MongoDB's Blog Post Gambit

The legal geography of open source, once navigated by a few well-mapped principles, had by late 2018 become a terrain of contested borders and shifting fault lines. From a high altitude, the view showed a landscape undergoing tectonic stress.

The stable continental plates—the GNU General Public License (GPL) continent of reciprocal freedom, the broad, permissive plains of MIT and BSD, the settled territory of the Apache License—were now being pressured by new, rising formations. These new formations were not the product of slow, consensus-driven erosion. They were the sudden, deliberate constructs of entities fighting for survival on the lowland, where the very ground was being pulled from beneath them.

The immense gravitational force causing this upheaval was the concentrated mass of the hyperscale cloud. As detailed in the previous chapter, the ability of providers like Amazon Web Services to absorb open-source projects into their service fabric and monetize them without returning commensurate value to their origins had created an existential pressure. From this pressure, a new class of legal architecture was being engineered.

The descent from this aerial view of a fragmenting ecosystem leads directly to a concrete point of publication: a blog post from MongoDB Inc. On October 16, 2018.

The company announced it was replacing the GNU Affero General Public License (AGPL) for its core database software with a new, custom-built license: the Server Side Public License (SSPL). The announcement was framed not as a routine update but as a necessary defensive fortification for the entire open-source community.

The gauntlet, thrown in the preceding period of industrial squeeze, now defined a new, legal battlefield. The license had been reconceived as a strategic weapon.

The SSPL’s design was that of a precision instrument for a specific form of economic warfare. It operated by radically extending the familiar “copyleft” principle. Traditional copyleft, as in the GPL, required that if you distributed a modified version of the software, you must also distribute your modifications under the same free terms. The AGPL had already closed the “application service provider” loophole by applying this requirement to software accessed over a network.

The SSPL launched the requirement into a new dimension. Its pivotal clause stipulated that anyone offering the licensed software “as a service” must also make available under the SSPL not only their modifications to MongoDB itself, but “all programs that you make available to users… that work together with the SSPL-covered software to provide the functionality of the service as a whole.”

In practical terms, this meant that if Amazon Web Services offered its MongoDB-compatible “DocumentDB” as a managed service, the SSPL could theoretically obligate AWS to open source the extensive, proprietary scaffolding that made DocumentDB a viable service—its scaling orchestration, management consoles, monitoring systems, and integration layers. This was a “share-alike” demand aimed at the entire service stack, a direct legal attempt to dismantle the cloud providers’ core competitive advantage by forcing them to share their operational innovations.

MongoDB’s leadership, including CEO Dev Ittycheria, argued this was a matter of fundamental fairness and sustainability. The cloud giants were harvesting billions in revenue from open-source labor while contributing disproportionately little back to the ongoing development of that software.

The SSPL was engineered to compel either meaningful contribution or commercial retreat, thereby protecting the economic model of the primary corporate steward.

This maneuver signified a fundamental philosophical shift in the purpose of open-source licensing. The dominant philosophy since Richard Stallman’s GNU Manifesto had centered on ensuring user freedom—“free as in speech, not free beer.” Licenses were shields to protect those freedoms from proprietary enclosure; the GPL functioned as a constitutional document for a republic of code.

The new licenses emerging from 2018-2020, beginning with the SSPL, inverted this logic. They were designed first and foremost as economic shields for their creators, tactical instruments to protect developer sustainability from the extractive logic of platform capitalism. Their primary imagined “user” was no longer the individual programmer seeking liberty, but the corporate entity seeking a defensible market position. Freedom was subtly redefined. It was no longer an absolute, unconditional right to copy, modify, and redistribute, but a conditional right contingent on not undermining the original producer’s commercial viability.

This was License Drift manifest in its starkest form: the evolution of licenses from universal ethical mandates into tactical business tools that defined competitive moats and regulated the power dynamics within ecosystems.

