About LWN.net

Interview with Richard M. Stallman

The GNU manifesto states "An initial kernel exists but many more features are needed to emulate Unix." What was that kernel, and what happened to it?

It was called TRIX, and was developed by someone at MIT (I don't remember who). I think that I eventually concluded it was not really usable as a starting point. One factot was that it ran only on an obscure machine, and would have required porting before we could even try to develop it further.

The GNU Project's kernel now, of course, is the Hurd; it is evidently getting close to ready for more widespread distribution. Free operating systems based on other kernels are now widely used; what will Hurd-based systems offer that will make them attractive relative to the others?

The Hurd offers the power of a microkernel-and-servers architecture. For instance, you can run two copies of the Hurd at the same time, debug the new one using the old one, even gradually switch from one version to another. You can even use GDB to debug the file system while the system is running--thread-specific breakpoints allow you to debug the file system's activity for certain files, while the same file server runs normally when GDB opens the source files of the file system.

These servers do not in general require special privileges. As an ordinary user you can write a new file system and attach it to a file name in your directory. Then anyone who accesses that file name talks to your file system. The file system can emulate the behavior of a single file, or the behavior of a directory.

One user-level benefit is that a command analogous to `su' can give root privileges to your existing jobs, including your Emacs. Another command can take root privileges away again. So you don't have to start a separate Emacs under your `su', which lacks your usual buffers etc. You can also assume several user identities at once for access purposes, adding and deleting them at will.

We hope to provide an extremely simple and convenient system for package installation and deinstallation, using specially written file systems for /bin and other such directories.

The release of a Hurd-based system will, in a way, represent the achievement of the GNU project's goals, set almost twenty years ago. What comes next? Will the project go into a maintenance mode, or are there interesting new initiatives in the works?

Our initial goal, a free Unix-like operating system, was achieved by the GNU/Linux system almost a decade ago--but just because a system is working, that doesn't mean it is finished. We have done a lot to improve GNU and GNU/Linux since then, such as developing GNOME, rewriting GNU libc to support GNU/Linux, and funding the start of Debian. More of this remains to be done: for instance, DotGNU aims to extend GNU facilities in a direction that many users are likely to want; GNU Classpath and GCJ are developing a free platform for running Java programs, while DotGNU is working to replace Microsoft.NET. The Trillium speech generation software has been donated to the FSF and is now being ported and improved under the GNU aegis. Beyond the operating system, we also hope to provide free application software for all the jobs users want. The point isn't that all these applications must be GNU programs; a good free program is sufficient for the job, no matter who developed it. Rather, our mission is to forward this development.

"Our initial goal, a free Unix-like operating system, was achieved by the GNU/Linux system almost a decade ago--but just because a system is working, that doesn't mean it is finished."

Also from the Manifesto: "Unix is not my ideal system, but it is not too bad." What is your ideal system at this point, and how do you see getting there?

I don't have grandiose ambitions in the area of technical changes. I would like to see the whole system made more coherent through use of Scheme as an extension language (supporting other popular scripting languages too by translating them into Scheme). I would like to see full and thorough support for internationalization.

At a higher level, I would like to see all the activities and customizations that non-wizards might want to use available in GUI form.

Is it your belief that "high-paying organizations" (i.e. proprietary software vendors) should be banned?

I would not ban high salaries, but I think they should have a high tax bracket. As for making software proprietary, I really don't care whether it is legal as long as in practice it is rare enough to have no significant impact on society.

Some companies have adopted a hybrid business model where proprietary software products are used to help fund free software development. In your view, is this an appropriate way to raise the money to pay for free software projects?

Proprietary software is antisocial, so developing it is wrong. In most cases, the user of proprietary software is expected to promise not to share with anyone else. It's wrong to make that agreement, wrong to keep it if you have made it, and especially wrong to lure someone else into making such a promise. Using part of the proceeds of this antisocial activity for a worthy cause cannot justify it.

We can easily identify the practical harm such companies do. When major institutions in our community develop non-free software, they tell the public that non-free software is ok. This weakens our community's resolve to maintain our freedom, and that weakness hurts our chances of surmounting each of the various obstacles that we face: hardware with secret specs, non-free tools and libraries such as Sun's Java platform, software patents, the DMCA, and the proposed SSSCA. When they make it tough to obtain free software for a certain job, will we persevere, or will we give in? Those who are willing to take the easy way out and use non-free software will not help us prevail.

