Open Source--The Unauthorized White Papers Chapter 1, The Origins of Open Source Software

Open Source: The Unauthorized White Papers

Chapter 1

The Origins of Open Source Software

Open Source is not a new way of doing things ó it is the original computer way of doing things. Linux did not suddenly appear like a mushroom ó it is (as Microsoft scornfully puts it) a "30-year-old operating system technology and architecture." Linux is not the first free mimic of UNIX ó that was BSD, which is still with us and growing stronger. Itís not an obvious joke when Linus Torvalds and his followers talk about World Domination and laugh ó they laugh because they enjoy sounding like comic-strip villains, and because they know they mean it. To follow all these threads in computer history and social history, we need to go back about 40 years.

In the BeginningÖ

International Data Corporation (IDC, http://idc.com/) currently rates the number of Linux users (not systems) in the world at anywhere from 7 to 21 million; the predicted growth rate is 200 percent for 2000. This large number did not spring up overnight, although it looks like that. The forces that brought on Linux go back many years, and to set the proper context we will have to go back and look in places besides computer companies and laboratories.

The 1960s and early 1970s ó datapriests and hippies

In the early days of computing, hardware manufacturers gave away their software, because their machines would not operate without it. Digital Equipment Corporation did not bother to copyright its operating system, and even provided customers with a large catalog of free software that was distributed originally on punched paper tape and later on "Linc tape." Users paid five dollars to cover the cost of copying a program and could then make their own copies for free. There might be jokes that came along with the operating system, and there were third-party games, such as Spacewar, which MIT computer students wrote and freely distributed. No one thought about rights to the software, let alone business practices; the software was a give-away needed to sell the actual article of commerce ó the expensive hardware. Because none of the software would run on a competitorís machine, no one gave "software piracy" a thought, let alone a name.

The machines were distant deities, dwelling in glass houses and approachable only through the intermediation of the datapriests. Students in the outer courtyard of the temple prepared forms from which keypunch "girls" prepared the actual cards for eventual offering to the computer. After this step the hopeful carried their box(es) of cards (it was very bad luck to drop them) to the datapriests, who fed them to Moloch. Results might be days in coming, and only then were results or errors known.

Life in California in those days had not reached the frenetic pitch of today, even in the emerging Silicon Valley. There were some workaholic engineers in the aerospace and silicon chip industries, but for the most part California remained the land that had invented modern leisure (or inherited it from the original Mexican settlers). Here the hippies arose and flourished and preached and practiced their main tenets of charity and antiauthoritarianism. They believed that the quest for material possessions was corrupting the world, and that life would be better if lived more simply, particularly if everyone cooperated and shared freely what they had. At the same time they were highly resistant to authority, and burned with a self-righteous zeal to smash the state. Self-dramatization was part of the ethos.

In the California of leisure, many young people believed that their jobs were merely the means to support their real lives. Living in an alternative reality (when not at work) was popular in a land with flourishing groups like the Society for Creative Anachronism and various sci-fi fan clubs, and with nature religions mixing varying proportions of masquerade, ritual, and belief. The recession at the end of the 1960s left many creative and energetic young people without jobs, and activities like the original Renaissance Faire flourished because its promoters took full advantage of the many volunteers eager to play at living in another world. After the spread of the microcomputer industry across the Santa Clara (now Silicon) Valley, the Renaissance Faire went into a decline and eventual takeover.

Computers were not absent from this world; the first big microcomputer show in San Francisco called itself the Computer Faire. Marshall McLuhan was the chief of the prophets of a coming convergence of media, computing, and telecommunications that would enable truly public discussion and give real "Power to the People!" Ted Nelson began his Xanadu Project, which aimed to place in hyperlinked cyberspace everything that anyone had ever written or said, or written or said in comment thereon. Xanadu would be powerfully democratic because it would be electronically accessible to everyone. In Berkeley, cradle (and later, museum) of the Revolution, Lee Felsenstein toiled at the beginnings of such a network he called Community Memory as he worked to put computer terminals in public places so that anyone could read and post to an electronic bulletin board. The antiauthoritarian message of these projects was clear: no gatekeepers, no datapriests, no mediators.