MongoDB was not a solitary combatant in this defensive arms race. Just months earlier, in August 2018, Redis Labs had adopted a different, more blunt approach. It attached a “Commons Clause” to certain advanced modules of its Redis database. The Commons Clause was not a full license but a rider superimposed on an existing open-source license like the Apache License. Its language was starkly commercial: “The Software is provided to you by the Licensor under the License, subject to the following additional limitation: Neither the Software nor any other portion of the License may be used for any commercial purpose… by any organization with more than a certain number of employees or annual revenue.”

This was not a nuanced extension of copyleft; it was an explicit revocation of commercial use rights for large entities, with cloud providers squarely in its sights. The community backlash was swift and severe.

Many argued this clause effectively closed the software, violating the bedrock definition of “open source.”

By early 2019, Redis Labs retreated from this model to the more tailored Redis Source Available License (RSAL), but the precedent had been set: the old social contracts were fracturing under commercial strain.

The most strategically significant follow-on move came from Elastic N.V., steward of the ubiquitous Elasticsearch search engine and Kibana analytics platform. For years, Elastic had watched AWS successfully build and monetize its own Elasticsearch-compatible service, fully permitted by the project’s permissive Apache 2.0 license.

The tension escalated until, in January 2021, Elastic acted. It announced a transition for Elasticsearch and Kibana from Apache 2.

0 to a dual-license scheme: the Server Side Public License (SSPL) and a proprietary Elastic License. The Elastic License was explicitly “source available” and contained a clear field-of-use restriction: it prohibited offering the software as a managed service to third parties. The choice presented to users was deliberate and difficult.

They could adopt the SSPL version and face its stringent service-provider requirements, or they could adopt the Elastic License and avoid those requirements—but only if they themselves did not resell the software as a service. For any cloud provider, both paths were blocked. This was not merely a defense; it was a direct counter-offensive to reclaim commercial control.

The reaction across the open-source world was a storm of controversy that laid bare profound fissures in the very meaning of “open.” Proponents of the new licenses, largely from the commercial open-source company (COSC) sector, framed them as a necessary and overdue correction to a broken system. Some respected figures, like Bruce Perens (a co-founder of the Open Source Initiative), initially expressed sympathy for MongoDB’s stance, arguing the AGPL had failed to address the modern service-based loophole. The sustainability argument resonated deeply with many developers who saw their collective labor being leveraged for enormous profit by platforms that contributed little back. If open source was to remain an engine of innovation, its primary stewards needed economically viable models.

These licenses were portrayed as the defensive walls protecting the innovative citadel from parasitic siege.

Opponents, including many foundational community voices and free software advocates, condemned them as a betrayal of first principles. They argued these licenses categorically failed the standard definition of “open source” as maintained by the Open Source Initiative (OSI). The Open Source Definition (OSD) includes the critical mandate that “the license must not restrict anyone from making use of the program in a specific field of endeavor.” Licenses like the SSPL and Elastic License were deliberately engineered to restrict use in the field of “providing commercial software-as-a-service.”

This, critics charged, introduced a poison pill that shattered the universal reciprocity of true open source. It created a two-tiered system: one rule set for friendly users and another for competitive giants. By stepping outside the OSI-approved list, these companies were forking the concept of openness itself, creating categories like “source available” or “fauxpen source” software. These projects retained the outward appearance of community development—public source code, public repositories—while wielding proprietary control over key commercial pathways.

Richard Stallman’s long-standing distinction between “free software” (a social movement) and “open source” (a development methodology) suddenly gained acute, practical relevance. This was not freedom being defended, critics insisted; it was freedom being strategically suspended for market preservation.

The institutional arbiter at the center of this maelstrom was the Open Source Initiative. The OSI, guardian of the Open Source Definition, faced a direct crisis of authority and relevance.

MongoDB formally submitted the SSPL for OSI approval, seeking the community’s legitimizing imprimatur for its novel construct. What followed was a lengthy, public, and deeply divisive deliberation on the OSI’s license-review mailing list.

The debate became a philosophical battleground. Could a license that deliberately discriminated against a specific business model (SaaS) ever be considered “open”? Was the OSI’s role to guard immutable principles, or to adapt pragmatically to new economic realities? Supporters argued the cloud represented a novel form of “distribution” not contemplated in the 1990s, necessitating an evolution in the OSD.

