source: non-gtk/emoglen/ @ 829

Last change on this file since 829 was 829, checked in by Александър Шопов, 15 years ago

Най-накрая - валиден DocBook? 4.2

File size: 81.9 KB
1<?xml version="1.0" standalone="no"?>
2<!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN" "" [
4<article lang="bg">
8<!-- <html><head>  --> 
9<!-- base href="" --><title>Anarchism Triumphant</title>
11<meta name="Description" content="This paper shows why free software, far from
12being a marginal participant in the commercial software market, is the
13first step in the withering away of the intellectual property system.">
14<meta name="Keywords" content="anarchism triumphant, free software, death of copyright, Linux operating system kernel, software as property, article">
15<meta name="DC.Title" content="Anarchism triumphant">
16<meta name="DC.Title" content="Free software and the death of copyright">
17<meta name="DC.Creator" content="Moglen, Eben">
18<meta name="DC.Subject" content="anarchism triumphant, free software, death of copyright, Linux operating system kernel, software as property, article">
19<meta name="DC.Description" content="This paper shows why free software, far from being a marginal participant in the commercial software market, is the
20first step in the withering away of the intellectual property system.">
21<meta name="DC.Publisher" content="Valauskas, Edward J.">
22<meta name="DC.Publisher" content="Dyson, Esther">
23<meta name="DC.Publisher" content="Ghosh, Rishab Aiyer">
24<meta name="DC.Date" content="1999-08-02">
25<meta name="DC.Type" content="text">
26<meta name="DC.Format" content="text/html">
27<meta name="DC.Identifier" content="">
28<meta name="DC.Language" content="en">
29<meta name="DC.Relation" content="IsPartOf First Monday, vol 4, no. 8"></head><body alink="#ffee99" bgcolor="#ffffff" link="#bb7777" text="#000000" vlink="#7777bb">
31<blockquote><img src="anarchism_files/logo.gif" alt="First Monday" align="bottom" border="0" height="40" width="256"><br>
37<para><ulink url=""><!-- <img src="anarchism_files/moglen.gif" alt="Anarchism Triumphant: Free Software and the Death of Copyright" border="0">--> </ulink></para>
39<para><blockquote> <para>The spread of the Linux operating system
40kernel has directed attention at the free software movement. This
41paper shows why free software, far from being a marginal participant
42in the commercial software market, is the vital first step in the
43withering away of the intellectual property
44system.</para></blockquote> </para>
46<!-- Should be automatically generated
49<para><a href="#m1">I. Software as Property: The Theoretical Paradox</a><br>
50<a href="#m2">II. Software as Property: The Practical Problem</a><br>
51<a href="#m3">III. Anarchism as a Mode of Production</a><br>
52<a href="#m4">IV. Their Lordships Die in the Dark?</a><br>
53<a href="#m5">Conclusion</a></para>
56<title>I. Software as Property: The Theoretical Paradox</title>
58<para><emphasis>SOFTWARE</emphasis>: no other word so thoroughly
59connotes the practical and social effects of the digital
60revolution. Originally, the term was purely technical, and denoted the
61parts of a computer system that, unlike "hardware," which was
62unchangeably manufactured in system electronics, could be altered
63freely. The first software amounted to the plug configuration of
64cables or switches on the outside panels of an electronic device, but
65as soon as linguistic means of altering computer behavior had been
66developed, "software" mostly denoted the expressions in more or less
67human-readable language that both described and controlled machine
68behavior <footnote> <para>1. The distinction was only approximate in
69its original context. By the late 1960's certain portions of the basic
70operation of hardware were controlled by programs digitally encoded in
71the electronics of computer equipment, not subject to change after the
72units left the factory. Such symbolic but unmodifiable components were
73known in the trade as "microcode," but it became conventional to refer
74to them as "firmware." Softness, the term "firmware" demonstrated,
75referred primarily to users' ability to alter symbols determining
76machine behavior. As the digital revolution has resulted in the
77widespread use of computers by technical incompetents, most
78traditional software - application programs, operating systems,
79numerical control instructions, and so fort - is, for most of its
80users, firmware. It may be symbolic rather than electronic in its
81construction, but they couldn't change it even if they wanted to,
82which they often - impotently and resentfully - do. This "firming of
83software" is a primary condition of the propertarian approach to the
84legal organization of digital society, which is the subject of this
85paper.</para> </footnote> .</para>
87<para>That was then and this is now. Technology based on the
88manipulation of digitally-encoded information is now socially dominant
89in most aspects of human culture in the "developed" societies
90<footnote> <para>2. Within the present generation, the very conception
91of social "development" is shifting away from possession of heavy
92industry based on the internal-combustion engine to "post-industry"
93based on digital communications and the related "knowledge-based"
94forms of economic activity.</para></footnote>.  The movement from
95analog to digital representation - in video, music, printing,
96telecommunications, and even choreography, religious worship, and
97sexual gratification - potentially turns all forms of human symbolic
98activity into software, that is, modifiable instructions for
99describing and controlling the behavior of machines. By a conceptual
100back-formation characteristic of Western scientistic thinking, the
101division between hardware and software is now being observed in the
102natural or social world, and has become a new way to express the
103conflict between ideas of determinism and free will, nature and
104nurture, or genes and culture.  Our "hardware," genetically wired, is
105our nature, and determines us. Our nurture is "software," establishes
106our cultural programming, which is our comparative freedom. And so on,
107for those reckless of blather.<footnote><para>3. Actually, a moment's
108thought will reveal, our genes are firmware. Evolution made the
109transition from analog to digital before the fossil record begins. But
110we haven't possessed the power of controlled direct
111modification. Until the day before yesterday. In the next century the
112genes too will become software, and while I don't discuss the issue
113further in this paper, the political consequences of unfreedom of
114software in this context are even more disturbing than they are with
115respect to cultural artifacts.</para></footnote> Thus "software"
116becomes a viable metaphor for all symbolic activity, apparently
117divorced from the technical context of the word's origin, despite the
118unease raised in the technically competent when the term is thus
119bandied about, eliding the conceptual significance of its
120derivation.<footnote><para>4. <emphasis>See, e.g.,</emphasis>
121J. M. Balkin, 1998. <emphasis>Cultural Software: a Theory of
122Ideology.</emphasis> New Haven: Yale University
125<para>But the widespread adoption of digital technology for use by
126those who do not understand the principles of its operation, while it
127apparently licenses the broad metaphoric employment of "software,"
128does not in fact permit us to ignore the computers that are now
129everywhere underneath our social skin. The movement from analog to
130digital is more important for the structure of social and legal
131relations than the more famous if less certain movement from status to
132contract <footnote><para>5. <emphasis>See</emphasis> Henry Sumner
133Maine, 1861. <emphasis>Ancient Law: Its Connection with the Early
134History of Society, and Its Relation to Modern Idea.</emphasis> First
135edition. London: J. Murray.</para></footnote>.  This is bad news for
136those legal thinkers who do not understand it, which is why so much
137pretending to understand now goes so floridly on.  Potentially,
138however, our great transition is very good news for those who can turn
139this new-found land into property for themselves. Which is why the
140current "owners" of software so strongly support and encourage the
141ignorance of everyone else. Unfortunately for them - for reasons
142familiar to legal theorists who haven't yet understood how to apply
143their traditional logic in this area - the trick won't work. This
144paper explains why<footnote><para>6. In general I dislike the
145intrusion of autobiography into scholarship. But because it is here my
146sad duty and great pleasure to challenge the qualifications or
147<emphasis>bona fides</emphasis> of just about everyone, I must enable
148the assessment of my own. I was first exposed to the craft of computer
149programming in 1971. I began earning wages as a commercial programmer
150in 1973 - at the age of thirteen - and did so, in a variety of
151computer services, engineering, and multinational technology
152enterprises, until 1985. In 1975 I helped write one of the first
153networked e-mail systems in the United States; from 1979 I was engaged
154in research and development of advanced computer programming languages
155at IBM. These activities made it economically possible for me to study
156the arts of historical scholarship and legal cunning. My wages were
157sufficient to pay my tuitions, but not - to anticipate an argument
158that will be made by the econodwarves further along - because my
159programs were the intellectual property of my employer, but rather
160because they made the hardware my employer sold work better. Most of
161what I wrote was effectively free software, as we shall see. Although
162I subsequently made some inconsiderable technical contributions to the
163actual free software movement this paper describes, my primary
164activities on its behalf have been legal: I have served for the past
165five years (without pay, naturally) as general counsel of the Free
166Software Foundation.</para></footnote>.</para>
168<para>We need to begin by considering the technical essence of the
169familiar devices that surround us in the era of "cultural software." A
170CD player is a good example. Its primary input is a bitstream read
171from an optical storage disk. The bitstream describes music in terms
172of measurements, taken 44,000 times per second, of frequency and
173amplitude in each of two audio channels. The player's primary output
174is analog audio signals <footnote><para>7. The player, of course, has
175secondary inputs and outputs in control channels: buttons or infrared
176remote control are input, and time and track display are
177output.</para></footnote>. Like everything else in the digital world,
178music as seen by a CD player is mere numeric information; a particular
179recording of Beethoven's Ninth Symphony recorded by Arturo Toscanini
180and the NBC Symphony Orchestra and Chorale is (to drop a few
181insignificant digits) 1276749873424, while Glenn Gould's peculiarly
182perverse last recording of the Goldberg Variations is (similarly
183rather truncated) 767459083268.</para>
185<para>Oddly enough, these two numbers are "copyrighted." This means,
186supposedly, that you can't possess another copy of these numbers, once
187fixed in any physical form, unless you have licensed them. And you
188can't turn 767459083268 into 2347895697 for your friends (thus
189correcting Gould's ridiculous judgment about tempi) without making a
190"derivative work," for which a license is necessary.</para>
192<para>At the same time, a similar optical storage disk contains
193another number, let us call it 7537489532. This one is an algorithm
194for linear programming of large systems with multiple constraints,
195useful for example if you want to make optimal use of your rolling
196stock in running a freight railroad. This number (in the U.S.) is
197"patented," which means you cannot derive 7537489532 for yourself, or
198otherwise "practice the art" of the patent with respect to solving
199linear programming problems no matter how you came by the idea,
200including finding it out for yourself, unless you have a license from
201the number's owner.</para>
203<para>Then there's 9892454959483. This one is the source code for
204Microsoft Word. In addition to being "copyrighted," this one is a
205trade secret.  That means if you take this number from Microsoft and
206give it to anyone else you can be punished.</para>
208<para>Lastly, there's 588832161316. It doesn't do anything, it's just
209the square of 767354. As far as I know, it isn't owned by anybody
210under any of these rubrics. Yet.</para>
212<para>At this point we must deal with our first objection from the
213learned. It comes from a creature known as the IPdroid. The droid has
214a sophisticated mind and a cultured life. It appreciates very much the
215elegant dinners at academic and ministerial conferences about the
216TRIPs, not to mention the privilege of frequent appearances on MSNBC.