If all of this seems a product of fantasy, it was. In addition to the alternative existences mentioned previously, there were elaborate role-playing games (RPGs) such as Dungeons and Dragons (played without computers in those days). Young technical employees and computer students found these appealing, along with the science fiction that spoke of computer realms of the future. Tolkienís fantasy work, The Lord of the Rings, was particularly popular; it echoes today in the computer industry with names like Gandalf and Palantir. There were even more products of these grandiose and fruitful fantasies of power. Ted Nelsonís dream of universal access to the collective mind embodied itself in the World Wide Web; and Lee Felsenstein went on to design the Osborne 1 in 1980. The biggest change the hippies made in society was the personal computer, and the personal power and attitudes that still surround it.

The 1970s ó The MIT AI laboratory

Although California had several important sites contributing to the progress of computer technology and education, we need to move across the country to the Massachusetts Institute of Technology (MIT) for the next part of our story.

Hacker vs. Cracker

The Open Source world is hacker-driven, and hackers define themselves as clever software programmers who push the limits of the doable. This attitude ó "It wasnít meant to do this but I can make it do this" ó is the antiauthoritarian germ that flowers freely in the cracker, a person who illegally breaks into othersí systems and exploits their resources. Crackers believe they are the cleverest of the hackers because they are pushing the systems (computer and legal) hardest.

Law-abiding, ethical hackers (the majority) resent the popular use of the term hacker to describe the disruptive and lawless crackers, and proposed the term cracker to describe them. This book observes the distinction between the two terms.

 

For a few years, the Artificial Intelligence lab at MIT was a center of hackers ó young men who lived to program and freely share their clever exploits with others. During the close of these golden hacker years, it was the home of Richard M. Stallman, known familiarly by his computer handle of RMS. Stallman began the Free Software movement, whose extension is the much newer Open Source Initiative.

Some Resources on Hackers, UNIX, and Linux

If you are looking for some background to the hacker culture at MIT and other places in the 1970s, Steven Levyís 1985 book, Hackers, provides a history; the Internetís Jargon File carries some of its folklore ó in fact Stallman was one of the original authors. You can find the file in many places on the Web, but you might start at http://www.tuxedo.org/~esr/jargon/ because the site belongs to Eric S. Raymond, who does the printed versions known as The New Hackerís Dictionary. Here you will find entries such as the Hacker Ethic.

Raymond has written a number of pieces on "hacker anthropology," including "The Cathedral and the Bazaar," which contrasts Open Source software development with centralized, proprietary development. He recently published these pieces in book form, but they have existed for years on the Internet. You can find them at http://www.tuxedo.org/~esr/writings/.

UNIX veteran and historian Peter H. Salus published A Quarter Century of UNIX in 1994, if you want the facts; the mythic dimensions of Linux are handled in my "Copyleft and the Religious Wars of the 21st Century" at http://www.stromian.com/copyleft.htm. Don Libes and Sandy Ressler cover the early period in Life with UNIX: A Guide for Everyone.

 

As Richard Stallman tells it, the MIT AI lab was a computer paradise in the 1970s, a democracy of learning. Everyone shared the programs they had written; there were no passwords to exclude users, either from using the computer or reading anyoneís files; and people were always ready to help each other. Users were free to improve any program they felt needed improving, because with an open system the users were in full control, and free of central authority. When professors had terminals in their offices, and the students had none, they felt justified in breaking into locked offices to use them. It was self-evidently selfish or thoughtless for a professor to deny them the use of a terminal at 3 a.m. when the professor was hardly there to use it himself. When the AI lab introduced passwords to comply with government requirements, Stallman fought against them, urging people to use a single carriage return as a password, so that effectively there would be none.

New restrictions such as passwords were harbingers of coming rough winds in paradise. Previously, users had programmed the system so that the printer sent messages if the paper had jammed. The new computer, on the other hand, did not provide the source code for its printer driver, so that users could not hack it to add the summons. Even worse, Stallman was outraged to discover a person at Carnegie-Mellon University who admitted to having the driver source code, but who told him that the code could not be shared because the holder had received it under a non-disclosure agreement (NDA). This was Stallmanís first encounter with the secrecy of the proprietary software world, which guards its source code as an asset and releases only binary code, readable by machines but not by humans. Stallman regarded the signer as a traitor who had betrayed the hacker culture and humanity, depriving everyone of the benefits that would otherwise flow from being able to modify the driver.

