Firstly, I will agree, largely, with the discussion defining
ethics in Section 1. The last paragraph laments that we do
not live in an ideal world. Granting that, I would point out that
there is still a place for idealism in an imperfect world.
Section 2, nominally, defines the term "free software", as it is
used in the rest of the article. Its true purpose, evidently, is
to set the tone of hostility that Meyer depends on for the next
several paragraphs. The section opens with a discussion of
two papers on the GNU web site. These are discussions of why we use
the GPL to license our software. They are emphaticly not just rants
about the deficiencies of other licenses. Perhaps Meyer missed the
point of these articles. Rereading them in the context of GPL
advocacy might lead to a different viewpoint.
Meyer summarizes his definition of "free software" with this list.
- Is available from at least one source without payment (which does not
preclude other sources from offering it for payment, for example to people
who want a distribution on CD rather than downloaded, or require commercial
- Can be used for commercial as well as not-for-profit development, even by
people who have not paid for it. (There may be some restrictions on
commercial uses, for example the requirement that additions to the free
software be free too.)
- Can be obtained in source code form.
Contrary to Meyer's subsequent contention, this is fairly close to
the GNU definition of Free Software. In simple terms, there are
four levels of software freedom for the GNU project.
- What we call "proprietary" software which, if the source
code is available at all, requires that the user not disclose any of
it to another party.
- Free Software, which is released under terms that look remarkably
like the ones listed above. It should be added that "restrictions on
commercial uses" is misleading, since these are general restrictions
which have nothing to do with whether the use is commercial or not.
- GPL software, developed independently of the GNU project but released
under the same license that GNU uses.
- GNU software, code for which the GNU Project holds joint or sole
copyright and is therefore able to assure its continued freedom.
There are actually two levels of GPL software, spelling out more specific terms
for derived works. Interested readers may want to look at a more
of this, including critical examinations of several popular licenses.
Closing Section 2 is an interesting paragraph which appears to be
an attempt at pushing well known alarmist buttons. It includes
gratuitous mentions of Stalinism and the Nazis, further buttressing
the hostile tone inaugurated in the first paragraph.
Section 3, which Meyer calls "The Economics of Free Software" has
virtually nothing to do with Free Software. It's 100% free beer and
0% free speech. This is the section where Meyer reveals his near
total misunderstanding of the Free Software movement in general and
the personal motivations behind it, in particular. His use of the
term "Donated software" is a rather clear example of this. One
"donates" his time or his money, not his ideas. Ideas are something
that are shared not donated. Donate is a terrible word to
use in the context of ideas. Did Lincoln "donate" the Gettysburg
Moving forward, we find a claim that Free Software is, in fact,
Taxpayer-funded software. Sigh, another free beer argument. Some
Free Software is taxpayer-funded, but that's hardly the point. The
way that a software project is paid for has nothing to do with
whether the source code is made available as part of the
I've worked in several proprietary shops over a 20 year career. A
typical business model goes something like this: "Write a cool
program, assure that your competitors can't look at how you wrote
it, sit back on your thumbs while waiting for the gravy train to
roll in." This model will, clearly, not work if the source code is
released with the product because your competitors will build on
your work. Free Software must be constantly improved if it
is to be economically successful. You must always stay ahead of
your competitors by having a better program, not by maintaining
secrecy about your methods.
Here's an interesting piece about the role of Universities in
"The story goes on to state that Stallman "resigned"-- presumably
meaning that he stopped using the MIT's machines, since it appears
from the above that he had already resigned --- because "sometimes,
universities take software written by their employees to sell them
as proprietary products". (What a shame indeed: that a university
would think it has any rights at all on products developed by
people it pays, on machines that it owns!)"
Universities, Mr. Meyer, do not develop products. They are
laboratories of human thought. Occasionally they release a public
gem for everyone to gaze at. In return for their support
Universities have granted some of the most structural advances in
human history. A University could claim all rights to the
creativity that takes place there, but then it could no longer call
itself a University. A shame indeed.
Meyer opens Section 4 with yet another round of free beer
"The best-known figure of free software, Richard
Stallman from GNU and the Free Software Foundation (FSF),
professes an absolute refusal of any notion of commercial
software. Software should be free, period."
Again, it's about the source code, not the price. And it has
nothing to do with whether the software is commercial or not.
A little later in this section we find this passage:
The GNU and FSF view is that it is OK to sell anything
except software. (To be precise, I have not found any example of
something else whose selling they find immoral; satellite signals
might seem a logical candidate.) Computers are OK; services are
OK. But if the work of your life is a great software package,
trying to make a living out of selling it --- unless you also give
it away, an immediate business-killer --- is a moral
I'm most interested in the assertion that distributing the source
code to a project is an "immediate business-killer". Since this
assertion is unqualified, the existence of one counter-example
should be enough to refute it. I have two. I present Michael
Tiemann at Cygnus, who founded one of the first commercial Free
Software businesses and grew it quite successfully for 10 years
before selling it (to another Free Software business), last year,
for almost $700 million. Second, Ghostscript, distributed under
the GPL by Aladdin Software, who, if you like, will sell you a non-GPL
version for a fee.
In another context, Section 5 might have been the most thought
provoking part of the essay. As a deliberately inflammatory, off
topic ramble, it's a distraction. Since I'd like to stay on topic
I'll be brief: Britain is a great example of a country where guns
are not readily available. In Britain they use bombs.