217It wants you to know that I'm committing the mistake of confusing the
218embodiment with the intellectual property itself. It's not the number
219that's patented, stupid, just the Kamarkar algorithm. The number
220<emphasis>can</emphasis> be copyrighted, because copyright covers the
221expressive qualities of a particular tangible embodiment of an idea
222(in which some functional properties may be mysteriously merged,
223provided that they're not too merged), but not the algorithm. Whereas
224the number isn't patentable, just the "teaching" of the number with
225respect to making railroads run on time. And the number representing
226the source code of Microsoft Word can be a trade secret, but if you
227find it out for yourself (by performing arithmetic manipulation of
228other numbers issued by Microsoft, for example, which is known as
229"reverse engineering"), you're not going to be punished, at least if
230you live in some parts of the United States.</para>
232<para>This droid, like other droids, is often right. The condition of
233being a droid is to know everything about something and nothing about
234anything else. By its timely and urgent intervention the droid has
235established that the current intellectual property system contains
236many intricate and ingenious features. The complexities combine to
237allow professors to be erudite, Congressmen to get campaign
238contributions, lawyers to wear nice suits and tassel loafers, and
239Murdoch to be rich.  The complexities mostly evolved in an age of
240industrial information distribution, when information was inscribed in
241analog forms on physical objects that cost something significant to
242make, move, and sell. When applied to digital information that moves
243frictionlessly through the network and has zero marginal cost per
244copy, everything still works, mostly, as long as you don't stop
247<para>But that wasn't what I was arguing about. I wanted to point out
248something else: that our world consists increasingly of nothing but
249large numbers (also known as bitstreams), and that - for reasons
250having nothing to do with emergent properties of the numbers
251themselves - the legal system is presently committed to treating
252similar numbers radically differently. No one can tell, simply by
253looking at a number that is 100 million digits long, whether that
254number is subject to patent, copyright, or trade secret protection, or
255indeed whether it is "owned" by anyone at all. So the legal system we
256have - blessed as we are by its consequences if we are copyright
257teachers, Congressmen, Gucci-gulchers or Big Rupert himself - is
258compelled to treat indistinguishable things in unlike ways.</para>
260<para>Now, in my role as a legal historian concerned with the secular
261(that is, very long term) development of legal thought, I claim that
262legal regimes based on sharp but unpredictable distinctions among
263similar objects are radically unstable. They fall apart over time
264because every instance of the rules' application is an invitation to
265at least one side to claim that instead of fitting in ideal category A
266the particular object in dispute should be deemed to fit instead in
267category B, where the rules will be more favorable to the party making
268the claim. This game - about whether a typewriter should be deemed a
269musical instrument for purposes of railway rate regulation, or whether
270a steam shovel is a motor vehicle - is the frequent stuff of legal
271ingenuity. But when the conventionally-approved legal categories
272require judges to distinguish among the identical, the game is
273infinitely lengthy, infinitely costly, and almost infinitely offensive
274to the unbiased bystander <footnote><para>8. This is not an insight
275unique to our present enterprise. A closely-related idea forms one of
276the most important principles in the history of Anglo-American law,
277perfectly put by Toby Milsom in the following terms:</para>
278<blockquote><para>The life of the common law has been in the abuse of
279its elementary ideas. If the rules of property give what now seems an
280unjust answer, try obligation; and equity has proved that from the
281materials of obligation you can counterfeit the phenomena of
282property. If the rules of contract give what now seems an unjust
283answer, try tort. ... If the rules of one tort, say deceit, give what
284now seems an unjust answer, try another, try negligence. And so the
285legal world goes round.</para></blockquote><para>S.F.C. Milsom,
2861981. <emphasis>Historical Foundations of the Common Law.</emphasis>
287Second edition. London: Butterworths, p. 6.</para> </footnote>.</para>
289<para>Thus parties can spend all the money they want on all the
290legislators and judges they can afford - which for the new "owners" of
291the digital world is quite a few - but the rules they buy aren't going
292to work in the end. Sooner or later, the paradigms are going to
293collapse. Of course, if later means two generations from now, the
294distribution of wealth and power sanctified in the meantime may not be
295reversible by any course less drastic than a <emphasis>bellum
296servile</emphasis> of couch potatoes against media magnates. So
297knowing that history isn't on Bill Gates' side isn't enough. We are
298predicting the future in a very limited sense: we know that the
299existing rules, which have yet the fervor of conventional belief
300solidly enlisted behind them, are no longer meaningful. Parties will
301use and abuse them freely until the mainstream of "respectable"
302conservative opinion acknowledges their death, with uncertain
303results. But realistic scholarship should already be turning its
304attention to the clear need for new thoughtways.</para>
306<para>When we reach this point in the argument, we find ourselves
307contending with the other primary protagonist of educated idiocy: the
308econodwarf. Like the IPdroid, the econodwarf is a species of hedgehog,
309<footnote><para>9. <emphasis>See</emphasis> Isaiah Berlin,
3101953. <emphasis>The Hedgehog and the Fox: An Essay on Tolstoy's View
311of History.</emphasis> New York: Simon and Schuster.</para>
312</footnote> but where the droid is committed to logic over experience,
313the econodwarf specializes in an energetic and well-focused but
314entirely erroneous view of human nature. According to the econodwarf's
315vision, each human being is an individual possessing "incentives,"
316which can be retrospectively unearthed by imagining the state of the
317bank account at various times.  So in this instance the econodwarf
318feels compelled to object that without the rules I am lampooning,
319there would be no incentive to create the things the rules treat as
320property: without the ability to exclude others from music there would
321be no music, because no one could be sure of getting paid for creating
324<para>Music is not really our subject; the software I am considering
325at the moment is the old kind: computer programs. But as he is
326determined to deal at least cursorily with the subject, and because,
327as we have seen, it is no longer really possible to distinguish
328computer programs from music performances, a word or two should be
329said. At least we can have the satisfaction of indulging in an
330argument <emphasis>ad pygmeam</emphasis>.  When the econodwarf grows
331rich, in my experience, he attends the opera.  But no matter how often
332he hears <emphasis>Don Giovanni</emphasis> it never occurs to him that
333Mozart's fate should, on his logic, have entirely discouraged
334Beethoven, or that we have <emphasis>The Magic Flute</emphasis> even
335though Mozart knew very well he wouldn't be paid. In fact,
336<emphasis>The Magic Flute</emphasis>, <emphasis>St. Matthew's
337Passion</emphasis>, and the motets of the wife-murderer Carlo Gesualdo
338are all part of the centuries-long tradition of free software, in the
339more general sense, which the econodwarf never quite
340acknowledges.</para> <!--<center><img
341src="anarchism_files/mog1.gif"></center> --> <para> The dwarf's basic
342problem is that "incentives" is merely a metaphor, and as a metaphor
343to describe human creative activity it's pretty crummy. I have said
344this before, <footnote> <para>10. <emphasis>See</emphasis> <ulink
346Virtual Scholar and Network Liberation.</ulink></para> </footnote> but
347the better metaphor arose on the day Michael Faraday first noticed
348what happened when he wrapped a coil of wire around a magnet and spun
349the magnet. Current flows in such a wire, but we don't ask what the
350incentive is for the electrons to leave home. We say that the current
351results from an emergent property of the system, which we call
352induction. The question we ask is "what's the resistance of the wire?"