The worst finally came: stalwarts of the AI lab broke away to form two rival companies to exploit the LISP language they had developed there, and by the early 1980s the companies had drained off the skilled personnel. As the expert programmers were hired away, the lab replaced its open system with a closed, proprietary system that was maintained by the vendor. Without source code, hacking was over. Stallman, who refused to join any proprietary venture, was left alone. Where he went, and what he did, we will take up in a moment.

The 1980s ó UNIX and the great betrayal

Besides the universities, there were other great centers of software development in the United States, among them Bell Labs. This was better than a university ó there didnít seem to be a lack of money. In this rarified atmosphere of abstract thought and pure research, work looked like play. A small robot might race down the halls carrying files, nudge your office door open, and demand to be put on your desk so its remote operator (your colleague) could look at you through the robotís camera, talk to you through its sound system, and discuss the files it had brought from the colleagueís office. One scientist recalled a Bell Labs meeting he had attended remotely from California: as he used the camera to scan those seated around the table, he noticed two other robots perched on the table, likewise sent by remote attendees.

This was the perfect place to invent the perfect computer system. On early computer systems, software was written in machine language specific to the machine it was run on. Development tools were also specific to a system. The result was duplication of effort; each machine had all its software written from the ground up. The idea of Bell Labsí UNIX was to write an operating system that would be cross-platform and that could be implemented on each type of machine. The software could simply be compiled as it moved from platform to platform. The theory of UNIX was simple, but the practice was complex; UNIX was not user-friendly, and chose its friends carefully.

Hardware manufacturers, who had originally seen software only as something that had to be supplied to sell machines, began to realize that not only could the software that came with the machine benefit the customer, it could also tie the customer to the machine. Programs could not be moved from one platform to another, forcing customers to be loyal. UNIX was to be a way around this problem. The resulting portability made it popular in the universities. Bell Labs made it easily available to other software researchers, and students studied its source code in their classes. From around the world, bug fixes and improvements poured into Bell Labs.

This open-source software paradise, like that of the MIT AI lab, ended because of commercial pressures. As AT&T and its Bell Labs were being broken up in the late 1970s and early 1980s, UNIX was recognized as a commercial asset, and a price tag was put on the source code. It was too rich for the universities. UNIX, however, did appeal to the new generation of hardware companies then rising: the makers of minicomputers. Licensing UNIX provided them with a dependable operating system developed by a responsible third party. It was scalable to large numbers of transactions for large companies. But rather than enjoy the benefits of portability and interoperability, the hardware vendors applied the lessons of lock-in. As each tweaked the source code they had received so that it would run faster on their particular machines, they also differentiated themselves so that their programs would run only on their own software/hardware systems. Customers were still locked in. Because the emphasis was still on hardware, software was merely the means to the end. The UNIX Wars saw as many varieties of UNIX as there were vendors.

The blighting of these twin paradises set new forces in motion. Stung that all their UNIX contributions had been taken from them, locked up, and sold to large corporations for large sums that they could not themselves afford to pay, developers in the universities quietly plotted their next move. In Amsterdam, Andrew Tannenbaum sat down and wrote a UNIX imitation called MINIX, so students would have a system to study. The University of California at Berkeley was a strong center of UNIX development, and had already begun distributing a version of its own: BSD. BSD at times surpassed Bellís UNIX, and its freely modifiable and distributable code made its way into numerous commercial versions of UNIX. It also formed the basis of Sun Microsystemsí operating system. Critics who say that Open Source software must always lag behind and imitate proprietary software should study the history of BSD, and see how many innovations and improvements originated there.

Over the years, and under legal pressure from AT&T, the BSD developers dropped every line of Bell Labsí code, and substituted their own. Meanwhile, Richard Stallman was contemplating the ruins of his programmerís paradise. What he had experienced there, he would recreate for all mankind. He resigned from the AI lab, though he was allowed to occupy space there. He founded the Free Software Foundation (http://www.fsf.org) to serve as the vehicle for making real his grand vision of an operating system that would be forever free for all developers, and which would turn the whole world into a programmerís paradise.

