acting on the foundation’s behalf, I rarely get to say, which is: I never really liked the GNU Free Document Licence very much. I didn’t like it because it did not have that property of elegant design that the GPL had. And I now understand why the FDL was a bad license – it’s because Stallman and I wrote it together. And I see that that’s the problem because I see GPL3 presenting many of the same challenges, and I wish I could just get out of the way and leave him to write some perfectly elegant license that would do all the work; it just doesn’t seem fated to be true.
To be more serious about it, the FDL wound up in the state in which, in order to attempt to protect the freedom of free reference manuals as intensively as possible, it got patched and repatched to the point at which it lost in simplicity and usability more than it gained in additional protectiveness. The appropriate response was to take it to pieces and rebuild. But for a number of reasons, including the early explosive success of the Wikipedia, that was not easy to do. It was unfortunately sailing at full speed in a high wind and taking it apart didn’t seem practicable.
We have, however, now 95% done that. We hope, within a very little bit of time, to be able to release a better Free Document License, which will actually, I think, be three free document licenses like nested dolls, with increasing levels of simplicity as you go in.
Because one of the problems about protecting freedom turned out to have to do with protecting freedom in different media of presentation. When the goal was to figure out a way to create a free document license that could be printed inside glossy covers by commercial publishers trying to make money out of selling as a commercial book what was also a free document, certain elements entered into the license that you wouldn’t otherwise have put there. When there was an attempt to combine two kinds of Stallman’s three kinds of content, that is to say, the political opinionated and the neutral technical information in a single physical binding, the result was the provisions about invariant sections that troubled the apostles of free media, and annoyed lawyers and engineers both.
In other words, the FDL as we have known it, and as it is currently applied to the Wikipedia, is an elegant demonstration – though not an elegant license – of the problem that you get into by attempting to balance protectiveness against facilitation in multiple media at the same time for works with fundamentally different purposes or intentions of creation. This is a problem that can be solved by brute force in legal technology but the brute force solutions are of limited range and imperfect utility.
The first job that Larry put forward, which is the unification of the CC‐by‐sa license and the FDL, is, I think, practically attainable. Lawyers for the Free Software Foundation and lawyers for Creative Commons at the Software Freedom Law Center and at Creative Commons have been talking about that particular task intensively for a couple of months now and I think we are going to achieve the right result. Fortunately I have the Free Document Licenses available for modification at this time, and so I think that it will be possible to do.
If it is possible to unify those two licenses in the form that Larry suggests, by permissive interoperation, rules that say: “Works under this license may, when they are turned new works based on this work, be released also under the other license or instead under the other license.”, and notice that the difference between “also” and “instead” may be very significant. That kind of exchangeability between two important licenses, I think is coming.
Now Larry says it will be necessary from the beginning to do that work in a generalized way. And if you are following the inside baseball of his remarks as well as the beautiful wrapping in which he puts them, you will know that that’s the real challenge that he’s putting forward. As I say, I noticed it slopping over the front row and into my lap in the middle of the talk in a way I hadn’t fully anticipated, but he points out, and I don’t therefore have to, that you do all the work anyway, whoever it’s assigned to; we’re all going to have to figure out how to create the kind of unified low barrier legal regime he wants, and I want, and he thinks we all want, and I hope he’s right about that.
There are some problems. I never met anybody who’d spent a lot of time inventing a license who wanted voluntarily to stop using it. I never met anybody who had, legitimately, any pride to take in any license – including the dumbest revision of yesterday’s proprietary license in the filing cabinet – who didn’t feel substantial pride of authorship. I will let you in on a secret: When you write a computer program, there is an enormous ecstatic result when it works. When you write a license, there’s an enormous ecstatic response when you think it might work, because you never really get rapid feedback. So for those of us who grew up with edit, compile, test, edit, compile, test, license making is more joy for less work, because you don’t have to test in the near term.
So there are a lot of guys out there who are very proud of licenses they have written which have not been in any sense, and I don’t mean only in a litigation sense, tested. But they’re proud of them, and I understand the nature of their pride and I understand the nature of their resistance to giving them up.
Moreover, up until this point, you will have noticed, in the history of free software licensing, the “how long it took the guys on the other side to figure out which licenses were dangerous” moment, OK? Microsoft began by thinking “All this open, free, whatever it is, don’t worry about it.” After a lengthy period of time they started worrying about it very much. After another lengthy period of time they figured out what their problem was – it was the GPL. Right? They had learned enough to understand that the problem was there was a particular license which implemented freedom in a way which was particularly threatening to their business model.
Now the bad news for us on this side is, the free culture problem presents to publishers a difficulty which doesn’t depend on which license it is, it doesn’t depend on how well the licenses work. As Larry pointed out to you with respect to our experiment in Eldred and with respect to everything that has followed from it, The real threat to commercial culture is the mere size of the public domain all by itself. Which it why, though he didn’t say it to you, governments are so damn resistant to mapping and publishing the metes and bounds of the public domain for their citizens, because proprietary culture will, in the 21st century, compete against free culture everywhere all the time, in a physical sense.