353So Moglen's Metaphorical Corollary to Faraday's Law says that if you
354wrap the Internet around every person on the planet and spin the
355planet, software flows in the network. It's an emergent property of
356connected human minds that they create things for one another's
357pleasure and to conquer their uneasy sense of being too alone. The
358only question to ask is, what's the resistance of the network?
359Moglen's Metaphorical Corollary to Ohm's Law states that the
360resistance of the network is directly proportional to the field
361strength of the "intellectual property" system. So the right answer to
362the econodwarf is, resist the resistance.</para>
364<para>Of course, this is all very well in theory. "Resist the
365resistance" sounds good, but we'd have a serious problem, theory
366notwithstanding, if the dwarf were right and we found ourselves
367under-producing good software because we didn't let people own it. But
368dwarves and droids are formalists of different kinds, and the
369advantage of realism is that if you start from the facts the facts are
370always on your side. It turns out that treating software as property
371makes bad software.</para>
375<title>II. Software as Property: The Practical Problem</title>
377<para>In order to understand why turning software into property
378produces bad software, we need an introduction to the history of the
379art. In fact, we'd better start with the word "art" itself. The
380programming of computers combines determinate reasoning with literary
383<para>At first glance, to be sure, source code appears to be a
384non-literary form of composition <footnote><para>11. Some basic
385vocabulary is essential. Digital computers actually execute numerical
386instructions: bitstrings that contain information in the "native"
387language created by the machine's designers. This is usually referred
388to as "machine language." The machine languages of hardware are
389designed for speed of execution at the hardware level, and are not
390suitable for direct use by human beings. So among the central
391components of a computer system are "programming languages," which
392translate expressions convenient for humans into machine language. The
393most common and relevant, but by no means the only, form of computer
394language is a "compiler." The compiler performs static translation, so
395that a file containing human-readable instructions, known as "source
396code" results in the generation of one or more files of executable
397machine language, known as "object code."</para> </footnote>.  The
398primary desideratum in a computer program is that it works, that is to
399say, performs according to specifications formally describing its
400outputs in terms of its inputs. At this level of generality, the
401functional content of programs is all that can be seen.</para>
403<para>But working computer programs exist as parts of computer
404systems, which are interacting collections of hardware, software, and
405human beings. The human components of a computer system include not
406only the users, but also the (potentially different) persons who
407maintain and improve the system. Source code not only communicates
408with the computer that executes the program, through the intermediary
409of the compiler that produces machine-language object code, but also
410with other programmers.</para>
412<para>The function of source code in relation to other human beings is
413not widely grasped by non-programmers, who tend to think of computer
414programs as incomprehensible. They would be surprised to learn that
415the bulk of information contained in most programs is, from the point
416of view of the compiler or other language processor, "comment," that
417is, non-functional material. The comments, of course, are addressed to
418others who may need to fix a problem or to alter or enhance the
419program's operation. In most programming languages, far more space is
420spent in telling people what the program does than in telling the
421computer how to do it.</para>
423<para>The design of programming languages has always proceeded under
424the dual requirements of complete specification for machine execution
425and informative description for human readers. One might identify
426three basic strategies in language design for approaching this dual
427purpose.  The first, pursued initially with respect to the design of
428languages specific to particular hardware products and collectively
429known as "assemblers," essentially separated the human- and
430machine-communication portions of the program. Assembler instructions
431are very close relatives of machine-language instructions: in general,
432one line of an assembler program corresponds to one instruction in the
433native language of the machine. The programmer controls machine
434operation at the most specific possible level, and (if
435well-disciplined) engages in running commentary alongside the machine
436instructions, pausing every few hundred instructions to create "block
437comments," which provide a summary of the strategy of the program, or
438document the major data structures the program manipulates.</para>
440<para>A second approach, characteristically depicted by the language
441COBOL (which stood for "Common Business-Oriented Language"), was to
442make the program itself look like a set of natural language
443directions, written in a crabbed but theoretically human-readable
444style. A line of COBOL code might say, for example "MULTIPLY PRICE
445TIMES QUANTITY GIVING EXPANSION." At first, when the Pentagon and
446industry experts began the joint design of COBOL in the early 1960's,
447this seemed a promising approach. COBOL programs appeared largely
448self-documenting, allowing both the development of work teams able to
449collaborate on the creation of large programs, and the training of
450programmers who, while specialized workers, would not need to
451understand the machine as intimately as assembler programs had to. But
452the level of generality at which such programs documented themselves
453was wrongly selected. A more formulaic and compressed expression of
454operational detail "expansion = price x quantity," for example, was
455better suited even to business and financial applications where the
456readers and writers of programs were accustomed to mathematical
457expression, while the processes of describing both data structures and
458the larger operational context of the program were not rendered
459unnecessary by the wordiness of the language in which the details of
460execution were specified.</para>
462<para>Accordingly, language designers by the late 1960s began
463experimenting with forms of expression in which the blending of
464operational details and non-functional information necessary for
465modification or repair was more subtle. Some designers chose the path
466of highly symbolic and compressed languages, in which the programmer
467manipulated data abstractly, so that "A x B" might mean the
468multiplication of two integers, two complex numbers, two vast arrays,
469or any other data type capable of some process called
470"multiplication," to be undertaken by the computer on the basis of the
471context for the variables "A" and "B" at the moment of execution
472<footnote> <para>12. This, I should say, was the path that most of my
473research and development followed, largely in connection with a
474language called APL ("A Programming Language") and its successors. It
475was not, however, the ultimately-dominant approach, for reasons that
476will be suggested below.</para> </footnote> .  Because this approach
477resulted in extremely concise programs, it was thought, the problem of
478making code comprehensible to those who would later seek to modify or
479repair it was simplified. By hiding the technical detail of computer
480operation and emphasizing the algorithm, languages could be devised
481that were better than English or other natural languages for the
482expression of stepwise processes. Commentary would be not only
483unnecessary but distracting, just as the metaphors used to convey
484mathematical concepts in English do more to confuse than to
488<title>How We Created the Microbrain Mess</title>
490<para>Thus the history of programming languages directly reflected the
491need to find forms of human-machine communication that were also
492effective in conveying complex ideas to human readers. "Expressivity"
493became a property of programming languages, not because it facilitated
494computation, but because it facilitated the collaborative creation and
495maintenance of increasingly complex software systems.</para>
497<para>At first impression, this seems to justify the application of
498traditional copyright thinking to the resulting works. Though
499substantially involving "functional" elements, computer programs
500contained "expressive" features of paramount importance. Copyright
501doctrine recognized the merger of function and expression as
502characteristic of many kinds of copyrighted works. "Source code,"
503containing both the machine instructions necessary for functional
504operation and the expressive "commentary" intended for human readers,
505was an appropriate candidate for copyright treatment.</para>
507<para>True, so long as it is understood that the expressive component
508of software was present solely in order to facilitate the making of
509"derivative works." Were it not for the intention to facilitate
510alteration, the expressive elements of programs would be entirely
511supererogatory, and source code would be no more copyrightable than
512object code, the output of the language processor, purged of all but
513the program's functional characteristics.</para>
515<para>The state of the computer industry throughout the 1960's and
5161970's, when the grundnorms of sophisticated computer programming were
517established, concealed the tension implicit in this situation. In that
518period, hardware was expensive. Computers were increasingly large and
519complex collections of machines, and the business of designing and
520building such an array of machines for general use was dominated, not
521to say monopolized, by one firm. IBM gave away its software. To be
522sure, it owned the programs its employees wrote, and it copyrighted
523the source code. But it also distributed the programs - including the
524source code - to its customers at no additional charge, and encouraged
525them to make and share improvements or adaptations of the programs
526thus distributed. For a dominant hardware manufacturer, this strategy
527made sense: better programs sold more computers, which is where the
528profitability of the business rested.</para>
530<para>Computers, in this period, tended to aggregate within particular
531organizations, but not to communicate broadly with one another. The
532software needed to operate was distributed not through a network, but
533on spools of magnetic tape. This distribution system tended to
534centralize software development, so that while IBM customers were free
535to make modifications and improvements to programs, those
536modifications were shared in the first instance with IBM, which then
537considered whether and in what way to incorporate those changes in the
538centrally-developed and distributed version of the software. Thus in
539two important senses the best computer software in the world was free:
540it cost nothing to acquire, and the terms on which it was furnished
541both allowed and encouraged experimentation, change, and improvement
542<footnote><para>13. This description elides some details. By the
543mid-1970's IBM had acquired meaningful competition in the mainframe
544computer business, while the large-scale antitrust action brought
545against it by the U.S. government prompted the decision to "unbundle,"
546or charge separately, for software. In this less important sense,
547software ceased to be free. But - without entering into the now-dead
548but once-heated controversy over IBM's software pricing policies - the
549unbundling revolution had less effect on the social practices of
550software manufacture than might be supposed. As a fellow responsible
551for technical improvement of one programming language product at IBM
552from 1979 to 1984, for example, I was able to treat the product as
553"almost free," that is, to discuss with users the changes they had
554proposed or made in the programs, and to engage with them in
555cooperative development of the product for the benefit of all
556users.</para> </footnote>.  That the software in question was IBM's
557property under prevailing copyright law certainly established some
558theoretical limits on users' ability to distribute their improvements
559or adaptations to others, but in practice mainframe software was
560cooperatively developed by the dominant hardware manufacturer and its
561technically-sophisticated users, employing the manufacturer's
562distribution resources to propagate the resulting improvements through
563the user community. The right to exclude others, one of the most
564important "sticks in the bundle" of property rights (in an image
565beloved of the United States Supreme Court), was practically
566unimportant, or even undesirable, at the heart of the software
567business <footnote> <para>14. This description is highly compressed,
568and will seem both overly simplified and unduly rosy to those who also
569worked in the industry during this period of its
570development. Copyright protection of computer software was a
571controversial subject in the 1970's, leading to the famous CONTU
572commission and its mildly pro-copyright recommendations of 1979. And
573IBM seemed far less cooperative to its users at the time than this
574sketch makes out. But the most important element is the contrast with
575the world created by the PC, the Internet, and the dominance of
576Microsoft, with the resulting impetus for the free software movement,
577and I am here concentrating on the features that express that
580<para>After 1980, everything was different. The world of mainframe
581hardware gave way within ten years to the world of the commodity PC.
