Linucon 2005: Pros and cons of the GNU General Public License

Synopsis from Linucon program book: “The most popular open source license, the GPL, inspires controversy to this day. Eric Raymond recently expressed some ambivalence about it, so he and his lawyer wife Cathy are moderating this panel, with Jay Maynard, a.k.a. The Tron Guy speaking out against the GPL and Rob Landley defending it.”

One of Rob Landley’s pro-GPL arguments is that it can prevent a project from forking. Jay Maynard claims credit for coining the term General Public Virus. His objection to GPL lies in its ideological agenda. Rob says GPL keeps companies from taking open source code, incorporating it into their products and making money off of someone else’s work without giving back to the community. Jay objects that even if companies did that, the good consequences of this action would outweigh the bad. Eric Raymond then inserts himself physically and ideologically between these two “nutcase friends” of his. His position is that GPL is slowing down the adoption of open source, because it is often incorrectly perceived that a company that uses open source software would be obligated to blow open their entire intellectual property. Furthermore, he says, GPL is based on the assumption that defecting from the open source community is attractive, whereas in reality it is its own punishment. Both sides use Linksys as an example to support their arguments. 🙂 They briefly debate whether the reason BSD did not become as popular as Linux was due to its license, or, as Eric Raymond argues, because they got their social machinery wrong.

Note: All the factual errors in this post — and there may be a few — are mine, and not the panelists’. I may have misunderstood something they’ve said. (Parts of conversation where people were talking over one another, or digressed too much, were paraphrased and put in parentheses, or condensed to convey the gist, or simply cut out and marked by (…). Names I didn’t catch were paraphrased as “person X” or So-and-so. While this may deprive potential readers of juicy bits of gossip, it may work in favor of the panelists, as they won’t go on record as having said certain things about those people.

However, they said Stallman looks more sane with distance!

Rob Landley. Introductions, starting with Cathy!

Cathy Raymond. Introductions. Hi, I’m Cathy Raymond. I’m an attorney who once thought that the open source movement might be a good excuse for me to move my law practice into licensing. That didn’t work out but it still provided me with some interesting experiences. If you ask my husband, and please, don’t, he will probably tell you that I am partly to blame for the GPL, because I reviewed and critiqued an earlier draft of it. I don’t remember much about it, but I’ll take his word for it.

I’m also married to this guy. This is Eric S. Raymond of open source fame, who can probably introduce himself much better than I could.

Eric Raymond. Hi, I’m Eric Raymond, I run around making trouble. One of the pieces of trouble that I have helped create is the open source initiative, which is a community certification authority for licenses. Big sites like (something-something) and SourceForge require an LST-conformant license (?) before they will allow new projects on their site. We have spent a lot of time thinking about licensing, we are now engaged in a project to actually crack the number of open source licenses in use, because there are too many of them, and collisions on the edges are causing problems. I recently caused a stir by arguing that we need to reexamine the question of whether the GPL is useful or not, but we’ll talk about it when we get into the panel itself.

Rob Landley. I’m Rob Landley. I’m the pro side of the GPL here. I’m for it. I’m here ’cause we needed someone to give a token conventional view. I am a programmer, I’ve done a bunch of other things. I got familiar with intellectual property issues years ago. I wrote a week worth (?) of columns for the Motley Fool on intellectual property, for example. And I’ve been interested in licensing the GPL. I was interested in the GPL and in LGPL before I was actually interested in Linux, because I was following it on gcc, and glibc for OS/2. So, I’ve been following it for years and years.

Jay Maynard. And I’m Jay Maynard. One of my lesser-known projects, I talked about it yesterday morning, is called Hercules. It’s an emulator for IBM mainframes. And it’s licensed under the least PPL-ish (did he mean GPL-ish?) of the open source licenses: the QPL. That’s my most recent involvement in this respect. But I’ve been arguing against the GPL and its philosophy for 15 years now. In a post on Usenet in 1989 I coined the term “General Public Virus”.

Eric Raymond. Did you? I didn’t know that!

Jay Maynard. Yes. That was my invention.

Cathy Raymond (apparently addresses Eric): Go update the jargon file.

Eric Raymond. I have to put a comment in the jargon file.

Jay Maynard. And catch me later, and I will point you to the Google News copy of that posting. So I have long believed that GPL is evil incarnate, or next best thing to.

Cathy. So, we have the good, the bad and the ugly, or however you wanna characterize them. Let’s have Rob start out with the pro-GPL arguments.