Imagine that airport book shop selling commercial novels written by robots for reading on the red‐eye, if right next to it there’s a guy with Brewster’s bookmobile, reduced to hand cart size, so that Anna Karenina for a dollar competes against everything in the proprietary book shop at $15.95 every time you’re about to catch an airplane.
Right. And they don’t want anybody to know, either, yet. That’s a competition that is difficult to win. And that’s the easy one.
Imagine what happens to the textbook publishers, when the whole immense profitable oligopoly of educational publishing collapses in the free educational materials – Wikiversity – model that Jimmy is shooting at them.
So our difficulty with respect to free interoperable cultural licenses is that in order to get interoperability we need to reduce protectiveness as against facilitation. But we need to do that knowing that the protectiveness of the licenses has yet to be tested and that the real pressure on them is still to come. That means that the architecture and the legal engineering are nontrivial.
The desire is to make, as he points out to you, a strong, powerful, flexible machine which is never seen in practice for the creator but which responds with the strength of steel at the moment that – as we can absolutely be certain will happen – proprietary culture identifies the licenses as potential weak places and tries to go after them. And for that purpose we need allies.
The GPL was a different license after IBM woke up one morning and realized there was billions of dollars of disruption and possibly billions of dollars in outright profit in it. Not because the words of the license had changed, but because, within an instant, the context had changed.
So I identify with Larry’s goal, things we can do among ourselves, and things you are going to have to demand of the outside world. The end of his talk suggested that the hardest work you’ll ever have to do will be the work of demanding that some license authors show a little bit of flexibility and respect. I agree that that’s not going to be the easiest work in the world, but I don’t think that that’s the hardest work in the world either.
I think the hardest work in the world is making governments believe what Larry told you we all believe, and I hope he’s right. Because your ally the next time out isn’t going to be IBM. Once the free culture starts competing effectively and destructively against proprietary culture, it’s not that Bertelsmann is going to decide to do a deal with you to put Random House out of business.
Instead you’re going to have to use that power of citizenship to make governments willing to watch as the big transformation happens. And it is a bigger transformation than the software transformation. Because the software transformation was only visible to geeks. The transition we’re talking about will be visible in every classroom and on every newsstand on Earth. And unless governments believe with us that the licenses we are making and the modifications to copyright law that we are instituting are valuable to citizens and should be defended against rent seekers, the rent seekers will have a powerful response.
We, on the other hand, can not afford to sit and write tightly crafted copyleft licenses that say “In order to protect freedom you may distribute derivative works under this license only.”, because as Larry has pointed out, that natural, simple, straightforward way of making a protective copyleft imposes autism in the license arena that we cannot long afford. So we need good, strong, flexible copyleft, we need social and political context for protecting the works more, because as we attempt to increase flexibility and facilitation we will inevitably make some compromises on protectiveness, and we’re going to have to think those things through with the slightest possible tinge of Not Invented Here.
That’s what went through my mind as I watched those slides and listened to that talk. I know that Larry is right. I’m ready to follow him. I’m even ready to follow him on the mission assigned to me. But, he’s right about one thing for sure, it can’t be done without you. That is certain.
So those are my comments. Now I would really much rather listen to some other people.
Eben, thank you so much.
Larry’s been scrawling down some notes that are likely relevant to what you were saying
This is a shopping list.
It’s a shopping list, he says, but… You always speak as asked…
But before I ask him to respond, let me just ask you to be a little more detailed about something. There’s some element of your remarks that had demean of a state department news conference to it, and I mean that with all…
That’s the… I can’t get out of the GPL3 mode, right?
It’s sort of… You know, “We’re aware of these problems, the diplomats are behind closed doors, we’re cautiously optimistic” – to borrow a term from Jimmy Carter on the Iranian hostage crisis – “that we can come to an agreement, and new licenses are around the corner that will basically solve these problems.”, and I’m just wondering if you can give us a little more detail of that process, of the time table, of how optimistic you are given Larry’s note that he’s not as optimistic.
Well, all right, I mean… Some portion of this, I guess I should say, does seem to me like legal work in progress and I need to be a little careful about it, but let me try and rip the covers away as far as I can.
The reworking of the GNU Free Document License began in earnest more than a year ago. I truthfully hoped that we would be finished and that the license would be out before the GPL3 process began, because I did not want to find us where we now are, trying to do two very complicated and different jobs at the same time.
We narrowly missed that opportunity, and I think one of the reasons that we missed it, to be perfectly clear about it, is that we already weren’t clear how far the Creative Commons’ breadth of licensing inventory was within the range of things we were trying to achieve interoperation with. Stallman’s protectiveness of the GFDL is a fact of life, he’s protective of the license as he’s protective of his other licenses, and there’s a reason; ’cause he emphasizes protection of freedom all the time, and it’s always OK with him to add another layer of acetate to the bulletproof vest. Given that that’s true, I experience some difficulty in coming to a final deal, and I think we are now at last moving on places where I was stuck last year.