582And, as a contingency of the industry's development, the single most
583important element of the software running on that commodity PC, the
584operating system, became the sole significant product of a company
585that made no hardware. High-quality basic software ceased to be part
586of the product-differentiation strategy of hardware
587manufacturers. Instead, a firm with an overwhelming share of the
588market, and with the near-monopolist's ordinary absence of interest in
589fostering diversity, set the practices of the software industry. In
590such a context, the right to exclude others from participation in the
591product's formation became profoundly important. Microsoft's power in
592the market rested entirely on its ownership of the Windows source
595<para>To Microsoft, others' making of "derivative works," otherwise
596known as repairs and improvements, threatened the central asset of the
597business. Indeed, as subsequent judicial proceedings have tended to
598establish, Microsoft's strategy as a business was to find innovative
599ideas elsewhere in the software marketplace, buy them up and either
600suppress them or incorporate them in its proprietary product. The
601maintenance of control over the basic operation of computers
602manufactured, sold, possessed, and used by others represented profound
603and profitable leverage over the development of the culture <footnote>
604<para>15. I discuss the importance of PC software in this context, the
605evolution of "the market for eyeballs" and "the sponsored life" in
606other chapters of my forthcoming book, <emphasis>The Invisible
607Barbecue</emphasis>, of which this essay forms a part.</para>
608</footnote>.; the right to exclude returned to center stage in the
609concept of software as property.</para>
611<para>The result, so far as the quality of software was concerned, was
612disastrous. The monopoly was a wealthy and powerful corporation that
613employed a large number of programmers, but it could not possibly
614afford the number of testers, designers, and developers required to
615produce flexible, robust and technically-innovative software
616appropriate to the vast array of conditions under which increasingly
617ubiquitous personal computers operated. Its fundamental marketing
618strategy involved designing its product for the least
619technically-sophisticated users, and using "fear, uncertainty, and
620doubt" (known within Microsoft as "FUD") to drive sophisticated users
621away from potential competitors, whose long-term survivability in the
622face of Microsoft's market power was always in question.</para>
624<para>Without the constant interaction between users able to repair
625and improve and the operating system's manufacturer, the inevitable
626deterioration of quality could not be arrested. But because the
627personal computer revolution expanded the number of users
628exponentially, almost everyone who came in contact with the resulting
629systems had nothing against which to compare them. Unaware of the
630standards of stability, reliability, maintainability and effectiveness
631that had previously been established in the mainframe world, users of
632personal computers could hardly be expected to understand how badly,
633in relative terms, the monopoly's software functioned. As the power
634and capacity of personal computers expanded rapidly, the defects of
635the software were rendered less obvious amidst the general increase of
636productivity. Ordinary users, more than half afraid of the technology
637they almost completely did not understand, actually welcomed the
638defectiveness of the software. In an economy undergoing mysterious
639transformations, with the concomitant destabilization of millions of
640careers, it was tranquilizing, in a perverse way, that no personal
641computer seemed to be able to run for more than a few consecutive
642hours without crashing. Although it was frustrating to lose work in
643progress each time an unnecessary failure occurred, the evident
644fallibility of computers was intrinsically reassuring <footnote>
645<para>16. This same pattern of ambivalence, in which bad programming
646leading to widespread instability in the new technology is
647simultaneously frightening and reassuring to technical incompetents,
648can be seen also in the primarily-American phenomenon of Y2K
649hysteria.</para> </footnote> .</para>
651<para>None of this was necessary. The low quality of personal computer
652software could have been reversed by including users directly in the
653inherently evolutionary process of software design and implementation.
654A Lamarckian mode, in which improvements could be made anywhere, by
655anyone, and inherited by everyone else, would have wiped out the
656deficit, restoring to the world of the PC the stability and
657reliability of the software made in the quasi-propertarian environment
658of the mainframe era. But the Microsoft business model precluded
659Lamarckian inheritance of software improvements. Copyright doctrine,
660in general and as it applies to software in particular, biases the
661world towards creationism; in this instance, the problem is that BillG
662the Creator was far from infallible, and in fact he wasn't even
663trying.</para> <!--<center><img src="anarchism_files/mog2.gif"
664hspace="0" vspace="0"></center>--> <para>To make the irony more
665severe, the growth of the network rendered the non-propertarian
666alternative even more practical. What scholarly and popular writing
667alike denominate as a thing ("the Internet") is actually the name of a
668social condition: the fact that everyone in the network society is
669connected directly, without intermediation, to everyone else
670<footnote> <para>17. The critical implications of this simple
671observation about our metaphors are worked out in "How Not to Think
672about 'The Internet'," in <emphasis>The Invisible Barbecue</emphasis>,
673forthcoming.</para> </footnote>. The global interconnection of
674networks eliminated the bottleneck that had required a centralized
675software manufacturer to rationalize and distribute the outcome of
676individual innovation in the era of the mainframe.</para>
678<para>And so, in one of history's little ironies, the global triumph
679of bad software in the age of the PC was reversed by a surprising
680combination of forces: the social transformation initiated by the
681network, a long-discarded European theory of political economy, and a
682small band of programmers throughout the world mobilized by a single
683simple idea.</para>
688<title>Software Wants to Be Free; or, How We Stopped Worrying and
689Learned to Love the Bomb</title>
691<para>Long before the network of networks was a practical reality,
692even before it was an aspiration, there was a desire for computers to
693operate on the basis of software freely available to everyone. This
694began as a reaction against propertarian software in the mainframe
695era, and requires another brief historical digression.</para>
697<para>Even though IBM was the largest seller of general purpose
698computers in the mainframe era, it was not the largest designer and
699builder of such hardware. The telephone monopoly, American Telephone
700&amp; Telegraph, was in fact larger than IBM, but it consumed its
701products internally. And at the famous Bell Labs research arm of the
702telephone monopoly, in the late 1960's, the developments in computer
703languages previously described gave birth to an operating system
704called Unix.</para>
706<para>The idea of Unix was to create a single, scalable operating
707system to exist on all the computers, from small to large, that the
708telephone monopoly made for itself. To achieve this goal meant writing
709an operating system not in machine language, nor in an assembler whose
710linguistic form was integral to a particular hardware design, but in a
711more expressive and generalized language. The one chosen was also a
712Bell Labs invention, called "C" <footnote> <para>18. Technical readers
713will again observe that this compresses developments occurring from
7141969 through 1973.</para> </footnote>. The C language became common,
715even dominant, for many kinds of programming tasks, and by the late
7161970's the Unix operating system written in that language had been
717transferred (or "ported," in professional jargon) to computers made by
718many manufacturers and of many designs.</para>
720<para>AT&amp;T distributed Unix widely, and because of the very design
721of the operating system, it had to make that distribution in C source
722code.  But AT&amp;T retained ownership of the source code and
723compelled users to purchase licenses that prohibited redistribution
724and the making of derivative works. Large computing centers, whether
725industrial or academic, could afford to purchase such licenses, but
726individuals could not, while the license restrictions prevented the
727community of programmers who used Unix from improving it in an
728evolutionary rather than episodic fashion. And as programmers
729throughout the world began to aspire to and even expect a personal
730computer revolution, the "unfree" status of Unix became a source of
733<para>Between 1981 and 1984, one man envisioned a crusade to change
734the situation. Richard M. Stallman, then an employee of MIT's
735Artificial Intelligence Laboratory, conceived the project of
736independent, collaborative redesign and implementation of an operating
737system that would be true free software. In Stallman's phrase, free
738software would be a matter of freedom, not of price. Anyone could
739freely modify and redistribute such software, or sell it, subject only
740to the restriction that he not try to reduce the rights of others to
741whom he passed it along. In this way free software could become a
742self-organizing project, in which no innovation would be lost through
743proprietary exercises of rights. The system, Stallman decided, would
744be called GNU, which stood (in an initial example of a taste for
745recursive acronyms that has characterized free software ever since),
746for "GNU's Not Unix."  Despite misgivings about the fundamental design
747of Unix, as well as its terms of distribution, GNU was intended to
748benefit from the wide if unfree source distribution of Unix. Stallman
749began Project GNU by writing components of the eventual system that
750were also designed to work without modification on existing Unix
751systems. Development of the GNU tools could thus proceed directly in
752the environment of university and other advanced computing centers
753around the world.</para>
755<para>The scale of such a project was immense. Somehow, volunteer
756programmers had to be found, organized, and set to work building all
757the tools that would be necessary for the ultimate construction.
