Chapter 7 The Pragmatic Turn and the Open Source Initiative (circa 1997-1998) (c. 1997)

The document was a single page, typed, with handwritten annotations crowding the margins. It bore no official letterhead, no corporate logo, only a title at the top: “Meeting Agenda.” The listed location was a nondescript office in Mountain View, California. The date was February 3, 1998. The subject was tersely stated: “Strategy session on free software advocacy.” The list of attendees included names that were, within a certain sphere, becoming coordinates of influence: Eric S. Raymond, Bruce Perens, Tim O’Reilly. The items for discussion were pointed, operational: “Problems with ‘free software’ terminology,” “Need for new label,” “Outreach to business press,” “Coordination of message.” This was not a manifesto born of moral outrage, nor a license crafted to protect a freedom. It was a campaign plan. It recorded, in the sterile language of a to-do list, the moment a movement decided to re-engineer its public face. The agenda was a blueprint for a coup in plain sight, a calculated effort to translate an ideology into a market proposition.

The pressure was multifaceted, born of both frustration and opportunity. The commercial ascent of the Internet, chronicled in the preceding chapter, had created a new class of technology executives and investors who were intensely pragmatic, focused on time-to-market, total cost of ownership, and return on investment. To them, the term “free software” was a minefield of misinterpretation. It whispered of anticommercial ideology, of software that was merely gratis, or worse, of a movement whose primary spokesman, Richard Stallman, spoke in terms of ethical imperatives that sounded alien in boardrooms. This perception gap was not merely theoretical; it had tangible consequences. Venture capitalists were hesitant to fund companies built around GNU/Linux. Corporate IT managers, intrigued by the stability of Linux servers, faced internal resistance from legal and procurement departments wary of anything labeled “free.” The community’s own successes—the demonstrable quality of the software being produced—were being hamstrung by the baggage of its founding nomenclature. The Mountain View meeting was, therefore, a crisis response to a ceiling of adoption, a gathering of tacticians who believed the underlying methodology deserved a wider stage and required new marketing to claim it.

The intellectual groundwork for this marketing, as the draft notes, had been laid by Eric Raymond’s “The Cathedral and the Bazaar.” Raymond had first presented his thesis at the annual Linux Kongress on May 27, 1997, in Würzburg, Germany. Yet Raymond’s essay was itself a product of a specific, fertile moment within the community. The explosive growth of Linux throughout the mid-1990s, fueled by the Internet’s proliferation, had provided a living, large-scale refutation of the old proprietary model. It was a proof-of-concept that demanded explanation. Raymond, an adept storyteller and keen observer, did not discover new principles so much as he articulated, in a compelling and accessible metaphor, the practices that had evolved organically within projects like the Linux kernel. His genius was in framing. By contrasting the slow, solemn, and secretive “cathedral” with the vibrant, chaotic, and productive “bazaar,” he translated a social phenomenon into a business case. The essay meticulously avoided the language of the GNU General Public License’s “freedom.” Instead, it talked of debugging efficiency, reliability through peer review, and accelerated innovation. It presented Linus Torvalds not as a philosophical leader but as a project management visionary. This was a language Silicon Valley could understand; it spoke the vernacular of engineering superiority and disruptive innovation. When Raymond presented these ideas in Würzburg, and later as his paper circulated on the nascent web, it provided a coherent narrative for corporate adopters who were already using the software but lacked a sanctioned rationale for doing so.

The catalyst that transformed this narrative into a formal campaign, however, was a single corporate announcement. On January 22, 1998, Netscape Communications, embroiled in the “browser wars” with Microsoft, made a startling decision: it would release the source code for its Netscape Communicator suite. This move was a desperate competitive gambit, but it was framed as an embrace of the “open” development model. For the free software community, it was a seismic validation and an unprecedented opportunity. Here was a major, publicly-traded software company seemingly adopting their playbook. Yet Netscape’s press releases and executive statements conspicuously avoided the term “free software.” They spoke of “open standards” and “community development.” This dissonance highlighted the very problem the Mountain View attendees had been discussing informally for months: the existing movement brand was too toxic for the mainstream commercial world, even when that world was trying to adopt its methods. The Netscape moment created a pressing need for a ready-made, corporate-friendly alternative philosophy and label. It created a market for the product the Mountain View group was preparing to sell.