Rob Landley. OK. The GPL is cool. From a purely pragmatic perspective, Richard Stallman annoys the hell out of me. But! The GPL is really cool. To a programmer, having your work co-opted by someone who won’t give you the modifications back really deeply sucks. Having your project fragment because people with not social skills can’t distinguish between technical arguments and when people don’t actually like them, or can’t forward their technical arguments properly and just don’t wanna go through the whole rigmarole, and decide to fork off a copy, which has some improvements, and which also has a bunch of things that don’t go back in the mainstream, and after they… They release a few binary versions, and they don’t release the source code for a long time. That’s really annoying.

Projects forking is really annoying. The GPL cannot prevent projects from forking, but it can seriously discourage it, i you cannot release even a beta version without a source code. One of the reasons there are so many BSD forks is that they have about as much friction as Debian has, and people occasionally go completely nuts like (person X) and wander off into a cave somewhere, and…

Eric Raymond. When (person X) goes nuts, how can you tell? (Laughter)

Rob Landley. It’s a matter of momentum, really. There’s (person X) being nuts and there’s (person X) going nuts. When he’s going nuts, he’s moving. And… I’ve never met the man. I don’t know, maybe he’s really nice in person.

(They exchange a few more opinions about X and another person, Y, comparing their levels of craziness. I didn’t catch the names of those people. Both of them are apparently open source developers. Eventually Eric Raymond declares this to be a sidetrack. — E.)

Eric Raymond. This is a sidetrack.

Rob Landley. This is a sidetrack, but what I wanted to point out is, (person Y) works on the Linux kernel, and the Linux kernel has not forked. (Person Y) has no intention of forking the Linux kernel. It wouldn’t come up. (Person X) had the very clear option of just going off and forking. He never needed to… he could keep it to himself for two years until he decided that other people were ready to share in the completed glory of his master vision.

Rather a lot of forks start when somebody decides “I need to go off into a hermitage for a couple of years and nobody will share in the completed glory of my master vision. No one will see my source code until I am ready for them to see it.”

And the longer a fork stays separate, the harder it is to integrate. The classic one of these, that is the reason the GPL evolved, was the emacs / xemacs split, which Eric knows way more about than I do.

Eric Raymond. I tried to prevent it. Unsuccessfully.

Rob Landley. And basically what happened is both emacs and xemacs are open-sourced now, but during their development they weren’t.

Eric Raymond. That’s not true. emacs was always open-sourced in the modern sense.

Somebody else. But xemacs wasn’t.

Rob Landley. The fork. xemacs was the fork. Might have been Lucent emacs or something like that?

Somebody else. Lucid.

Eric Raymond. Lucid, that’s it.

Rob Landley. Yeah. But it was a fork for a while, that was basically open-sourced upon its death as a closed-source project, which is fairly common. A lot of things get open-sourced rather than being abandoned. And we have had so much independent development by that time so that even though it’s open-sourced now, integrating it into emacs just can’t be done. They’ve tried. It hurts. Their design has just diverged too much, that their central design philosophy has just skewed way too much, even though back in the mists of time I’m under the impression they forked off the same version.

Eric Raymond. The GPL wouldn’t have done anything about this, though.

Rob Landley. The GPL would have prevent Lucid emacs from being an independent development project for many years without releasing source code, so that they could have started the integration work of porting over features much earlier.

Eric Raymond. The fork didn’t happen until after Lucid (…). GPL is orthogonal to the causes of this fork. It wasn’t the cause, and it wouldn’t have been the cure.

Rob Landley. I’ll have to look at it.

Eric Raymond. I was there.

Rob Landley. I’ll have to look into that. I read Jamie (…)-inski’s account, I’ve read a lot of this. But when this happened I was… what? Fourteen?

Eric Raymond. I think I was the last person present at the conversation that both Jamie (…)-inski and Richard Stallman were in.

Cathy Raymond. Was there blood on the floor afterward?

Eric Raymond. No, but it took a lot of work on my part to prevent it. And nobody did successfully afterwards.

Rob Landley. I know from reading the history of the GPL that the emacs license happened because Stallman wanted something to prevent what had been going on at the time.

Eric Raymond. The problem is that GPL 1.0 predates the emacs / xemacs fork.

Rob Landley. All I really remember about GPL 1.0 is that it was buggy.

Eric Raymond. Yeah, it was.

Cathy Raymond. So it was.

Eric Raymond. It was, but I’ve got a copy, and you can see the preoccupations and the basics of the logic that went into 2.0. Remember, Cathy and I reviewed 2.0. We were there when that transition was made.

Rob Landley. I think I still have your little yellow book that has the early emacs license in it.

Cathy Raymond. But in any event, it gives us at least a snapshot of the pro position. Let’s give Jay a few minutes on the cons.