758Stallman himself was the primary author of several fundamental tools.
759Others were contributed by small or large teams of programmers
760elsewhere, and assigned to Stallman's project or distributed
761directly. A few locations around the developing network became
762archives for the source code of these GNU components, and throughout
763the 1980's the GNU tools gained recognition and acceptance by Unix
764users throughout the world. The stability, reliability, and
765maintainability of the GNU tools became a by-word, while Stallman's
766profound abilities as a designer continued to outpace, and provide
767goals for, the evolving process. The award to Stallman of a MacArthur
768Fellowship in 1990 was an appropriate recognition of his conceptual
769and technical innovations and their social consequences.</para>
771<para>Project GNU, and the Free Software Foundation to which it gave
772birth in 1985, were not the only source of free software
773ideas. Several forms of copyright license designed to foster free or
774partially free software began to develop in the academic community,
775mostly around the Unix environment. The University of California at
776Berkeley began the design and implementation of another version of
777Unix for free distribution in the academic community. BSD Unix, as it
778came to be known, also treated AT&amp;T's Unix as a design
779standard. The code was broadly released and constituted a reservoir of
780tools and techniques, but its license terms limited the range of its
781application, while the elimination of hardware-specific proprietary
782code from the distribution meant that no one could actually build a
783working operating system for any particular computer from BSD. Other
784university-based work also eventuated in quasi-free software; the
785graphical user interface (or GUI) for Unix systems called X Windows,
786for example, was created at MIT and distributed with source code on
787terms permitting free modification. And in 1989-1990, an undergraduate
788computer science student at the University of Helsinki, Linus
789Torvalds, began the project that completed the circuit and fully
790energized the free software vision.</para>
792<para>What Torvalds did was to begin adapting a computer science
793teaching tool for real life use. Andrew Tannenbaum's MINIX kernel
794<footnote> <para>19. Operating systems, even Windows (which hides the
795fact from its users as thoroughly as possible), are actually
796collections of components, rather than undivided unities. Most of what
797an operating system does (manage file systems, control process
798execution, etc.) can be abstracted from the actual details of the
799computer hardware on which the operating system runs. Only a small
800inner core of the system must actually deal with the eccentric
801peculiarities of particular hardware.  Once the operating system is
802written in a general language such as C, only that inner core, known
803in the trade as the kernel, will be highly specific to a particular
804computer architecture.</para> </footnote> , was a staple of Operating
805Systems courses, providing an example of basic solutions to basic
806problems. Slowly, and at first without recognizing the intention,
807Linus began turning the MINIX kernel into an actual kernel for Unix on
808the Intel x86 processors, the engines that run the world's commodity
809PCs. As Linus began developing this kernel, which he named Linux, he
810realized that the best way to make his project work would be to adjust
811his design decisions so that the existing GNU components would be
812compatible with his kernel.</para>
814<para>The result of Torvalds' work was the release on the net in 1991
815of a sketchy working model of a free software kernel for a Unix-like
816operating system for PCs, fully compatible with and designed
817convergently with the large and high-quality suite of system
818components created by Stallman's Project GNU and distributed by the
819Free Software Foundation. Because Torvalds chose to release the Linux
820kernel under the Free Software Foundation's General Public License, of
821which more below, the hundreds and eventually thousands of programmers
822around the world who chose to contribute their effort towards the
823further development of the kernel could be sure that their efforts
824would result in permanently free software that no one could turn into
825a proprietary product. Everyone knew that everyone else would be able
826to test, improve, and redistribute their improvements. Torvalds
827accepted contributions freely, and with a genially effective style
828maintained overall direction without dampening enthusiasm. The
829development of the Linux kernel proved that the Internet made it
830possible to aggregate collections of programmers far larger than any
831commercial manufacturer could afford, joined almost non-hierarchically
832in a development project ultimately involving more than one million
833lines of computer code - a scale of collaboration among geographically
834dispersed unpaid volunteers previously unimaginable in human history
835<footnote> <para>20. A careful and creative analysis of how Torvalds
836made this process work, and what it implies for the social practices
837of creating software, was provided by Eric S. Raymond in his seminal
8381997 paper, <ulink
840Cathedral and the Bazaar,</ulink> which itself played a significant
841role in the expansion of the free software idea.</para>
844<para>By 1994, Linux had reached version 1.0, representing a usable
845production kernel. Level 2.0 was reached in 1996, and by 1998, with
846the kernel at 2.2.0 and available not only for x86 machines but for a
847variety of other machine architectures, GNU/Linux - the combination of
848the Linux kernel and the much larger body of Project GNU components -
849and Windows NT were the only two operating systems in the world
850gaining market share. A Microsoft internal assessment of the situation
851leaked in October 1998 and subsequently acknowledged by the company as
852genuine concluded that "Linux represents a best-of-breed UNIX, that is
853trusted in mission critical applications, and - due to it's [sic] open
854source code - has a long term credibility which exceeds many other
855competitive OS's." <footnote> <para>21. This is a quotation from what
856is known in the trade as the "Halloween memo," which can be found, as
857annotated by Eric Raymond, to whom it was leaked, at <ulink
860GNU/Linux systems are now used throughout the world, operating
861everything from Web servers at major electronic commerce sites to
862"ad-hoc supercomputer" clusters to the network infrastructure of
863money-center banks. GNU/Linux is found on the space shuttle, and
864running behind-the-scenes computers at (yes) Microsoft. Industry
865evaluations of the comparative reliability of Unix systems have
866repeatedly shown that Linux is far and away the most stable and
867reliable Unix kernel, with a reliability exceeded only by the GNU
868tools themselves. GNU/Linux not only out-performs commercial
869proprietary Unix versions for PCs in benchmarks, but is renowned for
870its ability to run, undisturbed and uncomplaining, for months on end
871in high-volume high-stress environments without crashing.</para>
873<para>Other components of the free software movement have been equally
874successful. Apache, far and away the world's leading Web server
875program, is free software, as is Perl, the programming language which
876is the lingua franca for the programmers who build sophisticated Web
877sites. Netscape Communications now distributes its Netscape
878Communicator 5.0 browser as free software, under a close variant of
879the Free Software Foundation's General Public License. Major PC
880manufacturers, including IBM, have announced plans or are already
881distributing GNU/Linux as a customer option on their top-of-the-line
882PCs intended for use as Web- and file servers. Samba, a program that
883allows GNU/Linux computers to act as Windows NT file servers, is used
884worldwide as an alternative to Windows NT Server, and provides
885effective low-end competition to Microsoft in its own home market. By
886the standards of software quality that have been recognized in the
887industry for decades - and whose continuing relevance will be clear to
888you the next time your Windows PC crashes - the news at century's end
889is unambiguous. The world's most profitable and powerful corporation
890comes in a distant second, having excluded all but the real victor
891from the race. Propertarianism joined to capitalist vigor destroyed
892meaningful commercial competition, but when it came to making good
893software, anarchism won.</para>
898<!--<para><img src="anarchism_files/quad.gif"></para><a name="m3"></a>-->
900<title>III. Anarchism as a Mode of Production</title>
902<para>It's a pretty story, and if only the IPdroid and the econodwarf
903hadn't been blinded by theory, they'd have seen it coming. But though
904some of us had been working for it and predicting it for years, the
905theoretical consequences are so subversive for the thoughtways that
906maintain our dwarves and droids in comfort that they can hardly be
907blamed for refusing to see. The facts proved that something was wrong
908with the "incentives" metaphor that underprops conventional
909intellectual property reasoning <footnote> <para>22. As recently as
910early 1994 a talented and technically competent (though Windows-using)
911law and economics scholar at a major U.S. law school confidently
912informed me that free software couldn't possibly exist, because no one
913would have any incentive to make really sophisticated programs
914requiring substantial investment of effort only to give them
915away.</para> </footnote> . But they did more. They provided an initial
916glimpse into the future of human creativity in a world of global
917interconnection, and it's not a world made for dwarves and
920<para>My argument, before we paused for refreshment in the real world,
921can be summarized this way: Software - whether executable programs,
922music, visual art, liturgy, weaponry, or what have you - consists of
923bitstreams, which although essentially indistinguishable are treated
924by a confusing multiplicity of legal categories. This multiplicity is
925unstable in the long term for reasons integral to the legal process.
