It’s been a while since we caught up with Stallman. But a couple months ago we took a look around at what’s happening with law, politics and technology and realized that he maybe perhaps his extremism and paranoia were warranted all along. So when we were contacted by an Iranian Linux publication and asked if we would like to publish an English translation of a recent interview they had done with Stallman, I thought that it was a particularly rich opportunity.
Iran is a particularly interesting case when it comes to the adoption of free software. Like many nations outside of Europe and North America, there is no ethos of paying for software, and in fact Iran is not a signatory of the WTO copyright treaty, which implicitly and de facto means that foreign software may be freely distributed within Iran without paying licenses. So Windows and other western commercial software is widely and inexpensively available. Nevertheless, Linux and other Free Software has a foothold among the computing community similar to its popularity elsewhere.
So without further ado, here’s the interview:
LR: Before we begin, let me mention that in countries like Iran, people
are not forced by the law to comply with copyright to use computer
software. Almost every piece of software can be obtained for a very
small amount of money or even for free.
Since I am committed to distinguishing “free as in free speech”
from “free as in free beer”, I’ve adopted the practice
of always saying “gratis” and never “free” when I mean “zero price”.
That way, when I say “free”, it always means “azadi”.
LR: Therefore, those who choose to
use and advocate free software do so for pure philosophical reasons
and due to real love for freedom, and they have great respect for you
and what you’ve done.
Since others may also read this text, I include here the definition of
free (azadi) software.
Free software means, software that respects the users’ freedom and
community. For any program, either the users control the program or
the program controls the users.
In order for the users to control the program, they need these four
0. To run the program as you wish.
1. To study the program’s source code, and change it, so it does
your computing as you wish.
2. To make exact copies and redistribute them to others when you wish.
3. To make copies of your modified versions and redistribute them to
others when you wish.
We call that “free software”.
If the users don’t control the program, then the program controls the
users, and the developer controls the program. This is nonfree
software. It’s an injustice, and our goal is to get rid of the
I launched the free software movment in 1983. In 1984 I started
developing a free software operating system called GNU. In 1992, GNU
was nearly complete, missing one essential component. In that year,
Torvalds liberated Linux, which filled the last gap in GNU. The
operating system that millions use is basically GNU, with Linux added.
LR: When you started the GNU Project, you stepped down from your
job at the university. Considering the fact that the project
needed investment, did you receive any financial support? Did you
have a major sponsor?
The term “investment” is not applicable here, because that implies
spending money on a business to obtain a larger subsequent profit. I
set out to do a large job, but it wasn’t a business and the purpose
was something more important than profit.
I quit my job at MIT when I started writing code for the GNU operating
system because I wanted to make sure MIT would not be able to claim
copyright on the code I wrote for GNU.
Evidently, financial support was not crucial at the beginning, because
I made progress on my own, which drew others to help.
LR: Does the GNU Project produce any revenues to cover it’s own
That question presumes a centralized project which is not how it
works. The GNU system is composed of many components, and each
component has its own developers. Some of those groups obtain funding
in various ways, but they do it separately and independently. Some
groups are composed solely of volunteers.
LR: Could you please update our readers on the current status of
HURD? Do you agree that HURD is now even more important than
before due to the rejection of GPL v3 by the Linux Foundation?
The Hurd needs lots of work to be competitive in practice with Linux,
but I don’t keep track of the details.
The problem caused by having Linux under GPL v2 is that it can be
tivoized — used in “tyrant” products that don’t allow the user to
replace the operating system with her own version.
Let’s assume the Hurd were practically as complete as Linux. The Hurd
does not allow tivoization, so it can’t be used in tyrant products.
But that would not stop them from making tyrant products — they would
use Linux instead.
Thus, in order to put a stop to tyrant products, the Hurd would need
to be not just equal to Linux, but far superior. That does not seem
to be a likely prospect. As a result, I don’t know any way we can
technically put an end to tyrant products that use free software.
What we can do, at least, is to stop them from using our code.
Releasing our code under GNU GPL v3-or-later achieves that.
LR: Mr. Torvalds has provided almost no sensible and acceptable
reason for refusing to update the license. In his opinion that’s
not necessary and the version 2 is enough. What do you think about
his decision? Will linux lose support from corporations’ side if
he agrees to use GPL v3?
He told me that he appreciates many of the improvements in GPL v3 but
objects to one of them: the requirement for manufacturers to tell
users how to install modified software in the products they buy.
Many of today’s computers are designed so the manufacturer can upgrade
the software but the user can’t change it. Under those circumstances,
the executable is not free software even though its source code may be
free. We call this “tivoization” or “lockdown”, and we call the
device a “tyrant”. Many Android phones are tyrants: they contain
executables of Linux compiled from free source code, but the
executables are nonfree since users are blocked from running
GPL version 3 prevents this abuse by requiring the manufacturer to
tell you whatever key is needed to sign your own version of the
program so it can run in the device. That is what Torvalds objects
to. He wants to permit tivoization of Linux, and according to what he
told me, that is his reason for rejecting GPL version 3.
