The Linux Community enjoys a popular reputation as a bunch of skylarking juveniles. The previous chapter tried to show where this picture comes from and how much truth might be in it. In this chapter we will look more closely at the Community’s ties to a project or set of projects with a much more sober public reputation: the creation of the Internet with public funds.
There is still an active community of computer experts who helped start the Internet and who keep it going now. One of their gatherings is USENIX, the meetings of UNIX users that have a club-like atmosphere because so many people have known each other for so many years. Here the newer Linux Community can look at the founders of the Internet and UNIX, and the UNIX crowd is more than a little aware that Linux is helping to revive UNIX.
When the Open Source Community looks at this side of its origins, it is both awed and emboldened: Open Source worked once and was universal; its time will come again. For the present the Community is astonished at the sudden wealth it has created, wealth that can be tangibly measured at the Red Hat Wealth Monitor (http://prosthetic-monkey.com/RHWM/). The ostensible motivation for posting the Monitor is to remind Red Hat to reinvest some of its capital in the Community; recently the Monitor’s computation showed the values listed in Table 2-1.
The Red Hat Wealth Monitor
Red Hat Market Capitalization
RH Market Capitalization Created by the Community
The values listed in Table 2-1 are subject to frequent fluctuation.
The capitalization is updated every quarter-hour while the market is open; the Community share in this wealth is reckoned as the percentage of code in Red Hat 6 not written by Red Hat (~87 percent). The number of contributors in the Community is reckoned at 5,000 persons, including documentation contributors and 1,000 kernel contributors.
Beyond its ostensible purpose, the Monitor operates at several levels. While it has a serious purpose of claiming some or most of Red Hat’s wealth as Community property, it is also an imitation or even parody of the Bill Gates Personal Wealth Clock (http://www.webho.com/WealthClock), which tracks Gates’s personal fortune and reckons the contributions to it on the basis of the entire U.S. population (some $376 per capita). The Red Hat Wealth Monitor is also awestruck at the size of the money pile measured, and shares in a tiny degree the resentment writ large on the Bill Gates Personal Wealth Clock site. At the same time, every rational Community member understands that the great wealth created by Linux IPOs is still measured in the perilous currency of start-up stocks.
The preoccupation with start-up wealth now rippling through the Community should adjust another popular idea concerning the Open Source Community: that it is a community of selfless or at least nonprofit bestowers, similar to the medieval monks who undertook public works like road repair. Of course there are men like Larry Wall, leader of an important Open Source software project (Perl) but not interested in becoming a CEO; but there are also men like Larry Augustin, founder and CEO of the recently IPO’d VA Linux.
The advantages of the nonprofit model include being able to improve and distribute code from many inspired sources; the advantages of the for-profit model include the ability to continue coding by using resources besides personal motivation. A hacker-centric view leaves out the noncoding part of the population, which includes people who can organize an economic engine to keep the coding going. Open Source software is always a work in progress. Hackers may see a program as an opportunity to become involved and contribute, but software vendors will see it as a deliverable product that must be fully developed to meet a sales deadline. Open Source software can benefit from both points of view.
The Internet goes back to the days of all-software-as-open-source; its origins lie in the national defense research and infrastructure building of the 1950s. It is the virtual parallel to the National Defense Highway System.
One of the early research centers funded by the federal government was at MIT; the technology emerging from the work there helped build the New England computer industry. Government-funded research on the other side of the country at Stanford gave us the early versions of e-mail, word processing, graphical interfaces involving windows and a mouse, and hypertext (the original and still popular form of linking on Web pages).
To tie together the various federal research sites, the Advanced Research Projects Agency of the Department of Defense (ARPA) spun the ARPANET across the country, eventually connecting universities and government research sites. To do so successfully, the network depended on interoperability standards such as TCP/IP and Ethernet. Not only were these standards developed by publicly funded researchers, but they received collaborative input in an Open Source manner as well. To guide standards development, standards bodies were formed. InterNIC, for example, regulated the address system on the network.
The development of open, interoperable standards that could evolve along with the progress in research was paralleled by the invention of an operating system that was not hardware-specific; that is, it was designed so that only a small part of its code would need to be written for the specific hardware it was to run on. Bell Labs was the home of this new operating system, UNIX. Its wide availability made it popular in universities and research centers, and it was rapidly improved by contributors in many locations.