Jay Maynard. Fundamentally, the biggest problem with the GPL is that it is a license designed to advance one political philosophy. At its root it’s not about programming, it’s not about sharing per se, it’s not about all things people like to think it’s about. It’s political. And to understand it, you need to read the GNU Manifesto. Hold your nose and read it. It reads very Marxian. It could be boiled best down to “from each programmer, according to his abilities, to each user, according to his needs”. It advances the proposition that while programmers should be able to make money for their services, the fruits of their labor themselves are not what money should be made from.

And the GPL was written and designed explicitly to advance that philosophy. The idea being that the GNU project would create so much good software under this license, that people would be drawn to the software and would be drawn to the license to take advantage of it, thereby advancing Stallman’s political vision, the end goal of which is nothing less than the destruction of the software industry as we know it today.

Eric Raymond. You say that like it’s a bad thing!

Jay Maynard. I’m certainly not going to defend Microsoft. I’m not sure they are defensible. I’m not gonna say I’m here to defend SCO. However, there are a lot of other software companies out there that do well and do good. And those too would go straight down the same toilet.

Rob Landley. Can I interject a point? When I was learning about business while writing about it — which is a really weird way to go about it — it does mean that an awful lot of people point out to you when you’re wrong. I had a column read by… I’m told that it’s 15 million people through the Yahoo syndication, and I had no idea what I was doing. And I was very upfront about this. Apparently they just thought I asked good stupid questions.

But one thing I did learn was: commoditization. Mature markets commoditize.

Eric Raymond. What does this have to do with the GPL, though?

Rob Landley. What does this have to do with it is: open source is commoditization of software. And with the GPL is the forced by license commoditization. This product cannot be made proprietary again without violating the license terms.

Jay Maynard. Unfortunately though, the argument there, you’re trying to make fungible something that is not. You cannot make operating systems fungible, fundamentally. And I’m not talking about one Linux versus another. I’m talking about…

Rob Landley. What does fungible mean?

Eric Raymond. Interchangeable.

Jay Maynard. Take bushels of corn in a silo. So you can’t replace Linux with OS/2, for example. Fundamentally that does not work. So the commoditization argument falls because it’s trying to attack the wrong problem.

Rob Landley. Well, no, what they did is they moved commoditization to a slightly different level. If the implementation is complex enough that you have to have a common implementation, then you have to make implementation a commodity.

Eric Raymond. This doesn’t take us anywhere on the GPL versus non-GPL.

Jay Maynard. And in fact it raises another issue. And that is, when GPL is applied to things that are standards, it hinders the adoption of those standards.

Eric Raymond. Ah! Which is a perfect segue to the position I want to state.

Jay Maynard. Let me finish a statement and (…) seamless transition.

The canonical example of this that GPL advocates come up with is the BSD TCP/IP stack. And they argue that the TCP/IP stack was hijacked by Microsoft and taken private and Microsoft didn’t contribute anything back to it, and this is a bad thing. I argue that this is a very good thing. Because when Microsoft did it — there is reasonably (…) it’s no longer there. But when they did it, they did it mainly to get rid of their own buggy code. GPL licensing, the TCP/IP stack would not have resulted in the infection of the single line of Microsoft code with the GPL. Microsoft would have simply ignored it. And we would today be condemned to working around Microsoft bugs all over the internet.

Eric Raymond. Like we’re not.

Rob Landley. As opposed to?

Jay Maynard. But not at the TCP/IP layer.

Rob Landley says something about winsocket

Pros and cons of the GNU General Public License panel at Linucon 2005
CIMG0928 Pros and cons of the GNU General Public License panel, left to right: Cathy Raymond, Eric Raymond, Rob Landley and Jay Maynard

Eric Raymond. This is the point at which I insert myself between those two zealots.

Rob Landley. Right! Let us swap the chairs.

Cathy Raymond. Right. Go insert, Eric, and in fact I was going to ask for your position.

They swap chairs so that Eric Raymond is now sitting between Rob Landley and Jay Maynard.

Eric Raymond. I sit somewhere physically and ideologically between these two nutcase friends of mine.

Jay Maynard. Which means, both of us attack him! I’ve known him 15 years, I don’t know how long Rob’s (known him).

Rob Landley. Not that long. Seven? Six?

Cathy Raymond. Stop stroking your grey beard (…)

Eric Raymond. I question the utility of the GPL, but not for ideological reasons. Jay is right: it was written to advance a political agenda. I don’t care. And I don’t particularly think anyone else should care.

Rob Landley. Yeah, I don’t care.