926The unstable diversity of rules is caused by the need to distinguish
927among kinds of property interests in bitstreams. This need is
928primarily felt by those who stand to profit from the socially
929acceptable forms of monopoly created by treating ideas as
930property. Those of us who are worried about the social inequity and
931cultural hegemony created by this intellectually unsatisfying and
932morally repugnant regime are shouted down. Those doing the shouting,
933the dwarves and the droids, believe that these property rules are
934necessary not from any overt yearning for life in Murdochworld -
935though a little luxurious co-optation is always welcome - but because
936the metaphor of incentives, which they take to be not just an image
937but an argument, proves that these rules - despite their lamentable
938consequences - are necessary if we are to make good software. The only
939way to continue to believe this is to ignore the facts. At the center
940of the digital revolution, with the executable bitstreams that make
941everything else possible, propertarian regimes not only do not make
942things better, they can make things radically worse. Property
943concepts, whatever else may be wrong with them, do not enable and have
944in fact retarded progress.</para>
947But what is this mysterious alternative? Free software exists, but
948what are its mechanisms, and how does it generalize towards a
949non-propertarian theory of the digital society?</para>
954<title>The Legal Theory of Free Software</title>
956<para>There is a myth, like most myths partially founded on reality,
957that computer programmers are all libertarians. Right-wing ones are
958capitalists, cleave to their stock options, and disdain taxes, unions,
959and civil rights laws; left-wing ones hate the market and all
960government, believe in strong encryption no matter how much nuclear
961terrorism it may cause, <footnote> <para>23. This question too
962deserves special scrutiny, encrusted as it is with special pleading on
963the state-power side. See my brief essay <ulink
965Much for Savages</emphasis>: Navajo 1, Government 0 in Final Moments of
966Play."</ulink></para> </footnote> and dislike Bill Gates because he's
967rich. There is doubtless a foundation for this belief. But the most
968significant difference between political thought inside the digirati
969and outside it is that in the network society, anarchism (or more
970properly, anti-possessive individualism) is a viable political
973<para>The center of the free software movement's success, and the
974greatest achievement of Richard Stallman, is not a piece of computer
975code. The success of free software, including the overwhelming success
976of GNU/Linux, results from the ability to harness extraordinary
977quantities of high-quality effort for projects of immense size and
978profound complexity. And this ability in turn results from the legal
979context in which the labor is mobilized. As a visionary designer
980Richard Stallman created more than Emacs, GDB, or GNU. He created the
981General Public License.</para>
983<!-- <center><img src="anarchism_files/mog3.gif" hspace="0"
984vspace="0"></center> --> <para>The GPL, <footnote>
985<para>24. <emphasis>See</emphasis> <ulink
986url="">GNU General Public License,
987Version 2, June 1991.</ulink></para> </footnote> also known as the
988copyleft, uses copyright, to paraphrase Toby Milsom, to counterfeit
989the phenomena of anarchism. As the license preamble expresses
992<blockquote><para>When we speak of free software, we are referring to
993freedom, not price. Our General Public Licenses are designed to make
994sure that you have the freedom to distribute copies of free software
995(and charge for this service if you wish), that you receive source
996code or can get it if you want it, that you can change the software or
997use pieces of it in new free programs; and that you know you can do
998these things.</para>
1000<para>To protect your rights, we need to make restrictions that
1001forbid anyone to deny you these rights or to ask you to surrender the
1002rights.  These restrictions translate to certain responsibilities for
1003you if you distribute copies of the software, or if you modify
1006<para>For example, if you distribute copies of such a program,
1007whether gratis or for a fee, you must give the recipients all the
1008rights that you have. You must make sure that they, too, receive or
1009can get the source code. And you must show them these terms so they
1010know their rights.</para>
1012<para>Many variants of this basic free software idea have been
1013expressed in licenses of various kinds, as I have already
1014indicated. The GPL is different from the other ways of expressing
1015these values in one crucial respect. Section 2 of the license provides
1016in pertinent part:</para>
1018<para>You may modify your copy or copies of the Program or any
1019portion of it, thus forming a work based on the Program, and copy and
1020distribute such modifications or work ..., provided that you also meet
1021all of these conditions: </para>
1025<para>b) You must cause any work that you distribute or publish,
1026that in whole or in part contains or is derived from the Program or
1027any part thereof, to be licensed as a whole at no charge to all third
1028parties under the terms of this License.</para></blockquote>
1030<para>Section 2(b) of the GPL is sometimes called "restrictive," but
1031its intention is liberating. It creates a commons, to which anyone may
1032add but from which no one may subtract. Because of §2(b), each
1033contributor to a GPL'd project is assured that she, and all other
1034users, will be able to run, modify and redistribute the program
1035indefinitely, that source code will always be available, and that,
1036unlike commercial software, its longevity cannot be limited by the
1037contingencies of the marketplace or the decisions of future
1038developers. This "inheritance" of the GPL has sometimes been
1039criticized as an example of the free software movement's
1040anti-commercial bias.  Nothing could be further from the truth. The
1041effect of §2(b) is to make commercial distributors of free software
1042better competitors against proprietary software businesses. For
1043confirmation of this point, one can do no better than to ask the
1044proprietary competitors. As the author of the Microsoft "Halloween"
1045memorandum, Vinod Vallopillil, put it:</para>
1047<blockquote><para>The GPL and its aversion to code forking reassures
1048customers that they aren't riding an evolutionary `dead-end' by
1049subscribing to a particular commercial version of Linux.</para>
1051<para>The "evolutionary dead-end" is the core of the software
1052FUD argument <footnote> <para>25. <ulink
1053url="">V. Vallopillil,
1054Open Source Software: A (New?) Development Methodology.</ulink></para>
1055</footnote> .</para></blockquote>
1057<para>Translated out of Microspeak, this means that the strategy by
1058which the dominant proprietary manufacturer drives customers away from
1059competitors - by sowing fear, uncertainty and doubt about other
1060software's long-term viability - is ineffective with respect to GPL'd
1061programs. Users of GPL'd code, including those who purchase software
1062and systems from a commercial reseller, know that future improvements
1063and repairs will be accessible from the commons, and need not fear
1064either the disappearance of their supplier or that someone will use a
1065particularly attractive improvement or a desperately necessary repair
1066as leverage for "taking the program private."</para>
1068<para>This use of intellectual property rules to create a commons in
1069cyberspace is the central institutional structure enabling the
1070anarchist triumph. Ensuring free access and enabling modification at
1071each stage in the process means that the evolution of software occurs
1072in the fast Lamarckian mode: each favorable acquired characteristic of
1073others' work can be directly inherited. Hence the speed with which the
1074Linux kernel, for example, outgrew all of its proprietary
1075predecessors. Because defection is impossible, free riders are
1076welcome, which resolves one of the central puzzles of collective
1077action in a propertarian social system.</para>
1079<para>Non-propertarian production is also directly responsible for the
1080famous stability and reliability of free software, which arises from
1081what Eric Raymond calls "Linus' law": With enough eyeballs, all bugs
1082are shallow. In practical terms, access to source code means that if I
1083have a problem I can fix it. Because I can fix it, I almost never have
1084to, because someone else has almost always seen it and fixed it
1087<para>For the free software community, commitment to anarchist
1088production may be a moral imperative; as Richard Stallman wrote, it's
1089about freedom, not about price. Or it may be a matter of utility,
1090seeking to produce better software than propertarian modes of work
1091will allow.  From the droid point of view, the copyleft represents the
1092perversion of theory, but better than any other proposal over the past
1093decades it resolves the problems of applying copyright to the
1094inextricably merged functional and expressive features of computer
1095programs. That it produces better software than the alternative does
1096not imply that traditional copyright principles should now be
1097prohibited to those who want to own and market inferior software
1098products, or (more charitably) whose products are too narrow in appeal
1099for communal production. But our story should serve as a warning to
1100droids: The world of the future will bear little relation to the world