But as we reworked the Free Document License, we came to believe that more than one license might be necessary. And one afternoon we found ourselves, after two hard days of work, looking at something called the Simpler Free Document License. which, for the first time in a long time, I liked a lot. It looked to me like we had at last succeeded in simplifying the document license to the point at which its use for something like the largest wiki in the world might actually make some sense, because pieces about “if you print more than 100 copies” or what to do with the covers no longer were essential to the nature of the document’s behavior.
So at that point the question became: “How many free document licenses are there in this family of ours, and how do we exchange content among them successfully so that FDL content can go to a wiki with a free wiki license on it?”, right? So first we found we had an interoperation process of our own to resolve. That took some time, and there were a few little things that happened in the mean time, like the onset of GPL3. I think we now know what the family of free document licenses is within the Free Software Foundation’s role, I think we know how to make those interoperate.
Then we undertake the problem, next, with some diplomatic element to it, I admit, but mostly in a fully straightforward way, we begin the process of looking at CC‐by‐sa, and figuring out with Larry and his licensing lawyer Mia Garlick what we can accomplish there. I think, and this is the place where optimism wanes and waxes as we work it through, I think that we’re going to get someplace. All the simple problems are resolved now.
I am working on one last question: “What do you do when you combine works which are partly under free licenses with material under non‐free licenses?” Whether you can anthologize free work and unfree work turns out to be a really important question. There’s an obvious motive to say “You shouldn’t do it at all, the purpose of having free work is to create a free world, when do you want people making anthologies of free and not free?”
Then you think about that for five more minutes and you think “Well, no, that sounds like giving the publishers everything they want, right? Let’s just have a complete division of free and unfree; they will advertise unfree using the revenue stream of proprietary capitalism and Anna Karenina will have to take care of herself.”
And then you think “Well, that’s not the right answer. Maybe we do want to interpenetrate free and non‐free work in order to use the proprietary infrastructure to advertise free culture by giving some to people who aren’t yet accustomed to looking in the free places.” But once you do that, to maintain protectiveness in your license is very difficult indeed.
That’s the terrain in which I think we are currently marching around. I believe, as I say, that optimism’s warranted, but if I’d been on the red‐eye I wouldn’t think that either.
Are these free licenses, these drafts of the new free licenses available online yet?
No. No, no, no. I’m still living in a world where my client thinks that you pour no wine before it’s time, is not into wiki development of licenses.
You don’t say.
We spend a lot of time, we figure out what the rules of the conversation are, we emit the conversational starters and then we have the conversation.
It’s the GPL3 model, and it has benefits and harms. In a room full of wikipedians I wouldn’t want to claim that it is superior, but some cathedral developers prefer it where some bazaar developers prefer the wiki.
Gotcha. Larry, over to you.
You know, this is… I don’t have a lot to say, because, as is usual, Eben and I, I think, are in fundamental agreement about every important matter.
And even, increasingly I’ve decided all the smart people in the world have beards, so maybe I should increase and get a beard.
I see, that’s good.
When Jon gets to be old enough to grow a beard you can grow one soon.
So I agree with him that the problem, you know, to the extent there’s a problem, is the instinct that we all have as humans, which is an instinct to exercise proprietary control over what we build. That’s what I describe is the natural instinct here. And I described that same instinct when I described what the first solution I thought there was to this problem would be, which is, basically, “switch everybody over to CC licenses”, that was my instinct too.
But I think that’s why it’s so great that we’re having this conversation here, because I think of all the institutions of free culture that has demonstrated the ethic in giving up on that instinct. It’s what the Wikipedia project has done, right? The ethic of a wikipedian is: You write things licensed in a way that guarantees you don’t control what’s going to happen to it. And, I mean, that’s a very important ethic that we all should learn from, and it took a little bit of beating in a subtle way from Jimbo before I got that, but I think I believe that now. So then the question is: How do we achieve that here?
I agree the compilation problem is a hard one. I my view the really hard question, though, that’s going to be complicating this, where I wax and wane over optimism, is the very subtle and careful distinction that Eben drew between being allowed to license something under another license in addition, or instead of the original license. So, when you have something on the FDL, can you relicense a derivative also under a CC license, or can you relicense a derivative instead under a CC’s license?
If it’s also under a CC license, then what we’re talking about is dual licensing. So then we create a world where we have FDL content and CC’d content, and then a new world where we have FDL and CC’d content. And then those two, that sort of amalgamation of dual licensed content, whenever it’s used again, needs to continue to have that dual licensing structure all the way down.
Why is that?