Eric Raymond. I don’t care (that GPL is quasi-Marxist). What I care about is the effects of the GPL. And what I think I’m seeing increasingly is that the GPL is a hindrance rather than a help. And the reason is that it is slowing down the adoption of open source. There are too many corporations out there that have no open source policies because they’re afraid that they’ll get infected with the copyleft license and have to suddenly blow open all of their business knowledge, all of their intellectual property, everything that they think is fundamental to their business model.

Now, hackers can say: “well, you’re wrong about that. It’s not as fundamental to your business model as you think.” I make that argument all the time. Sometimes I even succeed with it.

Rob Landley. We also say it doesn’t get infected.

Eric Raymond. I also make that argument too, and sometimes I succeed. It doesn’t matter. The point is that the perception is out there, it’s not going to go away, it’s easily exploited by our enemies, and I think GPL has become a net (?) drag.

Furthermore, I don’t think it’s necessary, because the GPL was erected on the assumption that open source cooperation is so fragile, such a sacrifice, such a difficult, painful, negative thing to do, that you have to protect it with teeth or it will fall apart.

Rob Landley. I don’t see that.

Eric Raymond. You don’t punish behavior unless you think that behavior is attractive.

Jay Maynard. And in fact people accuse me when I raise this argument all the time of wanting to take PPL (?) code and use it for master evil commercial products.

Eric Raymond. Furthermore, you don’t punish behavior if you believe that that behavior is its own punishment. So the GPL was erected on the assumption that defecting from the open source community is attractive and is not its own punishment. And what I’ve come to understand is that defecting from the open source community is in fact its own punishment. Here’s why.

When you do a proprietary fork of an open source project, here’s what happens. Now you have a small group of programmers who, because of the nature of non-disclosure agreements and proprietary lockdown, can’t get help from anybody else, and they are competing against the large project that they defected from. You can’t win that game. And in fact, intelligent IT managers these days don’t even try. They go the opposite direction.

Rob Landley. Yes, but there are a lot of unintelligent ones. There will never be a shortage of unintelligent IT managers.

Eric Raymond. Aha! But the market will punish that decision.

Rob Landley. And the next generation will make it all over again.

Eric Raymond. And the market will punish that decision, and it will keep on doing that until the business world learns better. We don’t need to punish defection, it punishes itself.

Rob Landley. Actually, you’re making an assumption. I do a lot of stuff in the embedded space. I’m one of the maintainers of the Busy Box project. I’m currently in charge of the stable fork. I put out 1.01, I’m gonna be putting out 1.02 after Linucon. I’m trying to get 1.1 to a release mode. I’m not the project maintainer, but the project maintainer doesn’t make releases, he just like putting more code in CVS forever. (The panelists cackle.) And somebody has to actually send him cakes. But that’s a long story. I sent him a cake for the one year anniversary of the uclibc release, which prompted the release of

In the embedded space there are all these products that exist for 2 years. The Linksys router and all the clones of the Linksys router. The only reason we got the Linksys code open, which was Linux-based, Busy Box-based, and had uclibc in it, I think, and a bunch of embedded stuff — the only reason we got this opened is that the people writing the embedded stuff, who had these things and couldn’t use them, basically did a polite GPL enforcement thing, saying: we’re the copyright holders of some of the code that you have, and we would rather like to see the source of your modified version as in the license terms.

Because of this, the code of the Linksys router — yeah, there are some proprietary kernel modules and stuff — but the source of the Linksys router is out there, and because of this, there are now companies… There is that guy who has a complete replacement for the Linksys firmware…

Eric Raymond. Sveasoft.

Rob Landley. Yeah, him. Which it might have happened without that, but it would have happened 5 years later, after we’ve picked apart and reverse-engineered it.

Eric Raymond. You picked a bad example. There were already open-sourced firmware loads (?) for the Linksys before they threw the code open.

Rob Landley. And, they couldn’t use the wireless hardware.

Jay Maynard. And, what’s happened now is Linksys has changed the hardware and gone to VxWorks-based software load on the thing and it’s closed up again! So the GPL hasn’t done a damn thing for the Linksys users if you weren’t lucky enough to get one of the older versions!

Eric Raymond. And this is a perfect example of the GPL retarding acceptance.

Rob Landley. Well, I would also like to point out that since Cisco bought Linksys, their market share went through the toilet for completely unrelated reasons, cause the hardware started to get really really crappy and unreliable.

Eric Raymond. They fixed that. I actually bought two Linksys’es that I had to junk just after the Cisco transition because they had Sudden Infant Death Syndrome. Then I went out and bought several more functionally similar boxes, and they were so crappy that with fear and terror in my heart I bought another Linksys, and it’s OK.

