The Term 'Open Source' Is Proposed and the OSI Forms in 1998

The phrase “open source” became a public advocacy label in 1998, but neither the practice of sharing source code nor the idea of software freedom began that year. What changed was the vocabulary and its organizing institutions. The catalyst was Netscape’s January announcement that it planned to publish source for its Communicator browser; Christine Peterson proposed the new phrase, while Eric Raymond, Bruce Perens, Tim O’Reilly, and others helped circulate it. The Open Source Initiative (OSI) formed later that February and based its Open Source Definition on Debian’s Free Software Guidelines. The record contains a small but meaningful date discrepancy over the naming meeting, so a careful chronology should preserve it rather than manufacture a single uncontested day.

People had shared source code and built software collaboratively long before 1998. The OSI’s own history places the label in a movement already gaining wider attention through Linux and Netscape’s decision. The history is therefore not “Netscape invented open development.” It is a story about a large commercial vendor’s public decision creating an unusually visible moment for explaining collaborative development to companies, users, and the press.

The word free caused a practical communications problem in English: some readers heard “zero price” rather than the freedom to study, modify, and redistribute software. Peterson’s retrospective account says the effort to find a clearer term began in late 1997 at meetings about computer security and reliability. She describes her own reason as reducing that ambiguity for newcomers, not rejecting the ethical case for software freedom. That distinction is important: her motivation, the strategy of other advocates, and the later philosophy of the organizations that adopted the term are related, but not identical.

The 1998 sequence and a date that is not settled

Date	What the sources establish
January 22	Netscape announced plans to make the next-generation Communicator source available; this was an announcement of a future release, not the day the code was published.
February 2	In Peterson’s recollection, Raymond arrived to work with Netscape on the release plan, and a discussion at Foresight Institute considered how to describe the development model.
February 3 / February 5	OSI’s institutional history dates the Palo Alto strategy session to February 3. Peterson’s first-person account, written in 2006 and published later, describes a February 5 meeting at VA Research where the phrase was introduced and informally accepted. The accounts agree on the people and the broad sequence, but not this exact meeting date and setting.
Late February	Raymond and Perens jointly founded OSI. Netscape also announced mozilla.org, giving the forthcoming code release a project site and an organizational home.
March 31	Netscape made the first developer release of the Communicator source code available and launched the Mozilla Project.
April 7–14	Tim O’Reilly’s gathering was announced in advance as a “Freeware Summit”; his April 14 release called it an “Open Source Summit” and recorded the participants’ agreement to use “open source” or “sourceware.”

In Peterson’s account, she had developed the phrase between meetings, then Todd Anderson tried it in conversation at VA Research. The group’s discussion later returned explicitly to terminology, with Raymond endorsing the label. This was not a formal standards vote, nor does the recollection show that one person alone created the broader project. The official OSI history summarizes the naming as a February 3 Palo Alto strategy session; Peterson gives a more granular account that places the key VA discussion two days later. Retrospective institutional histories and participant memories can differ in such details, and neither should be silently rewritten to match the other.

The distinction between announcement and release matters too. On January 22 Netscape said it intended to release source. In the following weeks, the company had to prepare code and licensing, and to create a structure for outside contributions. Mozilla’s historical timeline records March 31 as the actual first public developer release. Treating the January press announcement as if it were the public code release collapses planning, publication, and project launch into one event.

From Debian policy to an open-source definition

The label became useful only if people could tell what it meant. In 1997 Debian developers refined Bruce Perens’s Debian Free Software Guidelines (DFSG) as policy for deciding which works could be included as free software in Debian. Perens later removed Debian-specific references and adapted those criteria into the Open Source Definition during OSI’s launch in February 1998. The definition is a set of licensing conditions, not a statement that merely having source files online makes a program open source.

In broad terms, the criteria require permission for redistribution, access to preferred source, the right to make and redistribute derived works, and no discrimination against people, groups, or fields of use. Other clauses address how licenses travel with redistributed copies, avoid tying rights to a specific product or distribution, and avoid restricting unrelated software. These are the reasons the label has a technical meaning beyond “inspectable code.” Source-available programs can allow inspection while prohibiting redistribution, commercial use, or modified versions; that alone does not satisfy the OSI definition.

The OSI was founded jointly by Raymond and Perens in late February to advance the label and steward the definition. Its later formal license-approval process made the certification question more concrete: one asks what rights the license grants, not just whether a repository is public. In 1998 the term and the definition were still being explained and refined in public; the modern list of OSI-approved licenses came later, with the first formal list published in 1999.

The summit made the business case visible

O’Reilly’s April 7 meeting brought together developers and maintainers associated with Linux, Perl, Python, Tcl, Apache, Sendmail, BIND, Mozilla, and other infrastructure projects. The contemporary April 14 press release is especially useful because it captures the language at the time: the event had first been announced as a freeware gathering, but its organizers subsequently called it the Open Source Summit. Attendees argued that commercial activity such as support, services, customization, and products could coexist with source distribution. They also described peer inspection as a potential quality benefit.

Those were advocacy claims made by the participants, not a guarantee that public code is automatically secure, reliable, or superior. The relevant mechanism is conditional: users must have the legal rights to inspect and change the source; people with the necessary skills must actually review it; maintainers must integrate fixes; and projects need sustained resources. Visibility alone does not perform an audit or create a healthy development community. O’Reilly’s contemporary release is evidence of how advocates pitched the model, while modern analysis should distinguish that pitch from measurable outcomes in any specific project.

Shared licensing ground, different arguments

“Free software” and “open source” usually describe overlapping software licenses, but the labels developed partly to emphasize different reasons for supporting those freedoms. OSI presents open source as a practical development and licensing framework. The Free Software Foundation (FSF), whose movement predates the 1998 label, emphasizes users’ ethical freedom to run, study, modify, and share software. The FSF has explicitly criticized the open-source framing for foregrounding practical benefits while obscuring that moral argument.

That disagreement is not accurately reduced to “the same licenses, just different words,” nor to the claim that the two camps never work together. The OSI says the definitions largely converge in practice; the FSF notes that most open-source software is also free software while identifying some edge cases and differences in criteria. Individual developers may use both terms, one term, or neither. In historical writing, it is more precise to distinguish the shared body of code and licenses from the debate about what values the movement should communicate.

What the Netscape outcome proves and what it does not

The source release launched the Mozilla Project, but it did not produce a finished browser overnight. Mozilla’s own history records contributions during the first year, the release of Mozilla 1.0 in 2002, and Firefox’s 2004 launch and subsequent reach. Early Mozilla development also involved substantial architectural work: the project roadmap in 1998 already described replacing much of the old Communicator code with a new layout engine and front end. This is a useful counterweight to the simplified story that publishing source automatically turns an existing commercial product into a thriving community project.

The more defensible legacy is twofold. Netscape’s announcement made source release by a major vendor a headline event, and the advocacy campaign gave people a concise term tied to a written licensing definition. Neither event created the older traditions of free software or collaborative development. The episode instead shows how a label, a license test, an institution, a corporate release, and a community can reinforce one another while still leaving open the philosophical question of why users should have software freedoms in the first place.

Related:

Sources:

A label for a moment, not the birth of sharing

The 1998 sequence and a date that is not settled

From Debian policy to an open-source definition

The summit made the business case visible

Shared licensing ground, different arguments

What the Netscape outcome proves and what it does not

Comments