The discussions on February 3rd were thus intensely practical, focused on linguistic engineering. The term “free software” was diagnosed as suffering from fatal ambiguities: “free” as in speech versus “free” as in beer, a distinction Stallman constantly clarified to little broad effect. The group debated alternatives. “Sourceware” was considered. “Open,” however, emerged as the optimal adjective. It was already in positive use in commerce (“open standards,” “open systems”) and carried connotations of transparency, accessibility, and lack of obstruction. Paired with “source,” it precisely described the fundamental practice—making source code available—while sidestepping the ideological minefields of “freedom” and “free of charge.” This was not an accident but a calculated stripping away of moral complexity in favor of a clear, replicable criterion. The goal was to create a term a CEO could say to shareholders, a manager could write in a procurement report, and a developer could pitch to a venture capitalist without inviting a lecture on software ethics.

Bruce Perens, a key figure in the Debian project, played a crucial role in operationalizing this newly framed concept. Debian, a entirely free software GNU/Linux distribution, had already confronted the need for precise, enforceable definitions to guide its package selection. Perens had authored the “Debian Social Contract” and the “Debian Free Software Guidelines” (DFSG), a list of criteria that defined what constituted “free” software for Debian’s purposes. Recognizing the need for a stable foundation for the “open source” label, Perens simply removed the Debian-specific references from the DFSG. The result was the “Open Source Definition” (OSD), which established ten criteria, notably including non-discrimination against persons, groups, or fields of endeavor. This was a masterstroke of institutional translation. It took a practical, license-focused document born from the needs of a community distribution and repurposed it as the constitutional bedrock of a new commercial-friendly movement. The OSD provided the “open source” brand with immediate substance and legal credibility; it was no longer just a marketing term but a defined set of principles that software licenses could be measured against. This allowed the nascent Open Source Initiative (OSI), founded shortly after the February meeting, to act as a certifying body, granting its imprimatur to licenses that met the OSD criteria and thus creating a recognizable standard for the commercial world to rely upon.

Tim O’Reilly’s influence provided the crucial channel to that commercial world. As a publisher whose technical books served the developer community, OReilly & Associates was a nexus between the hacker ethos and the professional software industry. OReilly understood markets, messaging, and the power of convening. His role was that of evangelist and networker. He used his platform and his “Perl Conference” (soon rebranded the “Open Source Conference”) to bring corporate executives into the same room as community leaders, creating a dialogue that had previously been almost non-existent. O’Reilly advocated for “open source” not as a philosophy but as a superior business model, arguing that value in the network age would accrue to services, branding, and customization, not to hoarded code. This argument resonated deeply with companies struggling to adapt to the Internet’s disruptive force. O’Reilly’s pragmatism helped translate Raymond’s engineering arguments into boardroom strategy, framing open source as a smart competitive tactic rather than a moral compromise.

The formal launch of the Open Source Initiative and the promulgation of the “open source” label in the spring of 1998 was met with both enthusiasm and resistance within the existing community. For many developers who were focused on solving technical problems and building useful tools, the new terminology was a welcome relief. It allowed them to talk about their work without first engaging in a political debate. It provided a clear, non-confrontational way to advocate for their development methodology within their own companies. For the business press, always in search of a new trend, “open source” was a perfect story: a disruptive, geek-driven innovation that challenged Microsoft’s dominance. The label spread with astonishing speed through publications like The Wall Street Journal and Forbes, carrying with it the simplified narrative of superior, community-powered engineering.