He began by creating tools to build the new system, using (for the moment) proprietary tools because they were the only ones available. He created EMACS (an acronym for Editing MACroS), an editor and general-purpose tool still in use. Stallman supported himself by providing free copies of it, but charging $150 for actually making the copy and for support. He asked that all changes and improvements be sent back to him. In this way, he evolved the model of the GNU General Public License (GPL), which allows free use, modification, and distribution of the software and any changes to it, restricted only by the stipulation that those who received the software pass it with identical freedoms to obtain source code, modify it, and redistribute it. The products of a developerís mind must remain free for all developers to use and reuse. Free does not mean that no money changes hands, but it does mean that no authority or non-disclosure agreement will prevent developers from sharing code. Because of the freedoms conferred by the Free Software licensing method, and because the legal means of its enforcement is the copyrighting of the software itself, the use of the GNU GPL is often called "copylefting."

The 1990s ó the arrival of Linux

Stallman says that he chose UNIX as his model because that way he would not have to make any design decisions. He called the system GNU (GNUís not UNIX), and the GNU tools that he built remain important and widely distributed to this day. The task of building the kernel he left to last; it was intended to be much better than the UNIX kernel, and was called the HURD. By the time Stallman had reached this stage of the process, some ten years later, a Finnish computer student who was disgusted with MINIX had begun his own project to clone UNIX, using the popular GNU tools. This was the Linux kernel of Linus Torvalds, and when added to the GNU material already available it enabled the parts to add up to an operating system.

The rapid success of Linux resulted from a number of factors. First, Linux made it possible for developers to continue at home what they did at work. Instead of buying an expensive UNIX workstation along with an expensive operating system, they could use Linux on an ordinary personal computer. The fact that so many people around the world were working on Linux and talking about it created a buzz. The timing was also right because BSD was tied up in litigation with AT&T over whether the Berkeley Distribution contained proprietary Bell Labs material. And the HURD, however clever its architecture, was not yet ready for release.

The arrival of the Linux system in the early 1990s was also made possible by its Free Software licensing scheme, originated by Stallman. Whether the Linux system would have been as successful if it had been released under the BSD or X licenses (all these licenses are discussed in detail in Part III) will never be known. Torvalds pragmatically chose to use the GPL because it allowed freer source code distribution than did the MINIX license, and because it was the license used for the GNU software and it seemed to do the job; he does not claim to be a particular fan of it.

The GPL does have the virtue of making sure that any software changes or additions incorporated in GPLíd software are likewise distributed under the GPL; this gives only one licensing possibility for the Linux kernel. Proprietary material might be loosely linked to the kernel by means of the GNU LGPL (a second license allowing this concession that Stallman later regretted), or proprietary programs could simply run atop the Linux system. Torvalds made an important concession to the use of binary-only driver software by allowing the "kernel-loadable module." Stallman disapproved, but said that because neither he nor the Free Software Foundation held the copyright on the kernel software, Torvalds had the right to see things his way.

In contrast to the pragmatism of Torvalds, Stallman remained fixed on his idea of an entire system distributed solely under the GPL. Implacably hostile to proprietary or closed-source software, he steadfastly continued to promote Free Software and Copyleft as the one true faith. To keep the worldís attention focused on GNU, he insisted the system be called GNU/Linux (only the Debian distribution follows this convention), and he finished and released the HURD.

Stallmanís licensing vision is developer-centric, or hacker-centric, if we recall the Hacker Ethos. Stallmanís motive is to keep the code free for other hackers to use. Eric S. Raymond has explored the social motivations of hackers in freely giving away their code (the GPL merely enforces this custom), but in the dedication of Richard Stallman there is self-confessed religious fervor. On public occasions he is wont to dress up in robe and halo (a large disk-drive platter from some forgotten hardware) as St. IGNUtius of the Church of EMACS, and exorcise all proprietary software from those computers brought to him by the faithful. This is the sort of "ha-ha only serious" self-mockery much appreciated in hacker culture.