Well, it’s just the way the ar…
It doesn’t have to, in any logical sense, I’m just describing where the conversation seems to be right now. And this is the part that concerns me, because if, in fact, that’s the architecture it takes, doesn’t have to, but if in fact that’s the architecture it takes, we invite others to continue to create this amalgamation ethic, and what we’re going to eventually have is, you know, everything licensed under 45 different licenses at the same time. I don’t think that’s the most efficient way to do it.
I think the more efficient way to do it is to embrace – again I credit, you know, Jimbo completely for getting me to see this point – rather than imagining that any of us is great, at sitting down and writing the perfect license, not at imagining that any of us, you know, whether it’s Eben and Richard or Richard alone or me and my lawyers or whatever, any of us is bright enough to figure out what the best license is, embrace a strategy that allows you to learn what the best license is.
So rather than the king demanding “This is the best license, everybody follows it”, instead create a system where you’ve got some competition among licenses. So when you see people shifting from the CC license to the FDL, that’s a signal to CC, it says: “People don’t like your license, they want to get out of it.” Well, why? What’s wrong with our license? Well, it’s because three quarts said there’s some problem with it, and so you’ve got to fix that problem.
So the point is, the, you know… Jimmy’s always understated about this, but the understated Jimmy line here is: This information that comes from this market, this Hayekian market, is what produces the value here, and it’s a better system for finding truth than the “I’m a genius and I can tell you what the best license is”. Now, I don’t believe in the Hayekian system in lots of different contexts, I mean, I think there’s a lot of limitations, and I’m, you know, I’m not a Hayekian as deeply as he is, but I think that there’s wisdom about that here.
Now, I agree, this is not… You guys, solving this problem is not going to be the hardest problem you face. And the hardest problem you face is really going to be the problem of getting policy makers to begin to see this. But here’s where, strangely – this is probably just exhaustion and delusion – I feel more optimistic than Eben does about what that battle’s going to look like, because, you know, this is a fact that – I’m embarrassed that I see Yochai’s here, I’ve said such nice things about him and I wouldn’t have said it if I’d known he was here –
He wasn’t here when you said them.
There’s a point Yochai’s been making for a long time, and it’s an absolutely correct point, that once the capitalist market wakes up to it, will begin to have some consequence:
The market that’s supported by the read‐only Internet is a tiny market. compared to the market that could be supported by the read‐write Internet. There are huge interests who would win from a much more reasonable balance of copyright laws here. Huge set of interest, you know, from computer manufacturers, to bandwidth sellers, to software people, to all sorts of people who want to have to sell faster, better, systems, to support this creativity.
It’s a tiny Internet that supports the perfectly efficient iTunes model of how you get access to culture. It’s a huge Internet that’s got to facilitate my ability to send my 50 megabit film of my kid from this week to all ten thousand of my friends that I want to see it, right?
So the point is we can begin to teach these people why this other system is better, and the dynamics you’re seeing in the read‐write Internet now are beginning to do that. I mean, there’s a huge, you know, struggle about the YouTube controversies, right, where YouTube is basically taking a kind of Napster‐like position right now about content being placed up there. Tons of content up there is plainly in violation of copyright laws, and all sorts of people like Lucas, just in the months recently, come in and say “Take it all down!”, and then all of a sudden they say “Well, wait a minute, probably not a good idea to force them to take it down. OK, you can keep it up.”
So the point is to begin to teach this market process, potentially, and I think the fact is that’s a more valuable, powerful market that will be on our side eventually. Not quite the equivalent of IBM, because they don’t think as coherently, that’s not saying much, but thinking as coherently as IBM does, but I still think there’s a potential there.
OK, and I’m with you up until the place where there’s a qualification that needs introducing to the Hayek market signal theory.
When content moves from one license to another there could be two reasons: One is because the license they’re moving to is a better license, and the other is because the license they’re moving to is less restrictive for whatever it is they want to do. And if you can remove licenses that are protective downstream, there is always the risk of a race to the bottom. So the guys who concern themselves with protectiveness will not let go of branch A until branch B is shown to be equally protective. And this is why that process of making equivalence is so complicated, because I think the likelihood of failure is part of the process by which equivalence is measured.
Now, once again, it may be that the cost of failure is not as high, and so in that usual lawyer’s calculus of the risk of erroneous action and the cost of deprivation and the cost of getting a right answer, that it may be OK to let go of more branches earlier. That’s a question to be resolved by you based on what you think the risk is implicit in your own license failure.
Suppose FDL had a terrible problem in it, and the Wikipedia started showing up next week as appropriated into all sorts of proprietary contexts. If Britannica said, as Larry Ellison seems sometimes these days to be saying: “Oh, I love this stuff, there’s no intellectual property here at all, I can just take whatever I want!”, and Britannica started issuing monthly updates consisting of stuff ripped off from Wikipedia, would you guys say: “Oh my god, license failure, please somebody do something!”, or would you say “Well, you’ve got to take the bitter with the sweet; our license failed, but it’s OK, we’ll write new articles and put them under a better license.”?