LR: Does GPL v3 impose limits on installation of multimedia codecs
and proprietary applications?
I can’t see why it would do that, but the question is vague, so I have
trouble answering it. If you specify a more precise scenario, I would
be able to give a sure answer.
LR: Right now there are many non-free binary blobs inside the
linux kernel. Don’t you think that linux is moving off from
freedom towards being a proprietary software?
Linux moved off the path of freedom when the binary blobs were
included in it. Whether it is now moving even further from that path,
I don’t know. Either way, freedom calls for rejecting the blobs.
That is why we maintain Linux-libre, our own 100% free software
modified version of Linux.
LR: Right now, what’s the main focus of FSF and what exactly the
The FSF has four main activities: infrastructure, campaigns, GPL
enforcement, and education/fundraising.
Infrastructure means the servers such as www.gnu.org and
savannah.gnu.org that GNU uses.
GPL enforcement means writing to companies that violate the GPL (with
software copyrighted by the FSF) and informing them that that is
copyright infringement and they must stop.
Educating the public about free software and fundraising are combined
because the same activities usually achieve both results.
LR: There’s been a lot of debate on what is the proper title for
GNU/Linux. However many users still have a lot of questions in
this regard. Most references offer Ã‚ definitions for an operating
system that are more or less like this: “A system software that
connects the software with the hardware.” What that definition in
mind, why shouldn’t we call an operating system simply “linux” or
The right name depends on what collection of software you’re talking
With the definition cited above, you’re talking about the kernel and
drivers — only. Linux is a kernel, so perhaps you might be talking
However, if you’re talking about a distro such as Trisquel or Debian,
that’s a lot more than “software that connects the software with the
hardware”. Those distros are complete, usable systems — basically
the GNU system, with Linux added. So you should call them
I define “operating system” as “the collection of software that
implement the usual activities and provide a base for implementing
other activities.” GNU is an operating system in this sense, as was
Unix, and so are these distros.
By the way, “software” has no plural. It is like “literature” or
“petrol”. Any number of programs are “software”.
LR: Recently Digitizer published statistics (based on an analysis
on the number of lines of codes for all software inside Ubuntu
11.04 main repositories. They announced that only about 8 percent
of the code belonged to the GNU Project. What do you think about
I don’t know how they reached that figure. How did they claim to
determine which code belonged to the GNU Project? There is no simple
way to determine that from the code in any simple way. Perhaps they
omitted some GNU packages and undercounted.
However, it will probably be true in a few years that GNU packages
are only 8%. When that happens, Linux will probably be 0.8%.
These numbers both decrease, and the reason is that more different
projects contribute to the system. The decrease reflects our success
at attracting more free software developers.
This has no effect on the relationship between GNU, Linux, and Ubuntu.
Ubuntu is a development of the GNU/Linux system, which started in 1992
as the GNU system plus Linux.
LR: You announced that in gNewSense, GNU’s share was %15 while
Linux only had a %1.5 share.
I don’t remember the precise figures, but that looks about right.
What this shows is that the system is more GNU than Linux.
LR: Therefore, if we choose the amount of code as the criterion,
each distribution would probably have a different name, as many
projects other than Linux and GNU are involved in distros, and the
ratios are constantly changing.
If the question is whether to call the system “GNU/Linux” or “Linux”,
those things have no effect on that. However the rest of the system
may change, GNU remains a much bigger contribution to the system than
Linux is. See http://www.gnu.org/gnu/gnu-linux-faq.html#many.
LR: Do you think that the amount of
code is a proper index for naming a distro?
No, I don’t think that is the sole basis. I mention it because it is
an objective and verifiable fact about the system today.
However, I think the fact that the GNU Project started the development
of this system, and gave it the name GNU, is a stronger reason to
consider the system as a development from GNU.
LR: Aside from GCC, which part of GNU could be regarded Ã‚ part of
the Ã‚ operating system core? What actually justifies calling GNU
the base of GNU/Linux?
Binutils, Coreutils, Bash and GNU libc are just as central and
necessary as GCC and Linux. However, this concept of “core” seems
to be an artificial distinction. From the beginning, we aimed
to make GNU a complete Unix-like system, not merely a “core”.
LR: Then could we still call a GNU-free operating system (which
uses EKOPath instead of GCC and zsh instead of bas etc) GNU/Linux?
If you replace all of GNU, the result would not be GNU/Linux. But why
seek to do such a thing?