Rob Landley. So after you bought 3 of them, you finally got one that wasn’t crap. And you believe that based on this experience this means they fixed their production problems due to 30% yield?

Eric Raymond. No, because it wasn’t a yield problem. It was the software actually being screwed up. The load I got on the last one seems to be OK.

Jay Maynard. Don’t look at me, I run a DI 624.

Rob Landley. No, there actually were hardware problems with at least one of them. I remember that the wireless thing went away.

Eric Raymond. Yeah, that was after it had been in service for a while. It wasn’t Cisco’s fault. I’ve been through a bunch of those boxes.

I didn’t know about the VxWorks load, but it’s a perfect example of GPL killing off a niche where we were doing OK.

Rob Landley. Oh, we are still doing fine in a bunch of the clones of that thing. The thing is, BusyBox and uClibc maintainer is Erik Andersen, and his father is a lawyer. And because of that he can basically get free legal time from his father to send out various messages to people who are using BusyBox and uClibc in the embedded space. He has the Hall of Shame, where basically he’s got lots and lots of devices that you can grab the firmware image, decrypt it sometimes, decompress it, seek into the thing and find the root partition that has BusyBox in it, or uClibc, and he basically sends out little “cease and desist” letters saying “You guys are using our stuff in violation of the license terms. We would like the source code.”

He doesn’t go after it too hard. He mostly just sends “cease and desist” letters unless somebody really pisses him off. Mostly they do release.

There’s a guy in Germany who actually got somebody’s production halted, because he has some of the network code in Linux kernel. He runs and he goes after German companies, because that’s where he lives, or companies that do business in Germany. And he’s actually getting very good precedents that GPL is enforceable. He’s gotten restraining orders against companies. You have to stop shipping until you resolve this issue.

But basically in the embedded space, it’s not like these are people… They don’t care one way or the other about shipping source code. It’s that they’ve outsourced it to cheap Indian or Taiwanese labor that had a 6-month contract, shipped them a finished tarball; they no longer have a business relationship with these people; they’ve shipped a product that they expect to manufacture for 6-8 months; it will be on the shelf for 2 years; and then no one will support it at all, because it was a 40 dollar item. It was cheap, plastic hardware and they have moved on.

We would rather like the source code to these so that we could support them ourselves. They’re not even worried about clones, because they’re not making it anymore.

Jay Maynard. And half the time it’s not the design so much as implementation of the manufacturer’s application note (?)

Rob Landley. Half the times the companies we were talking to are re-shipping a very lightly modified firmware image that they got from somewhere else. And basically we go through this company that’s in the US; they have to talk to their Taiwanese supplier that doesn’t even speak English. We would never see any of the source that we do get from these guys, if it wasn’t for the GPL. And we already had several people who… Oh, darn, Glen McGrath is an Australian developer…

Eric Raymond. Taking everything you’ve said as a given, we would never see any of the source code without the GPL. What worries me is the second order effects of GPL enforcement are worse than the gains we’re collecting.

Jay Maynard. Yes. And I think I see where you’re actually going with this position I happen to agree with, and that is that sure, these guys are winning access to source code for…

Eric Raymond. I’m afraid it’s winning us battles and losing us the war.

Jay Maynard. Yeah, they’re winning access and what will happen is manufacturers will simply do like Linksys did in the case of RT54G (?): they will put out new versions and simply not use open source software. So the overall effect will not be a gain in the open source access to firmware at all. The net effect will be that people will simply stop using the code.

Rob Landley. I actually don’t mind this. If these are people who don’t wanna play by these terms, they shouldn’t have been using it in the first place. And one of the interesting things about this is that (…)


You’re acting like forking is a temporary thing that would just go away once we have a sufficient market share, and I’m point out that there’s never gonna be a shortage of stupid people.

Cathy Raymond. I see an audience question.

A guy from the audience (Big O). Let me point something out. The reason why, these manufacturers use these library components was that it shortened their development cycle, so it has some value to them. If they wanted to not use it, they should have developed their own and not (…)

Rob Landley. You notice that Linksys didn’t go to VxWorks, Cisco went to VxWorks, which is a much, much larger company with money to license anything they darn well please, that it already integrated this into a different business unit.

Jay Maynard. You are assuming, though, that that decision was made at the Cisco corporate level and not the Linksys division one.

Rob Landley. The Linksys division had its management swapped out with Cisco management, so Linksys isn’t making decisions anymore, they’re drooling.

Jay Maynard. But the same comment applies. You’re not going to convince Cisco to produce things to which the source can be obtained. It’s not going to happen. You’re not gonna get Microsoft do it, you’re not gonna get Cisco do it, you’re not gonna get IBM to do it, outside of very few cases, and IBM is a special case.