The Intersection of Hackers and the Real World

Hacking and its folk-ways grew up in an academic atmosphere where wit and indirection are prized. The business community, established chiefly on trust and taking people at their self-stated face value, confines its humor to telling funny stories rather than exchanging playfully-bent perspectives on the world. The world outside Linux is currently wondering how seriously to take Linux and its promises for the future, and is trying to separate truth from what appears to be absurd boasting.

Watch and hearken, ye solitary! The winds of the future approach on stealthy wingbeats, and to sharp ears come good tidings.

Zarathustra, Teil II: Von der schenkenden Tugend, 3

What is Open Source and why does it crave world domination?

Popular accounts of the Linux Community (often called a movement) tend to focus on the bizarre; visits to hardcore Linux sites on the Internet turn up a colorful landscape showing traces of all the elements that have shaped the Community. If you want to draw a caricature, itís easy enough: What do you get when you mix the antiauthoritarian energy and millenarian idealism of hippies, the competitiveness and arrogance of elite hackers, and the role-playing tastes of technical employees? And if you mix in the ideal of service to the Big Idea, a desire for a final showdown between the Children of Light and the Forces of Evil like that at the end of Tolkienís saga, and a sprinkling of born-again Christians? And all of them speaking a dense technical language? Who have somehow taken a penguin as their totem? And whose undisputed leader is a slender young Finn who combines purpose, rationality, maturity, and political skills with a dry sense of humor?

In earlier days it was easier for an outsider to grab the world of Linux by the wrong handle because colorful, self-promoting types were what attracted the press. There is no organized community, but there are organizations within the community. The inner sanctum might be the actual contributors to the kernel itself. Paul Jones of the University of North Carolinaís Open Source Research Team and MetaLab (http://metalab.unc.edu) says that that there are fewer than 200 named contributors to the kernel (the majority of them in Germany), although there are thousands of contributors to the software surrounding it. There is a central industry organization, Linux International (http://li.org/), nominally supported by the dues of various Open Source vendors, but actually dependent on the energy and dedication of its executive director, Jon "maddog" Hall, a UNIX veteran who was at DEC for many years.

The vendors themselves provide further focus points, but the whole community is dispersed around the globe (it originated outside the United States and is probably stronger outside the U.S. than in it), and in constant contact over the Internet. The focus points are generally software development projects. The number of machines running Linux is estimated to be well over ten million worldwide, and while there are probably more machines than users, there must be several million Linux users of varying skill levels as well.

The number of Usenet lists and Web news sites serving these users is endless. One addictive spot is the flamboyant Slashdot (http://slashdot.org), which handles more topics than Linux, changes the news several times per hour, and allows all comers to post their comments on that news and on each othersí comments. Linux Weekly News (http://lwn.net) is more sedate and generally more technical, while LinuxWorld (http://linuxworld.com) is stamped by its IDG parent with a blend of commercialism and Linux advocacy. Links to the growing number of news items about Linux are continually posted at Linux Today (http://linuxtoday.com/). If the Community includes a wide spectrum of personality types and interests, nevertheless people in the Community know who is in it, and know who the leaders are and what they are doing. The Open Source world provides a look backwards into the software world of 15 or 20 years ago; in those days flamboyant personalities were making an industry that is nowadays dominated by blander corporate types.

The Linux Community itself has been undergoing changes over the past two or three years. The primary barrier between the Community and the rest of the world has been the technical competence needed to deal with Linux. For a long time many members of the Community were not only proud of this difference, but enjoyed its use as a barrier against outsiders. Furthermore, the more juvenile members of the community discouraged outside interest by furiously attacking any journalist who wrote something about Linux, and (naturally) got some of the technical details wrong. They would subject the writer to public (at least on the Internet) flaming as clueless and totally unworthy to appreciate, let alone write about the wonderful thing they hugged to their breasts. Proud of their command line interface, many scorned the idea of a graphical user interface (GUI), not just for themselves, but because it might admit the feeble-minded to the sanctuary.

This self-sufficient society, being developer-centric, assumed that all the people who counted could compile their own kernels, and find all the technical help they needed on the Internet. It had the atmosphere of the California of a generation before, when people joined food co-ops and called them "food conspiracies," one more proof of the cleverness of "us" as compared with the others. A number of Linux hackers follow Ayn Rand, and it never occurs to them that not everyone is a Hank Rearden who can invent Rearden Metal. This is a world in which only programmers count.