1101of the past. The rules are now being bent in two directions. The
1102corporate owners of "cultural icons" and other assets who seek
1103ever-longer terms for corporate authors, converting the "limited Time"
1104of Article I, §8 into a freehold have naturally been whistling music
1105to the android ear <footnote> <para>26. The looming expiration of
1106Mickey Mouse's ownership by Disney requires, from the point of view of
1107that wealthy "campaign contributor," for example, an alteration of the
1108general copyright law of the United States. See "Not Making it Any
1109More?  Vaporizing the Public Domain," in <emphasis>The Invisible
1110Barbecue</emphasis>, forthcoming.</para> </footnote> .  After all, who bought
1111the droids their concert tickets? But as the propertarian position
1112seeks to embed itself ever more strongly, in a conception of copyright
1113liberated from the minor annoyances of limited terms and fair use, at
1114the very center of our "cultural software" system, the anarchist
1115counter-strike has begun. Worse is yet to befall the droids, as we
1116shall see. But first, we must pay our final devoirs to the
1121<title>Because It's There: Faraday's Magnet and Human Creativity</title>
1123<para>After all, they deserve an answer. Why do people make free
1124software if they don't get to profit? Two answers have usually been
1125given. One is half-right and the other is wrong, but both are
1126insufficiently simple.</para>
1128<para>The wrong answer is embedded in numerous references to "the
1129hacker gift-exchange culture." This use of ethnographic jargon
1130wandered into the field some years ago and became rapidly, if
1131misleadingly, ubiquitous. It reminds us only that the
1132economeretricians have so corrupted our thought processes that any
1133form of non-market economic behavior seems equal to every other
1134kind. But gift-exchange, like market barter, is a propertarian
1135institution. Reciprocity is central to these symbolic enactments of
1136mutual dependence, and if either the yams or the fish are
1137short-weighted, trouble results. Free software, at the risk of
1138repetition, is a commons: no reciprocity ritual is enacted there. A
1139few people give away code that others sell, use, change, or borrow
1140wholesale to lift out parts for something else. Notwithstanding the
1141very large number of people (tens of thousands, at most) who have
1142contributed to GNU/Linux, this is orders of magnitude less than the
1143number of users who make no contribution whatever <footnote>
1144<para>27. A recent industry estimate puts the number of Linux systems
1145worldwide at 7.5 million. <emphasis>See</emphasis> Josh McHugh, 1998. <ulink
1146url="">"Linux: The
1147Making of a Global Hack,"</ulink> <emphasis>Forbes</emphasis> (August 10). Because the
1148software is freely obtainable throughout the Net, there is no simple
1149way to assess actual usage.</para> </footnote>.</para>
1151<para>A part of the right answer is suggested by the claim that free
1152software is made by those who seek reputational compensation for their
1153activity. Famous Linux hackers, the theory is, are known all over the
1154planet as programming deities. From this they derive either enhanced
1155self-esteem or indirect material advancement <footnote> <para>28. Eric
1156Raymond is a partisan of the "ego boost" theory, to which he adds
1157another faux-ethnographic comparison, of free software composition to
1158the Kwakiutl potlatch. <emphasis>See</emphasis> Eric S. Raymond, 1998. <ulink
1160the Noosphere.</ulink>.  But the potlatch, certainly a form of status
1161competition, is unlike free software for two fundamental reasons: it
1162is essentially hierarchical, which free software is not, and, as we
1163have known since Thorstein Veblen first called attention to its
1164significance, it is a form of conspicuous waste. <emphasis>See</emphasis> Thorstein
1165Veblen, 1967. <emphasis>The Theory of the Leisure Class.</emphasis> New York:
1166Viking, p. 75. These are precisely the grounds which distinguish the
1167anti-hierarchical and utilitiarian free software culture from its
1168propertarian counterparts.</para></footnote>.  But the programming
1169deities, much as they have contributed to free software, have not done
1170the bulk of the work. Reputations, as Linus Torvalds himself has often
1171pointed out, are made by willingly acknowledging that it was all done
1172by someone else. And, as many observers have noted, the free software
1173movement has also produced superlative
1174documentation. Documentation-writing is not what hackers do to attain
1175cool, and much of the documentation has been written by people who
1176didn't write the code. Nor must we limit the indirect material
1177advantages of authorship to increases in reputational capital.  Most
1178free software authors I know have day jobs in the technology
1179industries, and the skills they hone in the more creative work they do
1180outside the market no doubt sometimes measurably enhance their value
1181within it. And as the free software products gained critical mass and
1182became the basis of a whole new set of business models built around
1183commercial distribution of that which people can also get for nothing,
1184an increasing number of people are specifically employed to write free
1185software. But in order to be employable in the field, they must
1186already have established themselves there. Plainly, then, this motive
1187is present, but it isn't the whole explanation.</para>
1189<para>Indeed, the rest of the answer is just too simple to have
1190received its due. The best way to understand is to follow the brief
1191and otherwise unsung career of an initially-grudging free software
1192author.  Microsoft's Vinod Vallopillil, in the course of writing the
1193competitive analysis of Linux that was leaked as the second of the
1194famous "Halloween memoranda," bought and installed a Linux system on
1195one of his office computers. He had trouble because the (commercial)
1196Linux distribution he installed did not contain a daemon to handle the
1197DHCP protocol for assignment of dynamic IP addresses. The result was
1198important enough for us to risk another prolonged exposure to the
1199Microsoft Writing Style:</para>
1201<blockquote><para>A small number of Web sites and FAQs later, I found an FTP
1202site with a Linux DHCP client. The DHCP client was developed by an
1203engineer employed by Fore Systems (as evidenced by his e-mail address;
1204I believe, however, that it was developed in his own free time). A
1205second set of documentation/manuals was written for the DHCP client by
1206a hacker in <emphasis>Hungary</emphasis> which provided relatively simple
1207instructions on how to install/load the client.</para>
1209<para>I downloaded &amp; uncompressed the client and typed two
1210simple commands:</para>
1212<para>Make - compiles the client binaries</para>
1214<para>Make Install -installed the binaries as a Linux Daemon</para>
1216<para>Typing "DHCPCD" (for DHCP Client Daemon) on the command
1217line triggered the DHCP discovery process and voila, I had IP
1218networking running.  </para>
1220<para>Since I had just downloaded the DHCP client code, on an
1221impulse I played around a bit. Although the client wasn't as
1222extensible as the DHCP client we are shipping in NT5 (for example, it
1223won't query for arbitrary options &amp; store results), it was obvious
1224how I could write the additional code to implement this functionality.
1225The full client consisted of about 2,600 lines of code.</para>
1227<para>One example of esoteric, extended functionality that was
1228clearly patched in by a third party was a set of routines to that
1229would pad the DHCP request with host-specific strings required by
1230Cable Modem / ADSL sites.</para>
1232<para>A few other steps were required to configure the DHCP
1233client to auto-start and auto-configure my Ethernet interface on boot
1234but these were documented in the client code and in the DHCP
1235documentation from the Hungarian developer.</para>
1237<para>I'm a poorly skilled UNIX programmer but it was
1238immediately obvious to me how to incrementally extend the DHCP client
1239code (the feeling was exhilarating and addictive).</para>
1241<para>Additionally, due directly to GPL + having the full development
1242environment in front of me, I was in a position where I could write up
1243my changes and e-mail them out within a couple of hours (in contrast
1244to how things like this would get done in NT). Engaging in that
1245process would have prepared me for a larger, more ambitious Linux
1246project in the future <footnote><para>29. Vinod Vallopillil, <ulink
1247url="">Linux OS
1248Competitive Analysis (Halloween II).</ulink> Note Vallopillil's
1249surprise that a program written in California had been subsequently
1250documented by a programmer in Hungary.</para>
1253<para>"The feeling was exhilarating and addictive." Stop the presses:
1254Microsoft experimentally verifies Moglen's Metaphorical Corollary to
1255Faraday's Law. Wrap the Internet around every brain on the planet and
1256spin the planet. Software flows in the wires. It's an emergent
1257property of human minds to create. "Due directly to the GPL," as
1258Vallopillil rightly pointed out, free software made available to him
1259an exhilarating increase in his own creativity, of a kind not
1260achievable in his day job working for the Greatest Programming Company
1261on Earth. If only he had e-mailed that first addictive fix, who knows
1262where he'd be now?</para>
1264<para>So, in the end, my dwarvish friends, it's just a human thing.