The version of UNIX called the Berkeley Software Distribution (BSD) demonstrates the two themes of this chapter. First, Open Source software can spread rapidly and undergo development by many hands, while coexisting in a number of different versions. Second, this process does have its commercial application. A 1956 court decision kept AT&T (and with it, Bell Labs) out of the computer business, so that Bell Labs could be liberal in its distribution of UNIX code. In 1982 another court decision brought an about-face in AT&T’s attitude towards UNIX, and stiffer control and licensing fees. To telescope a complex story, the version of UNIX that had been developed at Berkeley was offered to the public, source code and all, but missing a few portions of code whose ownership was claimed by AT&T. It did not take many months before the open community of UNIX users supplied the missing code by writing it from scratch. These contributors allowed their work to be distributed under the BSD License.
Commercialization of UNIX had already taken place; the rising minicomputer industry had licensed UNIX to put on their hardware. Because those companies are largely forgotten now, it is interesting to look at one company that decided to put BSD on its hardware and proceeded to attack not the minicomputer, but the workstation market. That company, Sun Microsystems, is still with us. Because Sun based its original operating system on BSD, it has great affinity with Linux, which is an imitation of the BSD imitation of UNIX.
In the early 1990s the Internet began to be opened to the public. Originally intended only for researchers in research institutions, its use spread first inside those institutions to all sorts of persons who were not actually science researchers; e-mail was the primary motivation for its spread. Eventually a few institutions applied for waivers from the "appropriate use" policy of the Internet; that is, outsiders were allowed to connect. The most "inappropriate use" of all, commercial activity, soon followed. A new business was born, the Internet Service Provider (ISP), upon whose back many more Internet-based businesses would be built.
The Internet of today not only runs on the protocols developed long ago (such as TCP/IP); a number of the Open Source applications of that period are still with us. In all cases they are able to evolve because they are open. BIND (Berkeley Internet Naming Daemon) is the service that enables us to type a name instead of a series of numbers to reach a Web site, and Apache is the workhorse of the Internet, driving well over half the Internet servers in existence. The original Web server software was publicly funded; and the present Apache project and product was begun by a number of Internet technicians who freely traded their constant tinkerings and improvements to it. Realizing they had a tangle of patches ("a patchy program") on their hands, they decided to organize the project and turn out a new product, Apache.
Apache shows us several ways to build a business based on Open Source software. Although Apache is free and may be used and modified by anyone, much of the development on it is done by private companies. O’Reilly & Associates funds some of the development, making money in its turn by selling books on Apache; IBM funds even more Apache development as a means of extending the life of its legacy computer systems by making their data available. IBM and O’Reilly make their improvements to Apache publicly available. On the other hand, at least two companies have taken the Apache code and put out proprietary versions of the program that include their modifications and improvements, and some Apache experts have started a business that sells Apache support.
Apache and Linux (and many other programs) were made possible by the Internet, which enables rapid cooperation over global distances by an unlimited number of contributors. The Internet was and continues to be a subverter of old standards, today in commerce and yesterday in programming. In place of the carefully planned and sequenced programming of the mainframe ethos, the Internet encourages rough-and-ready improvisation, "No kings, no priests, just a rough consensus and running code." The Internet is based upon cross-platform interoperability standards, and encourages this open thinking. The ideal of Java is an Internet ideal.
Throughout the history of the Internet, however, there were always private firms seeking to take some portion of it into private and proprietary waters. The existence of so much publicly funded code that could be taken private and made proprietary is a rich temptation, and it can be done successfully. SAS Institute, for example, took a statistical analysis product that originated as a Department of Agriculture project and grew it into an enormous privately held corporation.
The Berkeley version of UNIX, BSD, was distributed with its source code and a license that allowed the recipient the options both of altering it and of distributing the altered version without the source code. Sun (for Stanford University Network) Microsystems was one early distributor. Originally the company did not charge for the software, but included it with every machine so that Sun users could run UNIX programs. Other entrepreneurs picked up BSD as well, leading to such varieties as FreeBSD, NetBSD, OpenBSD, and BSD OS. The latter version is never given away, and even charges extra to customers who want to have the source code; the others are available in both paid and unpaid versions.
The crucial difference between paid and unpaid versions is support. Researchers and hackers are delighted with Open Source software that costs nothing; if they have problems, they can always ask user groups on the Internet for assistance. Corporate users, on the other hand, face legal responsibilities to shareholders, and insist on clearly defined support programs and a vendor they can turn to for immediate help. This paid/unpaid split is a constant in Open Source software.