Often a change in a company's public message is the first sign we get that it is thinking of jettisoning its principles on this issue. The idea that free software has an ethical importance disappears, and often the term "free software" disappears as well. It is replaced by statements that appeal only to the practical convenience their software offers.

The usual cause of this change is accepting outside investment from people who are not dedicated to free software. Many a free software company has been maimed or destroyed by a VC ambush, though I can't give you a body count. For instance, Cygnus Support was started with its founders' money, and was profitable and growing for several years as a free software company; then it accepted investment from people who did not care about free software, and they decided to start developing non-free programs. (We were lucky that Red Hat acquired Cygnus and rereleased them free software, but you can't count on being rescued that way.) Although Cygnus continued developing free software as well, it was no longer a free software company; our community had lost its biggest example of business success. If you are running a free software company, beware outside investors!

In contrast, I think it is acceptable to do what MySQL AB and TrollTech do: release under the GPL, but sell alternative licenses permitting proprietary extensions to their code. My understanding is that all the code they release is available as free software, which means they do not develop any proprietary softwre; that's why their practice is acceptable. The FSF will never do that--we believe our terms should be the same for everyone, and we want to use the GPL to give others an incentive to develop additional free software. But what they do is much better than developing proprietary software.

(Footnote: RMS added later: "I later learned that TrollTech does develop proprietary software. I apologize for having mentioned it erroneously.")

What system do you run on your desktop and/or laptop now? Who was its distributor?

I use Debian GNU/Linux.

The large commercial tax programs require, for their production and maintenance, vast armies of lawyers and accountants to ensure that all the rules are followed. They also come with a warranty: if a bug in the program leads to tax penalties, the vendor will pay those penalties. It has said that this kind of software can never be free, since the users must pay for the lawyers, accountants, and warranty. In a society based on free software, how can such programs be developed and supported?

If warranties are so important, the same people who now pay for the right to use the program (to get the warranty) might pay the same amount specifically for a warranty. The same company might thus develop the same program as free commercial software with a business model of selling a warranty for it.

Alternatively, the IRS could release this software. Or it could provide a back end which does the calculations, so that others could develop interfaces for that back end, without risk that the calculations might go wrong.

How will the FSF respond if the SSSCA becomes law in the U.S.?

We are responding already--by helping to organize grass roots groups in several cities to oppose the proposed SSSCA and the existing DMCA. (I think they chose the unpronounceable new name CBDTPA on purpose to discourage people from talking about the bill, so we need not let them saddle us with it. Why let them make the rules?) Please visit digitalspeech.org if you want to help.

As for what we will do if such a law is adopted, it depends on the details of the law actually passed, so I cannot say. In the worst case, we will depart the US, saying "We shall return," and fight for freedom in other countries.

We think the SSSCA will be defeated; the bill is extreme. That's when the real battle will begin. The SSSCA is likely to be followed by a series of slightly weaker laws that will pretend to be "balanced" by comparison. We must use this occasion to build organizations to face the next wave of attacks. Whether they call it the SSSCA, or the CBDTPA, or the SS-SA, or the @%&#!A, we have to defeat it every time.

The USA is not the only battleground: the Free Trade Agreement of the Americas (FTAA) may extend DMCA-style anti-circumvention provisions from the southern tip of Chile to the northern territories of Canada--that is what the US demands. If you live in the Americas, please work to oppose the FTAA, particularly this April when it is debated again. The immediate battle here is that the USA is pushing to accelerate adoption by 2003 when the current slate is for 2005. Let's at least try to keep the schedule for 2005 so we have more time to fight.

There is, for example, some disagreement (among the copyright holders) over whether run-time loading of modules into the kernel, Linux, requires that the modules have a GPL-compatible license. As the creator of the GPL, do you feel that Linux kernel modules fall within the boundary?

They clearly are covered by the GPL; modules for Linux are extensions of Linux, so under the GPL these modules must be free.

However, anything the copyright holders of Linux give permission for in use of Linux is certainly permitted, regardless of what the GPL by itself would say. The license used on a program is legally a statement of what the copyright holders permit. Any statements they make that they permit this or that, once others rely on them, have the same legal force.

How about MySQL AB's claim that some uses of the MySQL server over a network connection constitute a derived work?

Progress Software put out a press release that suggested MySQL AB was making such a claim, but that was not so. The GPL violation involved static linking of non-free (at the time) code with the mySQL code and distributing the result as a single binary.