Ask how you respond to the problem presented by occasional license failure in order to decide for yourselves how much protectiveness you want. Having decided how much protectiveness you want, you’re in a position to think better about the question “Where will I trade protection for facilitation?”. That’s a social policy decision. I think Larry is correct in saying that the wiki model of making that decision is a better model than the model of votes, kingship, and all the rest. We do need to take an essentially Internet era approach to that question: proof of concept, rough consensus, running code, and we educate ourselves as we go along.
But let’s not begin by losing that sense of what protection is for. One of the reasons you go to your lawyer is to be told about all the terrible things that could happen that most of the time you don’t want to think about yourself: What if your kids are squanderers? What if your wife’s unfaithful? What if the marriage breaks up after you buy the house?
This is the uplifting part of your talk.
Nobody ever wants to think about those things, and of course, that’s the bad news that lawyers deliver. Jonathan’s right, this is the uplifting part of my talk.
Remember that proprietary culture wants to eat your lunch, remember that if they do you’ll be sorry, decide how sorry you’ll be, and how much protection you want, and then we can, in fact, begin to approach intelligently how much interoperation we can design for.
Thank you both so much for being willing to speak forthrightly and frankly about what obviously are ongoing, possibly even conflicts among friends on this front.
I want to have a change to open it up, make us read‐write. As people are lining up at the mike if they want to ask a question, let me just ask one other question, which is: Eben, you say if wikipedians end up, in the wiki way, making a decision about migrating, say, to another license – of course it’s not a clean slate here, the existing content in Wikipedia is under the Free Document License – is this a license that Wikipedia basically now, precisely because of its protections, are hands largely tied?
I think that that’s a political question, I can’t speak for Stallman about the making of licenses, and I can’t speak for Jimbo about how he would set his people free if they came to dislike the license that they’re under. But although it would be difficult to relicense, because there’s no authority in Jimbo to declare what the license is on everything that’s been contributed, I see no reason to believe that transition in the Wikipedia is impossible.
One of the things that we asked as lawyers to the Wikipedia for a study about was the pace of replacement of Wikipedia material. To try and figure out, in the natural course of attrition and replacement, how long it would take for new license terms to percolate through the license, in the long tail sense. You won’t be surprised that 80% of the Wikipedia replaces pretty rapidly and 20% does not. Relicensing is conceivable, and as a community it will happen, I feel absolutely certain.
Whatever happens with FDL, it will facilitate the migration of Wikipedia content to improved FSF free document licenses, and I certainly hope it will facilitate migration to other licenses as well, where again, Larry is quite right, “also” is very different from “instead”, and I don’t know where we are yet.
Gotcha. 15 minutes and five people, so with that in mind, shoot!
So I have a question for Eben and a question for Larry, and my first question for Eben is: Wouldn’t you – this is in regard to your comment…
Feel free to tell us who you are.
I’m Greg Maxwell, Wikipedia editor. But in regard to your commentary on considering the risks of license failure and the need for protection, you focus strongly on the idea of someone taking the content that I’ve written and using it in an unfree context, but wouldn’t you also agree that a lot of protection terms are really also about the greater social aspect of educating the society at large about free content, to create an environment where free content is able to survive, and it isn’t just about the selfish “Control my content, protect it from Britannica”?
Yes, I would agree that that’s true, but I would agree with Larry that it is unfortunate to allow the “educating people about free content” function to get in the way of the “making free content” function.
And so, at the margins, it seems to me, that would be an easier trade to make in policy terms. If, as a lawyer, I can go to you as clients and present a series of choices about that on a continuum, I think it’s an easier choice for you to make: “I’m trading off at the margins my education about free content to newbies on the one hand, with my ability to get my work done on the other.” I have confidence that you can make that choice.
The choice between protectiveness and facilitation is a more difficult choice, because it requires projection further into the future against a larger number of unforeseen contingencies.
OK. And, to Larry, we… So, the… So, consider, in a world where all free content licenses were functionally, if not grammatically, equal. I don’t think there would be any difficulty in making the licenses compatible, and I think that you would agree that compatibility in a world where the licenses are equal in all ways that everyone thinks about wouldn’t be a problem. So when we talk about license compatibility, we’re talking about compatibility with licenses that have fundamentally different approaches, not just different grammar. And so I wonder what would be the purpose for anyone to create a license with protection clauses that could be relicensed as work without protection clauses, because, obviously, why did you put the protection in it if they could just be removed by someone else.
Yeah, it’s a great question. So, Eben was pointing a little bit to this problem and the problem all gets defined by how you decide what equivalence are.
We at Creative Commons have six “core” licenses. Many people think that’s too many, I think that’s probably is too many, but we have six core licenses, and we think of those as license types. And the copyleft license of the GPL or the FDL or the by‐sa is one type, right? So it’s… By a “type” we mean it’s achieving a certain kind of functionality.