LR: The number of distributions is increasing rapidly. Although
this helps free software to cater to Ã‚ tastes and needs of more
people, some people say it is a waste of time and energy that
could have otherwise been spent on upstream projects. Do you agree
with them? What’s your idea about a unified, Ã‚ standard distro?
People and organizations often think that the first thing to do when
they join the free software community is to make a distro. If their
goal is to develop their skills, perhaps this is a useful way to do it,
but I think most of these distros don’t serve much other purpose.
Thus, I urge people to concentrate their efforts on helping the
existing free distros, rather than making new distros.
See gnu.org/distros for information about the free GNU/Linux distros.
LR: Contrary to the original ambitions of the Free Software
movement, many distros are shipping non-free programs. What harm
could this do to the movement? What can be done about this issue?
The direct effect of the use of nonfree distros such as Ubuntu
GNU/Linux is that some people install GNU/Linux but they don’t get all
the way to freedom.
If that were the worst consequence of nonfree distros. it would be
unfortunate but would not put the movement’s goal at risk. Even if
these users have not got all the way to freedom, they have taken a big
step towards freedom. However, the nonfree distros have another
effect that does much deeper damage: it gives people the wrong idea of
what our work is intended for.
The developers of Ubuntu don’t say, “We’re sorry that this distro does
not fully deliver the freedom that only free software respects.” They
don’t mention “free software” at all. Instead they talk about “open
source”, and they say that their goal is to deliver the “best possible
So what is our goal? Is it to win freedom, or to have an appealing
user experience? It is for you to decide what your goal is. However,
newcomers to the free software community are more likely to get their
ideas from Ubuntu than from the free software movement. Thus, most of
the people who come into our community learn to seek the “best
possible user experience” rather than their freedom.
To prevent this from washing away the free software movement is a
constant effort, and that’s why I am so strongly concerned about the
problem of nonfree distros.
LR: Our readers are really eager to know what distribution and
desktop environment you personally use.
I use gNewSense, which until recently was the only 100% free software
GNU/Linux distro that ran on my computer. This computer, a Lemote
Yeeloong, has a Chinese processor which is more or less a MIPS.
There is now one other distro that runs on the Yeeloong, Parabola,
and I hope to try it soon.
I do most of my work in a text terminal, because that is more
efficient for me, but I also have GNOME running.
LR: A number of institutes committed to the spread of free
software like Sourceforge and Google Code ban IP addresses from
Iran and other companies under US sanctions and restrict their
services to those countries. In addition, a few months ago Redhat
prevented active Iranian contributors from entering its managerial
board. What”s your opinion? Which one is more more important,
abiding by US laws or helping all users have access to and a say
in free software?
You’re comparing a purely legal question with an ethical question, and
there is no comparison between them.
US organizations can’t get away with disobeying US export control law.
They would be stopped. The FSF obeys US export control law too, and
would not sell products to Iran. Whether this law is just or not, I
can’t blame US organizations for complying with it.
But does US export control law require repositories to block
connections from Iran? The FSF’s lawyer told us that we are not
required to block connections, so we don’t block connections. Why the
staff of Sourceforge and Google think this is required, I don’t know.
It would be interesting to ask them, but I think they would give
Turning from the legal question to the ethical question, it is wrong
and absurd for US to apply trade sanctions to free software. That
doesn’t stop business or the state in Iran from getting and using free
software, it only inconveniences individual Iranians, while hampering
them from contributing to the world’s shared knowledge.
LR: As we assume that many software developers don”t actually care
about those sanctions, many Iranian users bypass the censorship
and download the code. Do you think they”re doing something bad,
both morally and legally?
Legally speaking, they are safe: the US cannot prosecute people in
Iran for doing this. Morally speaking, it is wrong for trade
sanctions to apply to free software.
LR: How can a software developer make a living out of free
software? Apart from donations and paid support, are there any
revenue models for free software?
Most paid software development consists of writing custom software.
That business could exist almost unchanged in a world in which all
software is free: the same clients would pay for the same work, and
get similar software, but they would receive it as free software.
There are numerous small companies which already do this.
This is not the only model, but it is a common one.
LR: As you”ve admitted before, the open-source model has won more
fans compared to the free software model.
There is a misunderstanding here: these are not two “models”. Free
(azadi) software is not a “model”. It is a philosophy, a campaign for
certain specific freedoms for software users on ethical grounds. (See
Open source is also a philosophy, but it’s a very different one, which
is not about ethics; its values are values of practical convenience.
Open source seems to have more supporters. I don’t know why, but I
have guesses. First, the media usually adopt the open source
philosophy’s assumptions and mention only that name. Thus, more
people encounter the open source philosophy than ours. Second, open
source disagrees less with the establishment than we do, and that
makes its ideas less controversial.