What has saved Linux from sinking into the navel of its own regarding is its usersí sense of mission. Linux users have always regarded Microsoft code as third-rate technology, and pitied those who are locked into it through their own technical ignorance. Many Linux zealots would like to see Microsoft wiped out, and all would like to see a choice of operating systems for users. At one time most seemed to believe that Linux would eventually dominate because of its technological superiority, but they have come to realize that market forces are stronger than technological excellence. The Community accordingly shifted its sights to see what it could do to make World Domination happen.

Some humorless observers seem to misunderstand Linus Torvaldsí references to Linux World Domination. This expression of manifest destiny is simply part of the joking that accompanies his technical talks (describing features of the upcoming 2.4 kernel, he said it "will not have a mauve screen of death. . . .Wintel has a lot of patents in that area"). Torvalds soberly says that a 15 percent market share for Linux would be fine ó then people would have a choice. Linux fans like the jokey drama implied in World Domination ó a secret conspiracy that eventually controls the planet, and they like the deeper irony that if Linux becomes the dominant operating system, users will have choices ó freedom ó and that freedom is the opposite of World Domination.

The microcomputer world was always a friendly one. In the early days of PCs nonusers often heard from their friends: "Youíve got to get one! Donít worry, Iíll give you all the software you need. Just get one ó youíll love it!" Now, not only have local Linux User Groups (LUGs) been holding Installfests to help beginners past the traditionally difficult but increasingly easier installation of Linux, they announce these Installfests as public events to which anyone can bring a computer and have Linux installed on it. Taking the World Liberation/Domination theme seriously meant that the Linux Community had to address seriously issues such as ease of installation (the various Linux distributions are making rapid progress) and a graphical user interface (GUI) that would enable former Windows users to function on Linux. As soon as Microsoft pointed out Linux performance problems, developers began to improve the weak points.

This joy in writing good code to earn honor from other hackers fits Eric Raymondís descriptions of the hacker community; the love of good work for itself sounds like Ayn Randís geniuses, except that these can withhold the products of their genius, and look down on the rest of humanity. A better image of the Open Source community in action is probably Nietzscheís idealization of the perfected person who has evolved beyond humanity, the Highly Evolved or even Super-Evolved Übermensch. This person transcends humanity by making himself the pure conduit of the Life Force (der Wille zur Macht) and expresses this transcendence by an endless, bubbling creativity that benefits all humanity. This Virtue of Bestowing (die schenkende Tugend) is the highest virtue.

Who Actually Writes Linux?

Metalab at the University of North Carolina hosts a number of Linux archives, including all major distributions of the kernel, the Linux Documentation project, and the Linux Archive. This latter group contains most of the auxiliary software that has been written around the Linux kernel. Taken together, they form one of the very largest collections of Linux material on the Internet, containing over 30GB of data dating back to 1992. Anyone may send in a piece of code, making Metalab the most open such site on the Web. Submissions of code to the site must include a piece of metadata called a Linux Software Map (LSM), naming the submitter and giving other information. Recently several researchers there (Bert Dempsey, Debra Weiss, Paul Jones, and Jane Greenberg) studied the credits inside the Linux kernel and the LSMs for the material around it to form a preliminary picture of who is actually writing the code: A Quantitative Profile of a Community of Open Source Linux Developers (http://metalab.unc.edu/osrt/develpro.html).

It is important to remember that LSMs, like the kernel, do not name contributors who submitted patches, regardless of whether these patches were accepted. We can thus assume a larger but unknown pool of Open Source developers.

The Kernel is a Worldwide Project from a Select Few

The software in the Linux kernel is the result of sifting many contributions and patches from a large number of people; unfortunately, we still do not know how large a group this is. We do know that as contributions pass up the kernel organization (eventually to Linus himself), more are rejected than accepted. The Metalab study shows that there are fewer than 200 accepted contributors to the kernel that is the heart of Linux. The largest single group of contributed code portions (about 40 percent) came from .com addresses, possible evidence that a substantial portion of the kernel contributions are corporate-based. Although we cannot assign nationality to the .com suffix, 42 percent of the contributions using a national suffix were European, and of this European group the Germans contributed a dominant share of 32 percent. Thus there is a substantial international input to the Linux kernel. The educational world (.edu), as might be expected, contributes about 28 percent of the submissions. The total kernel contribution picture is heavily international, especially German, and involving substantial input both from industry and academe.