1265Rather like why Figaro sings, why Mozart wrote the music for him to
1266sing to, and why we all make up new words: Because we can. Homo
1267ludens, meet Homo faber. The social condition of global
1268interconnection that we call the Internet makes it possible for all of
1269us to be creative in new and previously undreamed-of ways. Unless we
1270allow "ownership" to interfere. Repeat after me, ye dwarves and men:
1271Resist the resistance!</para>
1274<!--<para><img src="anarchism_files/quad.gif"></para><a name="m4"></a>-->
1277<title>IV. Their Lordships Die in the Dark?</title>
1279<para>For the IPdroid, fresh off the plane from a week at Bellagio
1280paid for by Dreamworks SKG, it's enough to cause indigestion.</para>
1282<para>Unlock the possibilities of human creativity by connecting
1283everyone to everyone else? Get the ownership system out of the way so
1284that we can all add our voices to the choir, even if that means
1285pasting our singing on top of the Mormon Tabernacle and sending the
1286output to a friend? No one sitting slack-jawed in front of a televised
1287mixture of violence and imminent copulation carefully devised to
1288heighten the young male eyeball's interest in a beer commercial? What
1289will become of civilization? Or at least of copyright teachers?</para>
1291<para>But perhaps this is premature. I've only been talking about
1292software. Real software, the old kind, that runs computers. Not like
1293the software that runs DVD players, or the kind made by the Grateful
1294Dead. "Oh yes, the Grateful Dead. Something strange about them, wasn't
1295there? Didn't prohibit recording at their concerts. Didn't mind if
1296their fans rather riled the recording industry. Seem to have done all
1297right, though, you gotta admit. Senator Patrick Leahy, isn't he a
1298former Deadhead? I wonder if he'll vote to extend corporate authorship
1299terms to 125 years, so that Disney doesn't lose The Mouse in 2004. And
1300those DVD players - they're computers, aren't they?"</para>
1302<para>In the digital society, it's all connected. We can't depend for
1303the long run on distinguishing one bitstream from another in order to
1304figure out which rules apply. What happened to software is already
1305happening to music. Their recording industry lordships are now
1306scrambling wildly to retain control over distribution, as both
1307musicians and listeners realize that the middlepeople are no longer
1308necessary. The Great Potemkin Village of 1999, the so-called Secure
1309Digital Music Initiative, will have collapsed long before the first
1310Internet President gets inaugurated, for simple technical reasons as
1311obvious to those who know as the ones that dictated the triumph of
1312free software <footnote> <para>30. See "They're Playing Our Song: The
1313Day the Music Industry Died," in <emphasis>The Invisible Barbecue</emphasis>,
1314forthcoming.</para> </footnote> . The anarchist revolution in music is
1315different from the one in software <emphasis>tout court</emphasis>, but here too -
1316as any teenager with an MP3 collection of self-released music from
1317unsigned artists can tell you - theory has been killed off by the
1318facts. Whether you are Mick Jagger, or a great national artist from
1319the third world looking for a global audience, or a garret-dweller
1320reinventing music, the recording industry will soon have nothing to
1321offer you that you can't get better for free.  And music doesn't sound
1322worse when distributed for free, pay what you want directly to the
1323artist, and don't pay anything if you don't want to. Give it to your
1324friends; they might like it.</para>
1327What happened to music is also happening to news. The wire services,
1328as any U.S. law student learns even before taking the near-obligatory
1329course in Copyright for Droids, have a protectible property interest
1330in their expression of the news, even if not in the facts the news
1331reports <footnote><para>31. International News Service v. Associated
1332Press, 248 U.S. 215 (1918). With regard to the actual terse, purely
1333functional expressions of breaking news actually at stake in the
1334jostling among wire services, this was always a distinction only a
1335droid could love.</para></footnote>.  So why are they now giving all
1336their output away? Because in the world of the Net, most news is
1337commodity news. And the original advantage of the news gatherers, that
1338they were internally connected in ways others were not when
1339communications were expensive, is gone. Now what matters is collecting
1340eyeballs to deliver to advertisers. It isn't the wire services that
1341have the advantage in covering Kosovo, that's for sure. Much less
1342those paragons of "intellectual" property, their television
1343lordships. They, with their overpaid pretty people and their massive
1344technical infrastructure, are about the only organizations in the
1345world that can't afford to be everywhere all the time. And then they
1346have to limit themselves to ninety seconds a story, or the eyeball
1347hunters will go somewhere else. So who makes better news, the
1348propertarians or the anarchists?  We shall soon see.</para>
1350<para>Oscar Wilde says somewhere that the problem with socialism is
1351that it takes up too many evenings. The problems with anarchism as a
1352social system are also about transaction costs. But the digital
1353revolution alters two aspects of political economy that have been
1354otherwise invariant throughout human history. All software has zero
1355marginal cost in the world of the Net, while the costs of social
1356coordination have been so far reduced as to permit the rapid formation
1357and dissolution of large-scale and highly diverse social groupings
1358entirely without geographic limitation <footnote> <para>32. See "No
1359Prodigal Son: The Political Theory of Universal Interconnection," in
1360<emphasis>The Invisible Barbecue</emphasis>, forthcoming.</para> </footnote> . Such
1361fundamental change in the material circumstances of life necessarily
1362produces equally fundamental changes in culture. Think not? Tell it to
1363the Iroquois. And of course such profound shifts in culture are
1364threats to existing power relations. Think not? Ask the Chinese
1365Communist Party.  Or wait 25 years and see if you can find them for
1366purposes of making the inquiry.</para>
1368<para>In this context, the obsolescence of the IPdroid is neither
1369unforseeable nor tragic. Indeed it may find itself clanking off into
1370the desert, still lucidly explaining to an imaginary room the
1371profitably complicated rules for a world that no longer exists. But at
1372least it will have familiar company, recognizable from all those
1373glittering parties in Davos, Hollywood, and Brussels. Our Media Lords
1374are now at handigrips with fate, however much they may feel that the
1375Force is with them. The rules about bitstreams are now of dubious
1376utility for maintaining power by co-opting human creativity. Seen
1377clearly in the light of fact, these Emperors have even fewer clothes
1378than the models they use to grab our eyeballs. Unless supported by
1379user-disabling technology, a culture of pervasive surveillance that
1380permits every reader of every "property" to be logged and charged, and
1381a smokescreen of droid-breath assuring each and every young person
1382that human creativity would vanish without the benevolent aristocracy
1383of BillG the Creator, Lord Murdoch of Everywhere, the Spielmeister and
1384the Lord High Mouse, their reign is nearly done. But what's at stake
1385is the control of the scarcest resource of all: our
1386attention. Conscripting that makes all the money in the world in the
1387digital economy, and the current lords of the earth will fight for
1388it. Leagued against them are only the anarchists: nobodies, hippies,
1389hobbyists, lovers, and artists. The resulting unequal contest is the
1390great political and legal issue of our time.  Aristocracy looks hard
1391to beat, but that's how it looked in 1788 and 1913 too. It is, as Chou
1392En-Lai said about the meaning of the French Revolution, too soon to
1397<title>About the Author</title>
1399<para>Eben Moglen is Professor of Law &amp; Legal History, Columbia Law School.
1400E-mail: <ulink url="">Mail:</ulink></para>
1404<para>This paper was prepared for delivery at the Buchmann
1405International Conference on Law, Technology and Information, at Tel
1406Aviv University, May 1999; my thanks to the organizers for their kind
1407invitation. I owe much as always to Pamela Karlan for her insight and
1408encouragement. I especially wish to thank the programmers throughout
1409the world who made free software possible.</para>
1414<ulink url=""><!--<img src="anarchism_files/contents.gif" alt="Contents" align="bottom" border="0">--></ulink> </para>
1416<ulink url=""><!--<img src="anarchism_files/index.gif" alt="Index" border="0">--></ulink> 
1418<para>Copyright <ulink url="">©</ulink> 1999, First Monday</para></blockquote>
Note: See TracBrowser for help on using the repository browser.