Section 6 is about quality, a subject where comments on the GNU
system are very positive. The variability Meyer discusses is far
more likely in adaptations of GNU software to non-GNU systems. The
examples he uses fall into this category.
As he probably knows, DOS compatibility is a rather tertiary
concern for the GNU project, our primary mission is the GNU
System, which is an
compatible. There are examples of very well done ports of GNU
software to DOS but all of these efforts have been carried out
independently of the GNU Project. Some of those independent
developers, I'm sure, would have been pleased to accept a consulting
contract for rather more prompt delivery of bug fixes.
It's fairly clear from this article that Meyer's position on
"free" is gratis and not liberty. I saw no mention of any efforts
to make his company's resources available to contributers, to solve
even their own problems.
I'm not sure what prompted Ken Thompson's claim that GNU/Linux
was, in his opinion, less reliable than Windows. I just know that
it totally contradicts my experience. I've run hundreds of GNU/Linux
machines since shortly after they became usable about 8 years ago.
In that time I've seen two examples of outright wedges. Both of
them were on a highly experimental machine that was well
over-clocked. We have one mission critical server in the project
that has been up for over a year with essentially zero intervention
Then we come to a comparative analogy between product F and
product P. After a little alarmist discourse about the life
threatening possibilities of this decision (subsequently retracted
and therefore gratuitous) we are asked to consider the ethical
considerations of the following choice:
- Product F is free software. It comes with the standard no-warranty
- Product P is proprietary software. It costs $50 for the binary-only
version. It uses the most advanced techniques of software engineering. It
never crashes, or departs in any way from its (mathematically expressed)
specification. The seller is, in fact, so sure of those qualities that he
will commit in writing that any violation of the specification during
execution will immediately lead to reimbursement of the purchase price and
compensation for any damages incurred.
As an idealist, product F is very attractive, suppose that product P
is, indeed, mathematically correct and perfectly implemented, but it does
not do precisely what I want. Having chosen product F, I fix the bugs
and add the changes that I need and I have the solution to my problem.
As a pragmatist, should product P ever become available, it will
probably not do what I want anyway.
I don't have much on Section 7, it reads a little like the USENET religious
wars of old, replete with references to "scripture" and "sainthood".
I'm a nuts and bolts man and I don't care much about the glorious nature
of the various personalities involved. Richard Stallman is my friend as well
as my boss. He is, above all, a human being, and a good one. He has some
towering successes to mitigate his agonizing failures. His attempts to lead
an ethical life will probably be discussed for a very long time. I respect
him for following his instincts.
Section 8 is, probably one of the more cogent parts of the essay, if only
for its accuracy. There are various reasons for writing Free Software and
many of them are well described here. Were it not for the openly hostile
prelude to this section, it might be an interesting catalog of the motives
behind the Free Software Movement. Unfortunately this has already been
addressed in the category-killer
Eric Raymond, who Meyer vilified
as a terrorist for several paragraphs earlier in the article. So what is
left is an attempt to discredit the author of a seminal article on the
social anthropology of Free Software, and then reassert Raymond's conclusions
as his own.
Section 9 deals, specifically, with Microsoft bashing. I, for
one, do not discount the contribution of Microsoft to the "Wintel"
alliance in driving the price of hardware down. I simply do not
recognize Microsoft as the vendor of the best system to run on the
newly cheap hardware. This is, in large part, because of
Microsoft's resistance to accepted industry standards and refusal to
release it's source code.
GNU was chartered to build a source code release of a UNIX
compatible system. UNIX was chosen because it was popular and
powerful. Far more powerful, in fact, than anything I have yet seen
from Redmond, after 20 years. Microsoft developed DOS to run on a
class of computers that were originally considered the toys of the
industry and made all of the compromises necessary to do so. UNIX
came down to PC class hardware from the glass room only after the
machines became powerful enough to accommodate the protected,
multi-processing system that UNIX always was. To appreciate the
difference, one needs only to compare COMMAND.COM to bash.
The origin of any publicly perceived arrogance on the part of
Microsoft rests entirely within Microsoft. They, clearly, have no
shortage of talent in software development. So the failure must be in
public relations or management. Since there is no apparent shortage of
public relations talent in Redmond, I suggest that the problem may be
Section 10 discusses the fact that many programs that run on the
GNU System are re-implementations of existing programs. This is not
significant since GNU implementations are released as source code
and are therefore more durable. This durability is an enhancement.
Adobe could tank at some point and that would be it for PhotoShop,
but Gimp will be still be there as a tarball on some ftp site. As
for acknowledging Adobe for its efforts on PhotoShop, why?
PhotoShop was based on well known image processing techniques,
packaged conveniently for Macintosh people.
Finally we get to the 11th and last section. Here Meyer, after having
blasted the Free Software Community with both barrels, extends an olive
branch in the form of a call for tolerance and cooperation. Well, sure,
I'd love to, except that you just blew away my first born son.
Frankly, I can't conclude from this article that Mr. Meyer has any
interest in tolerance and cooperation at all.
I'll stop now, after a couple of personal observations about GNU.
GNU is not something that someone gives you, it's the neighborhood
where you live. The inhabitants prosper by sharing code with each
other, and we do prosper. Free software is not going to go away
anytime soon. I hope to live another 30 or 40 years and it will
certainly be around as long as I am alive. We haven't found truth
Meyer, we've just found something that works, license and all, for
us. If it works for you, welcome aboard. If not, either fix it or
get out of the way.