Now, the details of how it achieves that are different. Just like when we port any of our licenses from United States to Portugal to Brazil, the actual details of the licenses are different, but they’re trying to achieve the same functionality. And in that context, what we do is we say: “Content created under a Creative Commons share‐alike license produced in Poland can be relicensed in a derivative form under a Creative Commons license produced in Japan”, even though we know there are, you know, there are differences because Japan might have consumer protection laws that don’t exist in Poland and vice versa.
So, you’ve got to admit, you’ve got to accept some sort of deviation, but the critical, the only way this makes sense is if you identify what the core elements of the license types are. So, it would be a total failure if a copyleft license could be relicensed outside of the copyleft universe, that would just defeat the purpose of copyleft. But for other licenses, you don’t care, if it’s just an attribution license, like the BSD, you know, you don’t care how it’s relicensed, you can be proprietary, you can be, you can have a copyleft relicense under it, right? So, the point is, you’ve got to be careful about the types, and nobody’s more aware than I am about how hard it will be to do that properly, there’ll be a million questions that are raised.
And while I want Eben to do all the work, I’m committed to help raise the money to help him hire the people to do the work.
There you go, that’s all it takes.
But I think that’s the solution.
Yeah, hi, David Mertz, I write words that I release to the public domain for a living. I’ve a couple of questions, one is related…
Try and just keep it to one, given the timing, I’m sorry.
It’s really one.
Okay. It’s one with multiple parts.
I’ve a concern with the use of the free document license on Wikipedia because of the ugly invariant clause, and I certainly hope that, you know, Free Document Licence 2 gets, I realize…
No form of the Free Document License will retain the invariant sections provision.
I said, no form of the Free Document License will retain the invariant sections provision. That’s for sure.
That’s so wonderful.
My concern with Larry’s idea of transferability of licenses chiefly concerns the dangers we get into with transitivity. You know, we can go from FDL to CC‐sa, maybe CC‐sa contains a similar transferability clause that lets you get to somewhere else that, you know, you couldn’t quite get to right from FDL, and in fact you can get, by the same steps to a later, not yet written, version of a license that, you know, is, of course, bad, and closes things, you know, closes things back off again. And if you transfer licenses enough times, you know, there’s… you might get somewhere you don’t want to be. And, I mean, I’m… Both of you, maybe you could speak to if there’s any way to guard against these kind of dangers, in principle.
It is a huge danger, right. And, you know, why accept the danger, why try to do something that’s dangerous? Just because the consequences of not trying to do it are even more dangerous.
So, there are two kinds of solutions here. Interestingly, Microsoft has adopted one of these two solutions. Microsoft’s equivalent, you know, attempt at a copyleft license and their other, quote, “open source” licenses, basically says “You can relicense under this or an equivalent license.”. That basically allows anybody to decide what the equivalent license is, and there’ll be courts that argue about what the equivalent license is, and so there would be a struggle about that.
The alternative is to imagine a trusted institution – this is a little bit counter to the kind of Hayekian view that I’ve just been espousing, but – a trusted institution that begins to create lists of equivalence. So the trusted institution would say: “FDL is equivalent to the by‐sa. And it’s also equivalent to the Free Art license.” And each of those licenses say: “If content is relicensed, if a derivative work is made, it can be relicensed under any, quote, ‘equivalent’ – where ‘equivalent’ means ‘a license deemed to be equivalent by this body’”.
So that’s why I say it shouldn’t be a CC body trying to create an intermediate body to do it. The danger of that is that, you know, the body might be captured, bla bla bla…
But, you know, the fact is, again, it’s plumbing. It’s plumbing. Once we get the values clear, it’s plumbing. So, I don’t really think there’s a lot of intrigue to sort of figure out how to control the plumbing market. I might be wrong, but… That’s why I think that you can be clearer by listing equivalent licenses and facilitate… Sorry?
What about the ones that don’t actually exist yet, what about FDL version 3?
Sure. But that too would have to pass the test of equivalence. So it’s anything in the future that could be deemed to be equivalent in that way.
All right, so that’s a strategy, OK, with benefits and drawbacks, and you can evolve a couple more. Let me just present a couple of things that we’ll get to chance to think about collectively. Here’s another way you can go:
If you look at the current draft of GPL3, you will see that in the enhanced compatibility section we put a catch‐all, we said “Anything may be adopted into GPL3 which is a license term taken from another license that does not permit anything we forbid, and doesn’t forbid anything that is permitted by this license. So regardless of the words, you can adopt in to this license any term which is not incompatible with, or repugnant to, the existing set of terms.” Imagine, then, a structure where licenses merge over time, as they pull in the provisions which the market says are attractive.
Now take one more generalization out of that which will be familiar to wikipedians: Suppose we created a “stub” license for free documents, essentially void of terms, except the ones that we consider minimally necessary to the maintenance of the free document or the wiki or whatever. And then over time that stub comes to include the terms which people have imported into it, on the rules for importing terms, which would be a little different from the rules about removing terms. If you look at the way GPL is trying to accomplish that, you’ll see one possible approach.