The Rest of Linux is a Worldwide Project with Many Contributors

The kernel is useful only when surrounded by a host of auxiliary software that includes drivers, tools, applications, scientific tools, games, libraries, utilities, and the like. The contributors represented in the study are by no means the sum total of all Linux contributors: only those whose contributions were chosen for use by project leaders end up as contributors to the many files in the Archive. An unknown but probably substantial number of contributors had their work declined. The successful contributors number some 2,200 persons. Of these contributors, 75 percent provided one contribution, and 91.4 percent only one or two; only 2.2 percent contributed five or more pieces of code. These figures show that Open Source software is working the way its proponents claim: individual developers scratch their itch, and then contribute their code so that others may benefit from it as well.

Code contributions to the auxiliary archive show that 37 percent came from addresses with European suffixes, 23 percent from .com, 12 percent from .edu, and 7 percent from .org. The suffixes show that this software is more international than the kernel, and less dependent on the academic world.

Although the Metalab study makes the number of Linux developers appear small, there are other indications that a great many developers are capable of working with Linux. Netscape is said to have had 250,000 copies of the Mozilla code downloaded in the first month of its availability in 1998; a study of the Internet addresses of the downloaders would make an interesting study.

Top-Level Domains of Linux Contributions

Free software and Open Source software

The recognition that Linux and Open Source software actually have a chance for World Domination has shifted Community focus from inward to outward. Miguel de Icaza, who could respond to a hacker complaint about instability in the prerelease graphical desktop GNOME with a smiling "Linux: itís not for whiners," says that his attitude towards users has changed completely, and that he wants new users to have an easy, comfortable Linux experience. Even the clannish computer games experts are turning their graphics skills to end-user interface development, and it has been a long time since anyone flamed an outsider.

The Community has also noticed how resistant many businesses are to the GNU General Public License and to Richard Stallmanís insistence that no software can be completely free unless it is issued under the GPL. While Stallman takes this position to defend the freedoms of the user, he sees the user as a competent software developer, not a point-and-click end-user. This developer-centric attitude ignores the vast body of users who are happy with binary-only software, and it defines the only way of making money from Free Software (as it is officially called by the Free Software Foundation) as developer-provided services: program modification or extension, technical support, technical documentation, and training. An end-user who cannot program can hire a developer to make changes in a program provided the source code and permission to do so are available. This is a hacker-centric view of the world, one which defines computer literacy as programming facility.

To provide a more latitudinarian view of the software world, a group of Community leaders met in early 1998 to discuss the situation and to find a way of promoting Free Software to a skeptical business community. The Open Source Initiative (OSI, http://www.opensource.org/) was formed to carry the torch of Open Source software. By all accounts, the effort has been successful, and can take some of the credit for the rapid rise of interest in Linux and things Open Source. At the same time, the groupís touchstone for Open Source software, the Open Source Definition (or OSD, based on the Debian Free Software Guidelines by Bruce Perens) and its interpretation by the OSI add to the lack of clarity about what constitutes Open Source software.

 

Why the Penguin?

Linus Torvalds chose the Linux penguin, commonly known as "Tux," as the badge of Linux. Torvalds shared his vision of the ideal penguin emblem in a parodic mind-concentration exercise eventually posted in many places on the Web (http://www.linux.org/info/penguin.html): "OK, so we should be thinking of a lovable, cuddly, stuffed penguin sitting down after having gorged itself on herring. Still with me?"

Tux was not universally popular at first, and some objectors said that it was hardly an emblem likely to appeal to the business community. Torvalds met the objections by saying that the penguin was much better than the typical corporate emblem designed by an agency because Tux was flexible. Any Linux project or distribution is free to use Tux or its many variations: the bird appears with a policemanís cap and club (security); necktie, briefcase, and cell phone (business); boots and bierstein (Linux Bierwanderung, an annual procession through the forests and bierstuben of Germany, while packing laptops); wearing a biker t-shirt with "Born to Run UNIX," and also carries national banners. Larry Ewing at Texas A&M created Tux; designs are on file at http://www.isc.tamu.edu/~lewing/linux/.