If the Apache Web server story has a happy ending as Open Source software, and BSD lives on as both Open Source and closed, the Mosaic Web browser and its descendants sound like a Greek tragedy in midplot. Building on browser code originating at CERN, a research center in Switzerland, the National Center for Supercomputing Applications (NCSA) at the University of Illinois developed its version, the Mosaic browser. According to Netscape founder Jim Clark, Mosaic was not even an official project, just something that some young people at NCSA loved to work on. Made available for free as a download, the Mosaic browser spread across the Web like wildfire. NCSA recognized its commercial potential and licensed Spyglass, Inc. as the exclusive master distributor for the purpose of licensing it to companies who could develop its full commercial value.
Jim Clark also recognized the potential of Mosaic, and hired the original Mosaic development crew to come to California and build an improved version for his new firm, Mosaic Communications (later called Netscape Communications), to market. Although Clark avoided using the NCSA code, which the developers could easily have brought with them, he had the advantage of their expertise, and so duplicated the functionality, if not the code, of Mosaic. Knowing that they had to beat the NCSA Mosaic browser to survive, the company adopted the classic proprietary strategy of product differentiation. The new browser, code-named Mozilla (a monster to destroy Mosaic), would improve and extend the functionality of the browser to take the market away from Spyglass. Deprived of its inventors, the original NCSA Mosaic could not be expected to keep up.
To overcome NCSA Mosaic, Netscape introduced a scripting language into its new browser, and then licensed Java from Sun so that Mozilla could run Java applets. These improvements were de facto extensions of the HTML standard to which browsers were supposed to conform. From a vendor’s point of view the ideal customer lock-up would be a proprietary server whose output could only be read with a proprietary browser and whose content could only be developed with a proprietary tool. This perspective directly opposes the fact that the Internet was successful because of the open standards that govern and support it. These standards are no secret, and they are controlled by cooperative bodies.
There is always a tension between standards bodies and commercial enterprises because enterprises want to differentiate themselves by doing things that their competitors cannot do. If the product does not meet recognized standards, but fills a need, customers will still buy it. Given the speed of entrepreneurial innovation and the deliberate slow pace of standards bodies as they hammer out consensus, it is no surprise that the gap between the two should widen when a technology is hot. Structured Query Language (SQL) is an example; the official standard is less functional than the various versions of it delivered by competing database vendors. Vendors know that control of a de facto standard, that is, a standard accepted by the market in place of an official standard, is the best marketing tool of all.
If Netscape was having these thoughts, so was Microsoft. Microsoft licensed the NCSA Mosaic browser from Spyglass to have a place to start, and eventually produced its own browser on the basis of having studied Mosaic. The Microsoft approach to competition involved packaging the browser with the operating system, thus eliminating browsers as an application category in the Windows world. Netscape, meanwhile, doubled its bets on Java and was developing the Internet Foundations Classes (these were turned over to Sun and emerged as Java Swing) as part of its strategy to have the browser take over the operating system on desktop computers. It was a battle that Netscape could not win. Microsoft’s browser, Internet Explorer, developed its own exclusive features, and the Web entered an uneasy era in which developers had to prepare two versions of Web pages to suit whichever browser happened to be visiting.
Netscape had expected an attack by Microsoft from the very beginning; the only question had been how long it would take for Microsoft to field a browser. As part of becoming the de facto browser standard, Netscape planned to have its browser so widely spread across the Web that by the time Microsoft appeared on the scene Netscape would have vanquished Spyglass’s NCSA Mosaic and have the lion’s share of the market — an advantage they hoped Microsoft would be unable to overcome. Key to the strategy of overwhelming the market with copies of the Netscape browser was the pricing structure. Because Netscape expected to make most of its money from Web servers, the browser was priced nominally at $39.95, but was completely free to educational users, and to any other user on a 90-day trial basis. No one seriously expected retail revenues, but the corporate browser market would provide some useful gravy. In fact, browser revenues were far higher than expected.
Microsoft’s strategy of bundling Internet Explorer for free with Windows effectively set the price of all sales of the Netscape browser at zero dollars. If Netscape tried to collect money, the user would simply switch to Internet Explorer. Jawboning, elbowing, and copious cash helped Microsoft spread its browser as widely as possible, and the linking of Internet Explorer with Windows was eventually featured in a federal court case against Microsoft. By the time Netscape faced the music and made their browser a giveaway item, the company was failing. This was the moment for the Open Source gesture that caught the world’s attention.
Seen from a worldly perspective the opening of the Mozilla source was not a glorious moment: it was simple desperation in the face of overwhelming defeat. Seen in the mythic light that so often seems to shimmer over Open Source events: it was the sad Hebrew slave mother putting the Baby Moses in the basket to be borne downstream to future glory, or Sieglinde dying in childbed as she delivers Siegfried.