Now let me just call attention to one other aspect of strategy which may seem to be farfetched from here but which has an effect. Note the difference between two strategies of internationalization. Creative Commons internationalizes by using an abstraction layer. Right? The deed somehow covers for the fact that the licenses in legal code are in fact discrepant, and as Larry just pointed out, it’s a rule that you can move content from the Polish license to the Japanese license even though the code of those licenses is not, in fact, compatible.
GPL’s approach to internationalization is the other one: Make the only layer the abstractive layer, and try and use the same one everywhere. Note that the very task we’re trying to perform with respect to content licenses is equivalent to that problem of the globalization of any license, whether it covers executable code or it covers free culture. We’re going to see, in the next few years, a series of licenses that do a job like the one Larry is talking about, for code only.
For example, the evolving European Union Public License for code, which contains a whole lot of language necessary to be used by the European Commission, and contains a principle that says “Derivatives of works under this license may be relicensed under any license on exhibit A.”, where exhibit A then includes GPL and some other copyleft licenses, maybe. The result is to create a sort of one‐stage lifter; you get through the local legal environment, and then you get to some layer of compatibility up above when you have lifted the first version through the local legal requirements. GPL will accommodate that, other people’s licenses will learn to accommodate that, that’s a little bit like the task we’re trying to perform here.
I think Larry’s correct; these are questions of drafting strategy. They’re intricate, they’re complicated, it’s an interactive work between lawyers and clients, because you keep needing to go back to the client and say: “Now, if we do it this way, this is what will happen, this is what won’t happen, this is where the risks are, this is where the benefits are.”, and clients have to make choices, which means communities have to make choices. Negotiating how communities make choices about licenses is a complicated project. I’ve learned a little bit about how little I know about that this year; it’s very tough work.
But I’m with Larry for the proposition that it’s just work. Setting the policy, that’s the hard part. After that, after some iterations, we’ll get the licensing done.
Given the time and who we have, may I suggest that we batch all four questions; I’ll take notes as you ask them, and then let these guys figure out how they want to take on that cluster.
OK. Walter Bender from One Laptop per Child. It’s a question… I just wanted Eben to elaborate a little bit; he raised a point about the difficult problem, not the stuff you guys’ve been talking about now, but dealing with governments. And in particular, dealing with governments when there are large commercial interest lobbying them to perhaps do the wrong thing. So I don’t know if you’ve got any strategies for…
I haven’t been asked to elaborate by a client in so long I forgot what it felt like. Alright.
Terry Bollinger as the author, a few years back, of the MITRE report on DoD use ofFree and Open Source software.
It’s more a comment. The most success… The biggest success in interoperability, as Larry Lessig pointed out is an important goal here, is XML. XML does not take the strategy of defining a single fixed strategy, it captures the key fundamental ideas, puts them into one package and lets people build on top, then, whatever fashion they do.
Have you guys considered, instead of coming out version of version after license – which is a very proprietary approach if you think about it, it’s exactly what we did before XML – give a toolkit in which you capture those fundamental thoughts you want, make sure they’re absolutely airtight, can’t get around them, and then people can compose to their local needs on top of that? We… If we keep doing versioning, it’s never going to end, derivatives will go on forever. We need a toolkit, you guys could do a toolkit. I mean, I’m just stating the principles; you need to capture the principles.
I’m getting into a bit of a… OK, that’s my, quote, “question”.
My name’s Sam Cooper, I’m a Wikipedia editor and I was going to propose something very similar to what Terry just said, but also address the issue of compound works, where you have two works with different licenses… If you had a modular approach, you could, for example just have a rule where the resulting work takes the restrictions from the previous works and one is more liberal in one area and one is less liberal, overall the resulting work is less liberal according to the limiting case. Do you think that’s a sensible idea?
I’m Kelly Martin, Wikipedia administrator, and I was… Both of you mentioned, during your comments, of the difficulty when you have a mixture of free and nonfree works. This is very common, at least in the English Wikipedia as well as in others – the ones that do not forbid what we call “fair use” images – and if there’s anything you can expand in that area as to how we cope with that enough in the final… [Unintelligible] in the printed version, where we might mix free and nonfree. If that’s even… If there’s a possible way we can do that or if there’s any way that we have to… If we simply have to go without using such unlicensed media.
The problem of how to compose licenses in the mathematical sense, to overlap them and make a composite of them, has one answer, which is the answer Terry proposes, which is a reduction to a common language with primitives that are well defined and are used to reexpress every possible combination of license terms.
Then there are some sloppier legal means of the kind I was talking about a moment ago, which depend more upon lawyer’s logic, in the GPL sense, and then there’s an administrative way of doing the composing, which is the one that Larry proposed, which is: It’s an expert system, and it requires in the beginning carbon‐based intelligence to do the expertise, and maybe moves to silicon over time, but still basically, it’s judgment, and it needs judging, not merely mathematics to compose.