This rapid adoption, however, underscored the fundamental nature of the pragmatic turn. The rebranding was a strategic separation of methodology from ideology. The “open source” argument explicitly chose not to lead with the user’s freedom to study, modify, and share. Instead, it led with the producer’s benefit: better quality, faster development, lower cost, and reduced vendor lock-in. It was an argument tailored for the software vendor and the corporate IT consumer, not primarily for the individual user or the citizen in a digital society. This focus created a tension that would persist. Richard Stallman and the Free Software Foundation (FSF) viewed the new term with deep suspicion, arguing it threw the baby of freedom out with the bathwater of unpopular rhetoric. Stallman would famously call the two camps “open source” and “free software” separate political camps within the same community, a schism rooted in this very difference of emphasis: practical advantage versus fundamental liberty.

The success of the campaign, measured by its own pragmatic goals, was undeniable. Within months, “open source” became the default term in business and mainstream technology discourse for describing the development model of Linux, Apache, Perl, and countless other projects. It provided the conceptual cover corporations needed to invest heavily in what was once a fringe hobbyist activity. IBM’s subsequent, billion-dollar embrace of Linux would have been almost unimaginable under the banner of “free software.” The rebranding unlocked capital, legitimized the methodology in the highest echelons of industry, and set the stage for the dot-com era’s widespread exploitation of open source infrastructure. It achieved this by performing a careful act of social translation, repackaging a set of practices born from a radical ethos into a palatable, even compelling, proposition for the engine of global capitalism. The movement had not abandoned its tools or its licenses, but it had deliberately crafted a new story about them—a story of winning, not just of being right.

The Netscape announcement did more than create an opportunity; it imposed a deadline. The Mountain View group now had a concrete, high-profile case to which their rebranding could be immediately applied. Discussions moved swiftly from theory to implementation. In the weeks following the February meeting, the cadre worked to formalize the artifacts of their campaign. Bruce Perens’s adaptation of the Debian Free Software Guidelines into the Open Source Definition was the foundational legal step, but it required a supporting infrastructure of advocacy and explanation. Eric Raymond, leveraging his newfound prominence from “The Cathedral and the Bazaar,” began drafting a companion piece, “The Halloween Documents,” which would soon analyze leaked Microsoft memos to argue that the software giant itself feared the open-source methodology. This was strategic framing of the highest order: it positioned “open source” not as a fringe ideology but as a competitive threat significant enough to worry the most powerful player in the industry. The rebranding was thus being armed with both a positive definition of what it was and a negative case against what it opposed, a classic marketing one-two punch designed for maximum impact in trade press and corporate corridors.

The choice of “open source” over other contenders like “sourceware” or “freely distributable software” was subjected to a deliberate market-testing logic. The group understood that the term needed to survive the gauntlet of corporate legal departments, marketing teams, and press reporters. “Open” had a pre-existing positive valence in business computing, from “open systems” to “open architecture,” suggesting interoperability and vendor neutrality. “Source” was technically precise, pointing unambiguously to the availability of code. Crucially, the phrase was adjective-noun (“open source software”), making it a modifier of a product category rather than the name of a movement. This grammatical structure subtly reinforced the desired perception: this was a quality of software, like “high-performance” or “secure,” not a political affiliation. The linguistic engineering was precise. It aimed to make the methodology describable in a single, uncontroversial line in an annual report or a press release, divorcing the practical act of sharing code from the historical context of the Free Software Foundation’s crusade for user sovereignty.

This calculated framing inevitably produced friction with the community’s established guardians. Richard Stallman’s reaction was both principled and strategically acute. He recognized that the “open source” pitch deliberately marginalized the ethical dimensions he considered paramount. For Stallman, the issue was never merely about development efficiency or cost savings; it was about preserving user autonomy and preventing digital feudalism. He argued that by focusing solely on practical benefits, the open source advocates were making a Faustian bargain. Corporations might adopt the methodology, but they would do so selectively, embracing open source for infrastructure like operating systems and web servers while keeping application software firmly proprietary. The freedom of the end-user would remain unprotected. Stallman famously began advocating for the term “free and open source software” (FOSS) as a compromise, but the hyphenated phrase lacked the crisp, market-ready utility of “open source.” This tension was not merely semantic; it represented a fundamental split in objectives. The Open Source Initiative sought to make the methodology permissible within capitalism. The Free Software Foundation sought to make that methodology a tool for transforming capitalism.