The world is slowly coming to reassess the release of software assets as Open Source. Its early reactions to the practice ranged from skepticism to derision. Not so long ago certain VRML browsers were promised to the world as Open Source, even as they faded from sight, but nothing more has been heard about them. Some observers thought the release of a technology to Open Source by a failing company was an idea sold to management by developers as a public relations stunt, while the developers were planning all the while to pick up the technology once they had left the company. The adoption of Open Source by IBM and other healthy companies has changed public estimates of the practice.
Netscape was the first proprietary corporation to release its technology as Open Source; and from the point of view of Marc Andreessen, who had worked with Open Source software at NCSA, it was an obvious move. It was his experience at NCSA that helped shape the Netscape strategy of distributing the browser not in boxes in retail stores, but over the Web, and as freely as possible. Open Source practices were also the inspiration for Netscape to distribute beta software to all comers. While this strategy did not achieve the effect of development by a wide number of people, it did give an opportunity for a wide range of beta feedback.
The Open Source version of the Netscape browser returned to its aggressive internal name of Mozilla; this time Microsoft was the enemy. The only way of keeping a competitor browser alive was to release it as Open Source, so that even if Netscape perished, the browser could live on. Netscape also counted on the enthusiastic hands of a multitude of Microsoft-hating hackers to improve it. Towards the end of this book, we will take a look at what happened to the Mozilla effort, but it is now time to take a brief look at enmity to Microsoft as a motor of the Open Source movement.
The Open Source movement has its own objectives and ideals, and does not need to refer to Microsoft to practice them. Nevertheless, Bill Gates, Microsoft, and Windows appear throughout this book. The three are particular targets for many in the Community. Hackers deride "Windoze" as slow, bloated code that keeps users in a perpetual state of bondage (like the prisoners in the Apple 1984 commercial); software vendors fear the blistering competition, and even the federal government has taken exception to Microsoft business practices.
There is no need to go into details; a partial list of the casualties of the browser wars will do. Netscape’s plan was to develop its own browser so it would not have to pay royalties to Spyglass for NCSA Mosaic. This savings, like that achieved by relying on online distribution rather than on stores, was intended to keep Netscape’s price below the competition. Microsoft obtained a single-fee, nonroyalty license from Spyglass, negating the first part of Netscape’s competitive position. Putting Internet Explorer in the Windows box (and having it preinstalled on most new computers) solved the distribution problem for Microsoft, and further undercut Netscape. Making Internet Explorer completely free also made sure that the dwindling number of Netscape users were people willing to take the trouble to download it and (until Netscape finally decided to drop the browser cost to zero) pay for it. Finally, adding to Internet Explorer proprietary extensions that were based on Microsoft’s Web development tools meant that loyal Netscape users would find themselves visiting sites that would tell them to get Internet Explorer so they could enjoy the full functionality of the site, or (in the case of at least one site plugging a Hollywood movie) even be allowed to enter the site. By February 2000, site statistics indicated that market shares stood at Internet Explorer 80 percent and Netscape 20 percent.
Incidentally, but not unintentionally, Internet Explorer killed off NCSA Mosaic; faced with competition from a free browser, Spyglass customers ceased to deliver product and pay royalties. Other browsers and the businesses associated with them failed as well.
The software world is filled with the casualties of Microsoft competition. The return of Open Source provides an opportunity for those of them still able to lift a hand. Ray Noorda, whose Novell was battered by Microsoft, founded two Linux companies, Caldera and Lineo. Borland, now reorganized as Inprise, is converting its tools to Linux and contributing Open Source code. Corel, which assembled an office suite from battered pieces of Microsoft casualties, has decided that Linux is the best way to compete against Microsoft.
SGI has climbed aboard, and Hewlett-Packard is cooperating. Even the twin pillar of the Wintel universe, Intel Corporation, is taking positions in Linux companies. One benefit it may reap is that Linux is likely to be the only operating system ready to run on the new 64-bit Itanium chip when it appears later in 2000.
For many Open Source hackers, an assault on Bill Gates by commercial rivals using Open Source is especially welcome. They remember him as the young "Micro-Soft" executive who warned in his February 1976 "Open Letter to Hobbyists" that folks had better quit stealing (copying and giving away) his software. They also remember his statements in 1980, refusing to lower his profit margins to accommodate hobbyists, and pushing a new argument that the copyright laws also covered computer magnetic media. They find it easy to see him as the guy who invented proprietary software in the first place.
Next chapter: Benefits and Cost of Open Source in Your Business
Table of Contents
Copyright © 2000 by IDG Books Worldwide, Inc. Published under IDG Books Worldwide, Inc. Open Content License.