Eric Raymond. The question is, are the minor battles we’re winning by forcing this tiny pieces of code open worth the long-term bad effects of scaring lots of corporations and technology vendors away from open source? And I’m increasingly thinking not.

Rob Landley. 99% of the time nobody actually fights these battles, because we either convince people to cleanly use the stuff and release the source code, or not use it. We only really go after the people who probably would have released the source code if they could have been bothered; or like Harald Welte, people who are like, look, you really shouldn’t be using Linux if you don’t want release the source code; we’d rather you didn’t ship, and then (?) you ship with something else.

The thing is, most of these niches, there’s dozens of players in any of these niches. I think it’s WindRiver systems, the people who bought the corpse of, and they sort of had Slackware for a while, and spun it off, and had Free BSD for a while and spun it off… Those guys were one of the big players in the embedded space. And they still sort of are. But they had to switch over to Linux recently just because it was undercutting their margins.

Eric Raymond. So instead of threatening people with lawsuits, we should simply stand back and let the market do its thing. Help commodity software win. That way we win, and we don’t slow the adoption curve by scaring people with GPL!

Jay Maynard. I need to say at this point that I am not at all philosophically opposed to the concept of open source software. What I am opposed to is the concept that it should be in any manner, in any way, by any person, under any circumstances, mandated in any form.

Eric Raymond. Because no one wants to be forced to be virtuous.

Jay Maynard. Correct.

Rob Landley. And that seems to be the common objection. I’d like to point out that both of you are fairly radical libertarians, correct?

Jay Maynard. He would argue that…

Eric Raymond. No, I’m a radical libertarian. He (Jay Maynard) is a conservative.

Jay Maynard. I consider myself a conservative with libertarian leanings.

Rob Landley. I’d like to point out that Free BSD has been out for, like, six months less than Linux. And it has had basically every opportunity that Linux has.

Jay Maynard. No, it hasn’t.

Eric Raymond. Their failure was due to two main causes that have nothing to do with the GPL/BSD distinction. One was they lost a critical window of opportunity in the AT&T vs BSD lawsuit. If it haven’t been for that, the BSD people are well aware, and in fact Linus is well aware — I had this conversation with him — that BSD would probably rule the world now. I’ve been on stage with Linus when he admitted this publicly.

Rob Landley. Yeah, I’ve heard he said that. He said that he would have worked on BSD had it been out at the time.

Eric Raymond. That was problem number one. Problem number two is: the BSD people got their social organization wrong.

Rob Landley. Still is.

Eric Raymond. Yea, yea, yea, yea! The fundamental problem with BSD is there is a whole unitary distribution. You can’t change any piece of policy without forking the entire project model. They’re very proud of their single make system: you type “make” at the top level and it builds everything: kernel, utilities, the whole nine yards. They’re very proud of that. The problem is, it introduces fatal social rigidities. It means that any time you disagree with even the smallest little piece of the distribution decision, your only option is to do (as the person X did). That’s the problem with BSD.

Jay Maynard. And in fact, consider a Gentoo BSD, which would provide the same thing, except not mandate the distribution decision. You think that would have made a significant difference?

Eric Raymond. Oh, yeah! The problem is, the people — and I say this as a person who comes form that world myself — the people who founded the BSD project were excessively old school. They were still thinking in terms of the traditional software engineering, and toward (?) the critical points of organization, they thought you had to have one. And that’s what screwed them up. It wasn’t GPL versus BSD. They got their social machinery wrong.

Rob Landley. So why hasn’t it recovered if the GPL is dragging Linux so badly?

Cathy Raymond. Cause they still don’t have their social machinery right?

Eric Raymond. Heh heh heh. Well, for one thing, the relative size of their developer communities. Who wants to go play in an environment as authoritarian as, say, Free BSD?

Rob Landley. One interesting thing about the GPL is that it appeals to developers because GPL is designed to protect the interests of developers. So people who write open source code have a huge incentive…

Cathy Raymond. Slow down. Let’s slow down. I see a couple of questions that I’d like to get to. The gentleman way there in the back?

The gentleman in the back says something about how MacOS uses BSD and is now competing with BSD. That causes Rob Landley to segue into this:

Rob Landley. There would be a Microsoft version of Linux today, if there was a BSD license, or at least they would have incorporated huge quantities of code, and they would have embraced and extended it so that it wouldn’t…

Eric Raymond. If that were happening, we would be winning, because they’d have to use our networking stacks, our file protocols…

Rob Landley. And we’d have to buy it from them if we wanted to (…) We still wouldn’t have 3-D drivers.