The early efforts of the OSI, therefore, involved not only promotion but also boundary-policing. As the term gained currency, it risked being diluted. Companies with only a superficial commitment to the model might label anything with a viewable snippet of code as “open source,” a practice known as “openwashing.” The OSI’s certification of licenses against the Open Source Definition became a critical regulatory mechanism to prevent this. By maintaining a definitive list of approved licenses—the GPL included—the OSI provided a needed service to the corporate world: a trust mark. A company could adopt an OSI-approved license and be confident it was playing by the recognized rules, thus avoiding legal uncertainty. This institutional function was a direct response to the very corporate fears the rebranding sought to allay. It turned the philosophical complexity of software freedom into a manageable compliance checklist, another reflection of the pragmatic turn’s core ethos: reduce friction, accelerate adoption.

The speed of that adoption was catalyzed by the existing ecosystem of trade publishing and conference organizing, where Tim O’Reilly’s role proved indispensable. O’Reilly did not just evangelize; he created venues for the new narrative to be performed and normalized. The rebranding of his “Perl Conference” to the “Open Source Conference” in 1998 was a symbolic capitulation to the new terminology’s momentum, and the event became a crucial neutral ground. Here, corporate engineers from Intel or IBM could sit alongside volunteer Apache contributors, with the shared language of “open source” providing a common frame that “free software” might have fractured. O’Reilly’s publishing arm further amplified this, producing books that framed open source projects like Linux and Perl through the lens of practical utility and market disruption, further entrenching the new narrative in the professional developer’s mindset. This ecosystem provided the channels through which the carefully crafted message of the Mountain View group flowed outward, reaching decision-makers who would never have attended a Free Software Foundation event.

The success of this concerted campaign was visible within a startlingly short timeframe. By late 1998, major technology journals that had previously ignored or marginalized the free software community were running cover stories on the “open source revolution.” The rhetoric in these articles consistently echoed the points engineered months earlier: emphasis on reliability, cost savings, and innovation speed, with mentions of “freedom” relegated to quotes from community holdouts. This media amplification created a feedback loop. As more businesses publicly adopted the term, its legitimacy grew, making it easier for the next wave of companies to follow. The rebranding had effectively built a conceptual on-ramp for corporate America, allowing it to embrace the output of the collaborative community without having to confront the community’s original, more challenging philosophy. This was the precise outcome the February strategy session had plotted: a terminological vessel sturdy enough to carry the methodology into the heart of the commercial world it had once sought to undermine.

Its existence testified to a profound and pragmatic conclusion: the methodology worked, but its original name had become a liability. The pressure to scale within a capital-rich world demanded a new vocabulary. That vocabulary had already found its first and most potent articulation in another document, circulated less than a year earlier. The intellectual catalyst for the Mountain View meeting had been a paper presented by Eric Raymond on May 27, 1997, at the Linux Kongress in Würzburg, Germany. It was titled “The Cathedral and the Bazaar.” Raymond’s essay was a deliberate polemic, framed as an engineering case study drawn from his management of the Fetchmail project. Its power lay in its framing of a dichotomy. The “cathedral” represented the old, proprietary model of software construction: built slowly, in secrecy, by a small, closed order of experts. The “bazaar” was the open, Linus Torvalds-led model of Linux development: a noisy, public, and seemingly chaotic marketplace where code was developed by a vast, distributed network, its quality enforced by peer scrutiny.

Raymond’s central claim was pragmatic and revolutionary: “Given enough eyeballs, all bugs are shallow.” This maxim, dubbed “Linus’s Law,” was not a statement about liberty or ethics. It was an argument from efficiency, a claim that the bazaar model produced more robust, reliable, and ada.