As far as I know, the GPL violation by Progress involved static linking of MySQL with non-free code.

Code can remain in active use far longer than many of us would expect. Is the prospect of GPL-licensed code moving into the public domain something to worry about?

No. Even in 1997, copyright lasted a very long time. According to the laws in effect then, the early versions of GNU Emacs would go into the public domain 50 years after I die. So what? Newer versions of Emacs have major contributions by other people, most of them considerably younger.

Do you see the (seemingly continuous) extension of copyright terms as a good thing for GPL-licensed software or not?

I am surprised you would even entertain the notion that some minor secondary benefit could outweigh the great harm done by the Mickey Mouse Copyright Act. I supported the Eldred vs Reno case at the beginning, and still do.

In an earlier message, you said: "People often describe our work as "open source", but that is actually the slogan of another movement which was formed specifically to reject our views." Can you please explain why you believe the OSI was formed for that purpose?

Its founders said so, and if you compare www.opensource.org with the views of the free software movement in www.gnu.org/philosophy, you can see it for yourself. The free software movement is based on an ethical stance: that users are morally entitled to the freedom to share and change software, that non-free software is unethical and wrong. The open source movement rejects this basic view; it does not say or even hint that software should be free (or "open source"). It does not raise this as an ethical issue at all.

Some say that omission is just a strategem, that "open source" is a "marketing campaign" for free software, but I believe the open source movement is sincere and I take it at its word. Some of the OSI's leaders have told me they really don't believe that software morally ought to be free. As for Linus Torvalds, who is regarded as an open source celebrity, he recently condemned all political ideals, saying that anyone who would do without a tempting non-free program for the sake of his freedom is "thinking with his gonads". And here I thought that gonads had to do with sex.

For the open source movement, non-free software is a less than ideal solution to a problem. For the free software movement, non-free software is the problem, and replacing it with free software is the solution.

This is not to say that the open source movement is evil. Nearly all open-source software is also free software; they have convinced many people and companies to develop free software, and that is a contribution to the free software community. However, while contributing to our community in that way, the open source movement weakens the community in another way: by spreading an attitude that is less than firm, they undermine our resolve to win and keep our freedom.

Occasionally one hears rumors of work on version 3 of the GPL. Is there such an effort afoot? If so, what sort of changes to the GPL are anticipated?

We are working on GNU GPL version 3 now. The changes will only be in details, of course. Most are just clarifications, but one substantive change will make the GPL compatible with some additional free software licenses.

Probably the most substantial change would [give a] program's developer a way to insist that an ASP running the program and offering access to the public must make the source code for the program available for public download. Some developers have been unwilling to release software under the GNU GPL without this provision.

What, in your opinion, is the free software community's greatest accomplishment so far?

The GNU/Linux system, and other free operating systems, make it possible to use a computer without proprietary software. That was our initial goal, and that is our major accomplishment.

Now we have to defend this accomplishment, and extend it to the full range of software that users want.

...its biggest disappointment?

Our greatest failure lies in the widespread practice of adding non-free software to the GNU/Linux system: device drivers, libraries, applications, or whatever. The result is backsliding to partly-free systems, which convey to the users and the public the short-term thinking that freedom and community are less important than getting some job done today.

...and what is the biggest threat to free software in the future, and what should we be doing about it?

The biggest direct threats to free software are laws that prohibit release or use of free software for certain jobs.

The DMCA has been used to prohibit free software for a narrow range of jobs which are, however, important to many users, such as reading an e-book or playing a DVD movie. Software patents, monopolies covering ideas that can be used in software, can prohibit free software in all areas. The proposed SSSCA, designed to ban all software inconvenient that the movie companies find inconvenient, could do even more harm.

What we need to do about it is organize politically. We must all offer some of our precious time to defend our freedom from the SSSCA, and win back the freedom that the DMCA and software patents took from us. So please help to organize an anti-DMCA anti-SSSCA group in your area. Try to recruit people with experience from other grass-roots campaigns, and listen to their advice. Above all, recognize that this is more important than whatever technical job you are employed to do today. See digitalspeech.org for discussion groups and other information useful for organizing an effort in your area.

At a deeper level, though, the biggest threat to the future of free software is the idea that non-free software is acceptable. When people adopt that view, it saps their determination; they are less likely to make an effort to protect and extend free software. This weakens our abiity to resist any external threat or solve any difficult problem.