Chapter 4: Open as in Business

In January 1998, Netscape did something no major software company had ever done. It announced that it would release the source code for Navigator, its web browser — the product that had defined the early internet and that was now being crushed by Microsoft's Internet Explorer.

Netscape was losing the browser wars. Microsoft had bundled Internet Explorer with Windows, giving it an insurmountable distribution advantage. Netscape's market share was collapsing. Opening the source code was a Hail Mary — and it sent shockwaves through the technology world. For years, the free software movement had argued that closed, proprietary code was inferior to code developed in the open. Now a publicly traded company, under existential pressure, was about to test that theory at industrial scale.

The question was what to do with the moment.

On February 3, 1998, a group gathered in the conference room of VA Research in Mountain View, California — a short drive from Netscape's Palo Alto headquarters — to discuss exactly that. The meeting was organized by Eric Raymond, whose essay "The Cathedral and the Bazaar" had directly influenced Netscape's decision to open its source. Raymond had sent the essay to Netscape executives; they had circulated it internally; it had helped tip the balance toward what would become the Mozilla project.

The people in the room were among the most influential figures in the free software world. Raymond himself, the movement's most visible evangelist to the business community. Bruce Perens, who had authored the Debian Free Software Guidelines, which would become the Open Source Definition. Michael Tiemann, who had built Cygnus Solutions, one of the first companies to generate revenue from free software. Jon "maddog" Hall, executive director of Linux International and a longtime free software advocate. Larry Augustin of VA Research, who was hosting the meeting. Sam Ockman, another Linux entrepreneur.

The gathering had a practical question at its center: How do we capitalize on the Netscape moment? How do we convince other companies to follow Netscape's lead? The attendees knew that Netscape's announcement was an opening — possibly the best opening the movement would ever get to break into mainstream corporate adoption. But they also knew that the movement had a branding problem, and that branding problem had a name.

The name was "free."

The answer came from someone who was not a software developer at all. Christine Peterson was a nanotechnology researcher and co-founder of the Foresight Institute, a think tank focused on emerging technologies. She had been thinking about the naming problem for some time before the meeting, and she came prepared with a suggestion.

The problem, she argued, was linguistic. In English, "free" is hopelessly ambiguous. It means both "without cost" and "without restriction." Richard Stallman's careful distinction — "free as in speech, not free as in beer" — was philosophically precise and practically useless. It required a paragraph of explanation every time you said it. Worse, even after the explanation, the word "free" lingered in a businessperson's ear. It sounded anti-commercial. It sounded like the software was worthless. It sounded, to a corporate executive evaluating vendor relationships, like ideology.

Peterson suggested a replacement: "open source." In a first-person account she originally wrote in 2006 and published on Opensource.com in 2018, on the twentieth anniversary of the meeting, Peterson explained her reasoning: the term "free software" was distracting because of its focus on price. The industry needed a term that would focus on the source code itself and not confuse people who were new to the concept. "Open source" fulfilled those requirements.

The term was not hers alone in every sense — the phrase existed before the meeting — but she was the one who proposed it for the movement, in this room, at this moment. The group debated it. Todd Anderson, another attendee, helped refine the framing. Within days, Raymond and others began using it publicly. The term stuck. And in that renaming, something fundamental shifted.

It is worth pausing on the fact that a woman coined the most consequential term in the history of software. The free and open-source world has been, for most of its history, overwhelmingly male — in its demographics, its culture, and its public narratives. Stallman, Torvalds, Raymond, Perens: the canonical story is told almost entirely through men. Christine Peterson's contribution — the strategic insight that the movement's language was its greatest barrier to adoption — reshaped the entire industry. It deserves more recognition than a footnote.

"Open source" was a marketing decision. This is not a criticism — or not only a criticism. It was a strategic rebranding executed with remarkable skill, and it worked.

The word "free" carried baggage that the word "open" did not. "Free" implied ideology, radicalism, the FSF, Stallman's uncompromising moral stance. "Open" implied transparency, collaboration, pragmatism, good engineering. "Free" was a philosophy. "Open" was a methodology.