Eric Raymond. No, no, no! We’d still own the basic projects they were taking code from. If Microsoft tried to embrace and extend us, this would be a good thing.

Jay Maynard apparently addresses the guy in the audience who said Apple was competing with BSD. And I have to argue with one fundamental assertion that you made in it. Apple did not steal Free BSD! Get that thought out of your mind right now!

Somebody (maybe the same guy who mentioned Apple). I’m sorry. You’re right.

Rob Landley. They hired several developers.


Jay Maynard. One of the fundamental philosophies of BSDs plural is that people like Apple should be able to take that code and freely use it and freely build on it, and not be beholden to do a damn thing because of it.

Cathy Raymond. Why? No, seriously, why? I’m interested in this because… Well, I’ll get to my own opinion (in a minute).

Jay Maynard. It comes down to the fundamental definition of freedom. And true freedom must absolutely, necessarily include the freedom to do something that pisses other people off, as long as you don’t actually harm them.

Cathy Raymond. Right, like promulgating your software under the GPL.

Rob Landley. Nobody’s forcing anybody to use the GPL.

Jay Maynard. Hang on, hang on. The fundamental difference here is that the BSD types realize that allowing somebody like Apple to take their software and release the modifications without source, or release add-ons without source, etc. is going to piss off people. But there is no actual harm there. And so in order to maximize freedom for everybody you must grant them that freedom. If, as the GPL advocates claim, you must restrict freedom in order to maximize freedom…

Eric Raymond. Ignorance is slavery! War is peace!

Jay Maynard. … that is like the canonical example, fucking for virginity.

Rob Landley. I’d like to point out the that Bill of Rights is a list of things, of restrictions that people should not be doing (…)

Eric Raymond. No, that the government must not do. (…)

Rob Landley. So, laws against killing people: we’d be better off without them.

Eric Raymond. You stepped over a line. There’s actual harm involved.

Jay Maynard. So, Cathy, speak up. You…

Cathy Raymond. I was first going to find out what Mr. Big O wanted to say, if he still remembers it.

Big O. I don’t.

Cathy Raymond. OK. So let me start with my own recap of what I’m hearing here.

Rob is a programmer and he sees the GPL as important, because it’s a way for a programmer…

Rob Landley. Useful.

Cathy Raymond. Useful. Whatever. It’s a way for a programmer, as it were, to protect his ability to develop his code. Jay is looking at… Jay recognizes that there’s also a political agenda which can be used in the GPL. And he finds it offensive because he thinks that it interferes with commerce, with the ability to freely trade…

Jay Maynard. I’m a naked capitalist.

Eric Raymond. If you’re gonna talk the freedom talk, walk the freedom walk. He argues that FSF isn’t doing that.

Cathy Raymond. Eric is a visionary. Eric has this dream for open source, and he has looked at the GPL, and he thinks that at this point GPL is doing more harm than good to the vision.

I’m a lawyer. To me, a license, any kind of license is a tool. The purpose of a tool is to help somebody to get something done. When a tool is a license, the purpose is to effectuate the goals of whoever owns the thing that’s being licensed. You write a program, you own it. You have a right to set terms under which it can be bought, sold, distributed, whatever. That’s part of what a license is supposed to do for you. If you want to use your distribution of your code as means to push a political agenda, you can do that. It’s a free country. You will piss people off, but you can do that.

What I’m hearing here is not so much that GPL is ineffective, but that the GPL is, in a sense, too effective, notwithstanding that there isn’t a court in America that has yet said it’s an enforceable license. What I’m hearing is people saying that they disagree with the different agendas that are being discussed. And that’s very interesting to me.

I think that whether those agendas are good, bad or evil, may be a slightly different panel, but we can certainly go into that in the remaining five minutes.

Rob Landley. Can I clarify one thing?

Cathy Raymond. Please.

Rob Landley. I’d like to clarify one thing. I really don’t care about the political agenda of the Free Software Foundation. I consider them largely to be foaming loonies. They may be right in some things but that doesn’t stop them from being foaming loonies about an awful lot of it.
Most of the BSD people who object to GPL because it’s not “real” freedom, it’s Nutra-freedom (panelists emit shrieks of laughter) — I consider them to be just as zealous as people who say there should be no software that is not GPL’ed. I use BSD-licensed code, I just don’t write it. As a developer, I use the GPL because it’s the license I want on the code I write in my spare time that I’m not paid to write. I will not release BSD-licensed code without a darn good reason, unless I am paid to write it. It’s not what I do for fun.