I think that there is a real possibility for mechanical composition of licenses. That is, take the sum of the restrictions and the sum of the permissions, and work them out in a consistent way. I think that’s possible to do where there is broad general consensus, and where you are not worried about harm done by defectors internally.
If you’re worried about defectors internally, that is, people saying “No, no, you’re infringing my copyright because I never gave permission for that composite of license terms to apply to my work”, then you have to worry, because that system is vulnerable to internal dissent. How you estimate the importance of internal dissent goes back to a question I was asking this morning: “Do internal dissenters have moral rights in their work? Do they have a veto based on integrity concepts, or do they have only a property right which they have waived or consented to or in some other way traded off when they entered into – knowingly entered into – a cooperative activity?” That’s a hard problem because it is globally inhomogeneous.
And so in making a global license that solves that problem you have to pick something that works in France and in the United States, and in Germany, and in China. It’s really hard. I don’t know, Terry, if we’re going to get there by pure tech, in the sense of having an XML for license expression. Lots of people have thought about it, and some really smart work’s been done,
[Unintelligible] Scandinavian [Unintelligible]
Watch GPL3 as I do. I think that’s right. I think there is going to be some movement towards modularization in that form. It will solve lots of problems. It will solve a problem I heard Linus complaining about last week, for example.
We will get, I think, much more mixing and matching of licenses even within copyleft licenses, but the overall design still matters, and I think it still matters whether it’s a program or a sculpture, and I think we’re going to have to be attentive to that.
So when we were thinking about how to architect Creative Commons, this was the core problem that we had to struggle with, and our solution was to say: “We should recognize there are basically three different audiences that we’re speaking to.” One audience is the people who are not lawyers, and so that’s why we have the commons deed that tries to express the freedoms associated with the content.
But the second audience is lawyers and potential judges. And those lawyers and potential judges in different jurisdictions are not likely, anytime soon, to adopt a methodology that says “We have machines to pull together things and they function in the way the people want them to function.”, because, again, it’s humans ultimately, interpreting these licenses, not machines such as computers.
And then at the third layer, you know, we wanted to find a way to speak in a modular way to computers, so the RDF technology which we embedded facilitates exactly this kind of intelligence, modular intelligence, but it’s just indexes, pointing back to certain types of licenses. So you can, in principle, develop technologies that look at the RDF and, for example, say “Can these two licensed content objects be mixed together?”, and the system figures that out by thinking about the logic of the modules, not by reading the legal code; or build search engines that begin to filter on the basis of this.
But at least in the current state of legal development, we thought we had to speak three languages at once in order to deal with the three different audiences.
Right, now, look, that engineering – I’m sorry, Jon, just one more moment to turn the crank one more time – that point of Larry’s about the architecture is presently being understood in the software world, it’s basically the black duck theory, right? Give us some tools that we can use to answer a question like this: “We want to achieve a certain result, here’s some code we think might do the job, munge the licenses, look at the provenance, consult the block comments, could we distribute this, yes or no, and if not, why not?”
In other words, people are learning to try to navigate that with respect to Open Source and Free Software licenses in code in automated ways, and we’ll see more of that. It’s conceivable that you could imagine, right, asking creators “Just write down in plain language, in your native language, what you want to do with this work. We will attach that to your work in such a way that will generate the appropriate licenses on the fly for whatever it is that the work is contextually required to do, including to combine itself with other works with different licenses or different languages of intention by authors.” That would represent the full mechanization of copyright law.
So, I’m sorry. I’m sorry.
Let me end this panel the way it began – with a question to each of you, answer to exceed no more than 30 seconds.
You both had a call to arms to this audience. Both of you said: “We’re counting on you!” to the people in this room. And I want to help the people in this room understand the mission with which they have been charged.
What is each of you asking the people in this room, if they care about Wikipedia and free culture, to do?
One: Get involved in the license process for the Document Licenses when public drafts are announced for discussion.
Two: Take at least a quick look at the GPL3 discussion approach and ask yourself: “How will the Wikimedia structure do better at discussing the license when the time comes for public license discussion?” That’s to say, there’s only one model in being for hundreds of thousands of people to discuss a license, it’s jerry‐built crap, we made it up in order to get GPL3 done. You can do better, help us plan it.
And then, three: Figure out how to get involved in local politics. Not in Senator Ted Stevens’ truck‐and‐tubes problem, but how to make the city council care about free educational materials in the public schools, how to make the board of education care about municipal Wi‐Fi. In other words, how to stimulate organs of local government to see past the ends of their own noses about the broad issues of information freedom and access. Those are the people we’re going to need ten years from now, and we’re not going to be able to educate them ten years from now.
All of that, plus just one. Which is to hold out as long as you can for the principles you believe in, and not to compromise them. Because, there’s a lot of pressure to compromise on the ideals that will build the infrastructure that will enable the free culture movement to take off, and you can afford to hold out until the right answer is selected.
Thank you both so much.