In February 1998, Raymond and Perens founded the Open Source Initiative to promote the new term and steward its definition. Linus Torvalds endorsed it the following day — a critical stamp of legitimacy from the most famous developer in the world. Torvalds had never been comfortable with the ideological weight Stallman attached to software freedom. He had always described his motivations in practical terms: Linux was a technical project, not a moral crusade. The term "open source" suited him perfectly.

In April 1998, Tim O'Reilly organized the "Open Source Summit" — a gathering of the leaders of major projects. Torvalds was there. Larry Wall, creator of Perl. Brian Behlendorf, co-founder of the Apache web server. Guido van Rossum, creator of Python. The summit was a coming-out party for the rebranded movement, and it placed the emphasis squarely on practical excellence — these projects produced software that corporations actually depended on.

The pitch to business was straightforward: open source produces better software, faster, at lower cost. You get transparency (you can inspect the code), reliability (thousands of eyes finding bugs), and no vendor lock-in (you're never beholden to a single company's roadmap). These are business arguments. The Four Freedoms didn't come up much.

Richard Stallman watched this unfold from Cambridge, Massachusetts, and refused to participate.

His objection was not tactical but philosophical, and he has articulated it with extraordinary consistency for nearly three decades. The open source camp, he argues, asks: "How do we make better software?" The free software camp asks: "How do we respect users' freedom?" These are not the same question. They sometimes produce the same answer — the same license, the same code, the same development practices. But they diverge precisely at the moments that matter most: when respecting freedom is inconvenient, expensive, or commercially disadvantageous.

Stallman saw the rebranding as a deliberate amputation. The Four Freedoms — to run, to study, to redistribute, to modify — were not engineering principles. They were ethical principles, grounded in a vision of what human beings owe each other when they share tools. To strip those principles out and replace them with efficiency arguments was, in Stallman's view, to gut the movement of the only thing that made it a movement rather than a methodology.

In his canonical essay on the subject, "Why Open Source Misses the Point of Free Software," Stallman drew the line with characteristic precision: "The two terms describe almost the same category of software, but they stand for views based on fundamentally different values." When we call software "free," he wrote, we mean that it respects the users' essential freedoms. The idea of open source, by contrast, is that allowing users to change and redistribute the software will make it more powerful and reliable — a claim about engineering outcomes, not about rights. The main initial motivation of those who split off the open source camp, Stallman observed, was that the ethical ideas of free software made some people uneasy — and by keeping quiet about ethics and freedom, they might be able to sell the software more effectively to business.

The "free as in speech, not free as in beer" formulation, which Peterson and the open source camp found cumbersome, was precisely the point for Stallman. The distinction was supposed to be difficult. It was supposed to force a conversation about what "freedom" means — not freedom-to-download, but freedom-to-control-your-own-computing. If explaining the distinction was awkward, that was because the concept itself required moral seriousness. Removing the awkwardness meant removing the seriousness.

Stallman also objected to something subtler: the implicit message that the movement needed corporate approval to succeed. The free software movement, whatever its limitations, was rooted in an ethical claim that stood on its own. You didn't need IBM or Netscape to validate it. The open source rebranding inverted this: it made corporate adoption the measure of success. And once corporate adoption became the goal, the movement would inevitably reshape itself to serve corporate interests.

He was right about that, at least.

In the short term — roughly 1998 to 2020 — the open source rebranding was an unqualified triumph. It opened the floodgates of corporate participation. Companies that would never have touched "free software" — with its whiff of anti-capitalism — embraced "open source" enthusiastically.

The milestones came fast. In 1999, Red Hat went public. Its IPO was the eighth-largest first-day gain in Wall Street history at the time — a company built entirely on free software, valued by the market at billions of dollars. The message to corporate America was unmistakable: there was real money in open source.

Then came IBM. In 2000, IBM announced it would invest one billion dollars in Linux — an almost incomprehensible sum to commit to a project that no single company owned or controlled. IBM's bet was strategic: it saw Linux as the platform that would undermine Microsoft's dominance in enterprise computing, and it was willing to pay a billion dollars to accelerate that outcome. The investment legitimized open source in boardrooms where the word "free" would have gotten you escorted out.