This does not explain Torvaldsí choice of a penguin in the first place. It took a summertime expedition to Helsinki to find the answer. In 1997, Jon "maddog" Hall, executive director of Linux International and eventual godfather of Torvaldsí children, went to Finland to gather material on the young man for a Linus Roast scheduled as a benefit dinner at the first Atlanta Linux Showcase. He learned a great deal about Torvalds, including his fondness for cats and ice cream; in fact, nothing could interrupt Torvalds at programming except the bell of the ice cream truck, which would invariably turn him out. As maddog wandered the Helsinki streets, he became aware that he was seeing penguins everywhere: the trademark for a popular ice cream!

 

The OSI was founded not to commercialize Open Source software (Red Hat and Cygnus had successfully done that), but to promote the commercialization of Open Source software. The Initiative believed that the way to do this most advantageously to the Open Source Community would be to issue the Open Source Definition and grant seals of approval to licenses that met the Definition. Although the Initiative carefully replaced the term Free Software with the term Open Source software, and recognized licenses beyond the X License and the GNU GPL, it nevertheless kept the Free Software Foundationís hacker perspective: an individual must be free to modify, reuse, and redistribute the code. Accordingly its list of approved licenses expanded to include those that favored individual use of the software material, even at the expense of commercial enterprises, while rejecting licenses that served commercial ends, even if they gave developers great latitude but infringed in some small way on an individual developerís freedom with the code. On top of this hacker bias in the Initiative, its powerlessness to control the very term Open Source led a number of parties to describe their own licensing arrangements as Open Source, whether they conformed to the Definition or not; among them are the Sun Microsystems licenses that the Community resents for their severe distribution restrictions.

As a result of these controversies, the public is confused about the actual meaning of "Open Source software," and the only common definition seems to be that the term means "source-inspectable software," with varying degrees of freedom beyond that, depending on the license. There is a large section in this book dealing with the ins and outs of various licenses and licensing strategies, but the mention of just a few mysteries of the approved license list of the OSI is not out of place here.

* The X License permits a developer to receive source code, modify it, and then to redistribute the derivative product without any accompanying source code at all; it is on the OSI-approved list.

* The license for the Apache Web server, mainstay of the Internet and pride of the Open Source Community, does not appear on the approved list.

* The Q Public License (QPL) is an OSI-approved license that does not permit any modification of the source code at all, but only the distribution of patches accompanied by source code. This latter requirement is unlikely to appeal to commercial software distributors, who will then have to buy a commercial license to keep their source code a secret. This restriction is unlikely to bother individual developers who are giving their products away, and because it seems to be there to push commercial users into paid licenses, it appears to violate the OSD restrictions against discriminating on the basis of intended use or groups of persons.

* Sun licenses tend to approve modifications to the code, provided the derivative passes a compatibility test, but none of the Sun licenses has been approved.

* The Zope server is very popular in the Community, and permits modification and redistribution of its source code, but its license (the ZPL) does not appear on the approved list, either.

* The Mozilla license permits modification to its basic code only in the form of patches, but allows extensions to the basic code to be shipped as binary-only; it is on the approved list.

It doesnít seem possible to draw any firm conclusions from studying the OSD, the licenses on the approved list, and licenses which have been rejected; for that matter, there is no list of rejected licenses. There is no public explanation of acceptance or rejection, nor is there an open process of review. It is not a case of the OSI reckoning that its work as a promoter of Open Source has ended, and fading quietly; it is currently said to be rejecting the most recent of the Sun licenses, and Sunís open challenge to the rejection may possibly open up the process. Unless the OSI clarifies its standards, it runs the risk of becoming irrelevant.

Entering the world of Open Source software, then, means entering a territory with indefinite borders controlled by more or less complex licenses. This should not dismay the adventurer; with care, there are worthwhile commercial opportunities.

Next chapter: Commercializing Open Source Software

Table of Contents

Copyright © 2000 by IDG Books Worldwide, Inc. Published under IDG Books Worldwide, Inc. Open Content License.