LR: Some people believe that the free software movement and its
philosophy is n ot limited to computers and could be generalized
to other branches of science. Do you agree with them? Do you
believe that something called “free science” or “free knowledgeÃ¢
could be pursued? How, the way you see it, is it possible to
trigger similar movements in other fields of knowledge?
The free software movement applies to software. What is software?
A program is a work of authorship which is used to tell a computer how
to do a job. In other words, it’s a work used to do practical jobs.
Users deserve to have control of the jobs they do, so they need to have
control over the work. That requires the four freedoms.
The same argument applies to other works that are designed to do
practical jobs. For instance, recipes for cooking, textbooks for
learning or studying, and reference works for looking things up.
All these should be free.
“Knowledge” is a very different kind of thing from software. Works
can relate to knowledge: a program can embody specific knowledge, and
a textbook can teach specific knowledge, but knowledge is not a work
like a program or a textbook. Thus, I don’t think we can talk about
“free knowledge” in this particular sense of free. Knowledge raises
different ethical issues. Likewise science.
LR: What”s your idea about adding a new term to free licenses that
bans the use of free software for military purposes? As you know
free software is being widely used by governments for military
purposed and sometimes for oppressing people. Can anything be done
Freedom 0 is the freedom to run the program as you wish. If a license
restricts how you can run the program, the program is not free
This criterion is crucial. We cannot accept programs in the GNU
system which have limits on what they can be used for. If we did,
different programs would come with different limits. One program,
perhaps written by Muslims, might ban use by restaurants that serve
alcohol; another program, perhaps written by the Munich Oktoberfest
committee, might ban use by restaurants that do not serve alcohol.
Continuing along these lines, we might end up with a system that
nobody would be allowed to use.
The idea that we could stop governments from launching wars of
aggression, or from torture, by conditions in software licenses is
absurd anyway. Who enforces these licenses? Governments do. If we
sued a government for using our software in those activities, the
government would say, “We were compelled to go to war to protect the
country, and that wasn’t torture, merely ‘enhanced interrogation’.”
Or it would simply legislate an exception for itself.
LR: What do you know about free software community in Iran? if
you”re invited to events related to free software will you accept
I know very little about the free software community in Iran. Even in
some countries which I visit often, and in which I know some of the
local issues and activists, I would find it hard to answer such a
I would not take the risk of visiting Iran under its current
government. I make statements that criticize that government, and I
fear I might be imprisoned or worse if I went there.
LR: Looking at the past, what do you think have been your best and
Sorry, I don’t like confessing my biggest mistakes to strangers.
LR: What should be the next step in helping the society to care
about and protect its freedom?
I have trouble responding because the question is not really
meaningful. This doesn’t work by “steps”. Rather, there are many
different areas where we are pushing, and many areas where our
adversaries are pushing, and it all happens in parallel.
However, it is clearly crucial to make a tablet and a mobile phone
that can be run without nonfree software.
LR: Your comments about Steve Job’s death created a lot of debate
among Apple fans and Free Software advocates. A number of Free
Software and Open-source figures like Eric Raymond criticized your
tone and blamed you for what they called being too harsh and
insulting Apple users.
Actually, Eric Raymond said I was right about this. He seemed to
uncomfortable agreeing with me, and added gratuitous insults so the
tone would not seem friendly. But he said I was right on every point
of this issue.
Others seemed to have a knee-jerk condemnation of criticizing someone
who has died. One person told me it was unfair for me to criticize
Jobs when he could not defend himself — never mind that the entire PR
staff of Apple was busy telling us how much the world had lost.
Frankly, I don’t care about Jobs very much either way. I am concerned
about the people whose freedom is harmed by his products, and by other
products that follow the same avenue.
LR: What’s the best way to advocate Free Software? Some Free
Software users engage in technical debates with Microsoft and
Apple fans, trying to convince them GNU/Linux is more
powerful. Another group focus on philosophical and cultural
aspects of Free Software and try to make people care about their
freedom. Which of the two mentioned approaches are more effective?
They are both “effective” but they lead to different results.
If you convince people that some free software is technically
superior, they might run some free software, but they will remain
ready to use nonfree software in the areas where that is technically
superior. They will continue to judge an important question based on
superficial issues. This is just a partial success.
However, if you convince people that they deserve freedom, they will
start rejecting nonfree software whether it is technically inferior or
technically superior, because they will see that free software is
ethically superior. They will understand the important question and
judge it right. This is a full, deep success.
Another weakness of technical arguments is that nontechnical people
probably won’t care about them at all. But they can understand
ethical arguments. Ethical arguments are the only way we can convince
nontechnical people to become free software supporters.
I figure that users can judge for themselves whether program A is more
convenient than program B. So I don’t try to convince them about that
sort of question, except when someone has preconceptions about free
software and has not tried it. I focus on talking about freedom.