The most dramatic conversion, though, was Microsoft's. On June 1, 2001, Steve Ballmer, Microsoft's CEO, told the Chicago Sun-Times that Linux was "a cancer that attaches itself in an intellectual property sense to everything it touches." The metaphor was deliberate: the GPL's copyleft provision, which requires derivative works to carry the same license, was in Ballmer's view a contagion that destroyed intellectual property wherever it spread.

For years, Microsoft operated under this posture. Internal memos (leaked as the "Halloween Documents" in 1998) had laid out a strategy of fear, uncertainty, and doubt aimed at Linux. Ballmer repeated the "cancer" line. Microsoft's lawyers aggressively defended Windows' monopoly. The company was, by any reasonable measure, the open source movement's primary adversary.

The reversal took fifteen years and a change of leadership. Under Satya Nadella, who became CEO in 2014, Microsoft began contributing to open-source projects. It open-sourced .NET, its flagship development framework. It released Visual Studio Code, which became the most popular code editor in the world, under an open-source license. And in 2018, Microsoft acquired GitHub — the platform where virtually all open-source collaboration happens — for $7.5 billion in stock, announced on June 4, 2018, and closed that October.

From "cancer" to a $7.5 billion acquisition in seventeen years. The journey tells you everything about what the 1998 rebranding accomplished.

The business case worked because it was true. Open-source software genuinely was better for many purposes — more secure, more reliable, more adaptable. The "bazaar" model of development, stripped of its countercultural trappings, turned out to be a superior engineering methodology for infrastructure software. The enterprise world didn't need to believe in the Four Freedoms to see the value in Linux, Apache, and PostgreSQL.

The numbers by the 2020s were staggering. Virtually every Fortune 500 company ran on open-source infrastructure. GitHub hosted over 400 million repositories. Linux ran on 100 percent of the world's top 500 supercomputers. Every Android phone. The vast majority of web servers. The cloud infrastructure of Amazon, Google, and Microsoft itself. The rebranding unlocked all of this.

But something was lost.

By framing open source as a methodology rather than an ethic, the movement surrendered the vocabulary it would later need to confront the hardest questions. When the discussion is about efficiency and quality — "open source produces better software" — there is no principled basis for saying "some things should not be opened." Methodologies don't have moral limits. Ethics do.

This distinction barely mattered when the technology in question was web servers, compilers, and databases. Nobody needed a moral framework to decide whether to open-source a load balancer. The question was purely pragmatic: does open development produce a better load balancer? Usually, yes. End of discussion.

But methodologies are tools, and tools are agnostic about the hands that hold them. The open source framework, stripped of Stallman's ethical architecture, had no way to distinguish between opening a web server (which makes everyone's life easier) and opening a surveillance system (which makes some lives easier and other lives much worse). The methodology says: open is better. The methodology does not say: better for whom?

Stallman's framework had an answer to that question. The Four Freedoms were centered on the user — the individual human being who runs the software. Freedom 0 was not "freedom for the developer" or "freedom for the corporation." It was freedom for the person whose life the software touches. This centering was the ethical core that the open source rebranding excised as an inconvenience.

The excision was understandable. It worked. It produced two decades of extraordinary growth and innovation. But it left the movement structurally unable to articulate why some kinds of openness might be dangerous — because danger is a moral category, and the movement had spent twenty-five years cultivating a vocabulary that was deliberately, proudly amoral.

The historical record suggests that the ethical dimension was deliberately set aside, not debated and lost. The attendees at VA Research were strategists, not philosophers. Their explicit goal was to capitalize on the Netscape moment by making free software palatable to corporations. Stallman was not invited and would not have attended if he had been. The decision to drop the ethical framing was not a compromise — it was the premise of the meeting. Bruce Perens, who co-founded the OSI with Raymond, would later express some ambivalence and briefly attempt to reconcile the two camps, but the strategic direction was set from the first day.

The rebranding also changed who the movement attracted — and who led it.

Stallman's free software movement was, for all its flaws, rooted in an ethical vision accessible to anyone. You didn't need to be a programmer to understand that users should have freedom. The open source movement, by contrast, was built for and by an engineering elite. Its arguments were technical. Its language was corporate. Its heroes were CTOs and VCs, not philosophers and activists.