The elephant in the room is that Linux is GPL’ed right now, and the license is not changing, because there are way too many stakeholders. The license on most of the GPL’ed software out there — you’re not gonna get it re-licensed. You’re either arguing that new software should not be GPL’ed, which, I admit… I admit that GPL 3.0 is somewhat scary, because, well, Stallman is involved in doing it, and to me he seems more sane with distance. The longer ago he said stuff, the more…

Eric Raymond. The less crazy it seems.

Rob Landley. The less crazy it seems. I don’t know if this is history proving him right, or the fact that he used to be more sane. I don’t know.

Eric Raymond. I’ve known him long enough to answer that question. He used to be more sane.

Jay Maynard. I’m going to object to one thing there, and that is, that talking about freedom makes me a zealot. What was it (So-and-so) said? Extremism in the defense of liberty is no vice? Freedom is so thoroughly important a concept, it’s so central to what we are as a society, it’s so vital to what we want to become, that polluting it by discussing freedom in a manner that FSF is doing, is actively harmful to the concept of maximizing freedom in our society.

Eric Raymond. And I will say… I will make my closing statement. I will start by saying, no, I will not concede that he is a zealot about freedom. If he were a zealot about freedom, he wouldn’t be a conservative, he’d be a libertarian. (Everybody laughs.)

But my closing statement about the GPL is the thought I wanna leave you all with, is that I think the GPL was fundamentally a confession of (…) weakness. That it was written at a time when we saw open source and free software development as a fragile, endangered phenomenon that was surrounded by predators much more powerful than us. What I want to point out is that it’s 2005, we don’t live in that world anymore.

Rob Landley. Yes, Microsoft imploded years ago, they (…) They admit they haven’t shipped anything (?) (Everybody laughs.)

Eric Raymond. We don’t live in that world anymore, and I don’t think it’s possible, I think that open source is a superior development method. We will win in the market place. And I don’t think it’s possible to both believe that and believe that the GPL is necessary.

Rob Landley. I’d like to live to see it. DCSS, binary-only Nvidia drivers, 802.11g was a huge fight, we still have problems and I don’t think even if we had a lot of market share, someone who had an end-all, be-all graphics card that was still a monopoly piece of hardware, you know, because it was just so much cooler that anything else out there, they would love to have a binary-only driver, they would love to dictate the terms to the rest of the world. That’s what monopolies do.

Eric Raymond. Yeah, well, and you know what markets eventually do to monopolies, unless there’s a government propping them up. They kill them.

Rob Landley. And it can take 60 years.

Eric Raymond. No, actually, the half-life of monopolies is about 12 years.

Cathy Raymond. When did Microsoft become a monopoly, again?

Jay Maynard. In 1995.

Rob Landley. More or less since the release of Windows 3.1. It was more like ’92.

Eric Raymond. The half-life of monopolies is about 12 years. This is very hard number, derived from studying monopoly market shares, and location on the (…) over the last 150 years.

Rob Landley. That is an interesting argument, and I’d like more information on that, because my main objection is, you know, it’s like, yay BSD, it can be stable once it’s dominant and it’s not stable when it’s not dominant.

Eric Raymond. There is one exception. If you get government to prop you up the way (so-and-so) did with the AT&T monopoly, then you get to exceed the 12-year half-life. Unless government a silent partner — 12 year half-life. It’s a rule.

Rob Landley. How long did Henry Ford last?

Cathy Raymond. Well, back then it wasn’t called a monopoly. It was…

Eric Raymond, addressing a member of the audience. Back there! We’ll finish up with this guy’s question.

A guy in the audience. (If a company can run away with commercially viable open source software, they will.) And the only thing that’s stopping them from running away with all the stuff that is Linux, and keeping them…

Eric Raymond. And running a small development group competing with the large group?

Jay Maynard. You are making a fundamental assumption that makes my blood boil. And that is… The assumption you are making is that by doing so, they will make the original code that they are starting from unavailable. And that is flat 100% wrong of any OSD licence. BSD, QPL, GPL, no matter what, you cannot take existing code private.

Rob Landley. You can just render it irrelevant.

Eric Raymond. When has that ever…

Rob Landley. You can also hire the developers away.

Eric Raymond. When has “render it irrelevant” ever been accomplished, even once? Give me even one example.

Rob Landley. I remember a number under OS/2, but (…)

Jay Maynard. If that could be done, we wouldn’t have BSD today, cause Sun would have taken it private.

Rob Landley. Yeah. Well, Sun hired away Bill Joy, and then… it didn’t stop it, it just cost them five years.

Cathy Raymond. I think we will all have to agree to disagree. It’s 3 o’clock, I am expected at another panel, if there is no one else in the room after this, you’re welcome to continue the debate.