This isn't inherently bad. Technical excellence matters. Corporate adoption brought resources, stability, and reach that the free software movement alone could never have achieved. But it created a cultural shift: the movement's center of gravity moved from "what is right" to "what works" — and from there, inevitably, to "what pays."

Eric Raymond and Bruce Perens, the co-founders of the OSI, represented this shift. Raymond's "Cathedral and the Bazaar" was fundamentally a management argument: here's why decentralized development produces better outcomes. Perens wrote the Open Source Definition — a technical standard for what qualifies as an open-source license. Both contributions were valuable. Neither was about ethics.

The corporate world responded by developing its own sophisticated relationship with openness — one governed entirely by strategy. Google open-sourced Android and Chromium; it did not open-source its search algorithm or ad-targeting systems. Facebook open-sourced React and PyTorch; it did not open-source its news feed algorithm or its content moderation models. Amazon built AWS on open-source databases; it did not open-source the infrastructure that made AWS profitable.

The pattern was consistent: companies opened their infrastructure and closed their competitive advantages. This was perfectly rational under the open source framework, which has no principle requiring otherwise. If openness is a methodology for producing better software, then you apply it where it serves your interests and decline to apply it where it doesn't. There is no hypocrisy here — only strategy. And strategy is exactly what the 1998 rebranding promised.

The result of the rebranding was a paradox that would take two decades to fully manifest.

Open source won. It became the default infrastructure of the digital world. It enabled a generation of startups to build billion-dollar companies on free foundations. It proved that collaboration at scale could produce extraordinary results.

But open source also became a tool — a strategy to be deployed when it served corporate interests and set aside when it didn't. This is the world the 1998 rebranding made possible. A world where "open source" is a business decision, not a moral commitment. A world where the question isn't "Is this right?" but "Does this serve our interests?"

For two decades, that was fine. Infrastructure software benefits from being open. The more people use Linux, the better Linux gets. The incentives were aligned, and nobody much needed to ask whether the alignment was a coincidence or a principle.

Then came AI.

And for the first time, the thing being potentially "opened" wasn't infrastructure that makes everyone's life easier. It was a technology that could, in the wrong hands, make everyone's life dramatically worse. A technology capable of mass surveillance, autonomous warfare, and the concentration of power at a scale the world has never seen. Suddenly, the question Stallman had been asking since 1983 — the ethical question, the question about freedom and responsibility and what humans owe each other — was the question that mattered most.

The movement that had spent twenty-five years cultivating a vocabulary of efficiency, quality, and business value found itself without the words it needed. When Meta releases the weights for a model capable of generating biological weapon instructions, the open source framework offers no guidance. "Will this produce better software?" is not the relevant question. "Should this be open?" is — and that is an ethical question the movement had deliberately, systematically, and with great commercial success trained itself not to ask.

This is where the 1998 story connects to the 2026 story. In Chapter 1, we watched Anthropic draw two ethical redlines — no mass surveillance, no autonomous weapons — and enforce them precisely because its model was not open. Anthropic could say no because it controlled the technology. That control was the ethical firewall.

The movement that Christine Peterson renamed in a conference room in Mountain View — the movement that stripped out Stallman's ethics to win corporate hearts — is now living in a world where those ethics were the thing it needed most. The pragmatism worked. The victory was real. And the bill is coming due.

Stallman has, in fact, weighed in — though not in the way either side might have hoped. In his 2024-2026 speaking tour, he launched a characteristically blunt critique of large language models, refusing to use the term "artificial intelligence" and preferring phrases like "pretend intelligence." On the specific question of whether the Four Freedoms apply to AI models, his answer was surprising: "Software licenses are not the right way to prevent a program from being used harmfully. That's not what software licenses are effective for." He argued that while learning from data is fair use, if an AI generates code that replaces the programmer, it violates the social contract of the commons. He also pointed out that almost all AI is delivered as Software as a Service — a model he has opposed for decades as fundamentally hostile to user freedom, regardless of whether the underlying code is open.