Opponents held that bending the definition to accommodate one group’s commercial interests would destroy its universal utility and invite endless, self-serving license proliferation.

After nearly a year of debate, the OSI board rejected the SSPL in early 2020. Their formal determination was that it violated the OSD’s prohibition on field-of-use restrictions.

This was a landmark ruling. It drew a bright line, declaring that this new breed of defensive licenses existed outside the canonical realm of open source.

For MongoDB and Elastic, the rejection was a setback but not a stop. They proceeded without OSI approval, calculating that their user base and commercial imperatives outweighed the ideological certification.

The OSI’s authority, once largely tacit and unchallenged, was now openly contested. Its list of approved licenses was no longer the sole authoritative map of the permissible world.

The strategic calculus driving these moves was rooted in cold, institutional logic. For a publicly traded company like MongoDB, with over 1, 500 employees and a market valuation measured in billions, the threat from AWS’s DocumentDB service was concrete and material.

While MongoDB operated its own successful Atlas cloud service, the strategic fear was that AWS could commoditize the core database layer, capturing the vast, undifferentiated middle of the market and leaving MongoDB competing only in narrower enterprise niches.

The license change was a preemptive strike designed to radically increase AWS’s cost of adoption. Similarly, Elastic confronted an AWS service reportedly generating hundreds of millions in annual revenue—a revenue stream that did not flow back to Elastic.

Weaponizing the license was an attempt to force AWS into a binary choice: undertake the expensive and complex work of building a truly independent fork, or negotiate some form of partnership or revenue share with Elastic. It was a high-stakes game of legal and economic chicken.

This period provides a decisive answer to a persistent counter-argument: that open source’s ascendance was primarily a deterministic outcome of superior networked engineering efficiency and economic logic, with its institutional conflicts being mere superficial drama. The licensing wars of 2018-2020 demonstrate that the conflicts were central, not epiphenomenal.

The economic logic of modular production did indeed favor open-source models, but that very success created a critical vulnerability: the platform’s ability to capture and monetize the modular components without reciprocity. The institutional response—the deliberate, contested reshaping of the legal frameworks themselves—was a pivotal intervention that altered the trajectory. It was not an inevitable, smooth next step in evolution; it was a painful, explicit renegotiation of the rules of engagement. The efficiency of decentralized production collided with the concentrated power of centralized platforms, and the license was the chosen instrument for that collision. The aftermath was a permanently fragmented ecosystem. A new taxonomy solidified. “Open source” remained the designated term for software under OSI-approved licenses like GPL, Apache, or MIT. “Source available” became the accepted descriptor for software like MongoDB and Elasticsearch under the SSPL or Elastic License—visible and modifiable, but encumbered by explicit commercial restrictions.

Another model, the Business Source License (BSL) used by MariaDB, offered a time-based compromise: software began as source-available with usage restrictions, then automatically converted to a standard open-source license after a predetermined period (e.g., four years).

This was license-as-temporary-competitive-moat. The landscape was no longer unified by a shared philosophical north star but partitioned by strategic business considerations.

The concrete consequence of this fragmentation was a subtle but pervasive chilling effect on collaboration and institutional trust. When Elastic changed its license, every company, startup, and integrator built upon Elasticsearch was forced into a sudden legal reassessment. Was their core infrastructure now a potential liability? Would they be compelled to abandon their stack or switch providers? This uncertainty created new business risks and opportunities. It directly catalyzed community forks driven not by technical divergence but by licensing strategy. When Elastic changed its license, AWS swiftly supported the creation of OpenSearch, a fork of Elasticsearch and Kibana that remained under the Apache 2.0 license.

The developer community itself became a strategic asset to be mobilized in a corporate legal standoff, its cohesion sacrificed to the new logic of license-as-weapon. The actor Zhu Yilong first gained widespread recognition in television series such as Guardian in 2018, the very year this legal reordering began—a coincidental marker of how this specialized conflict unfolded within a wider world largely oblivious to its stakes.

Yet for the digital infrastructure upon which that world increasingly depended, these years marked a point of no return. The ideal of a universally accessible commons, governed by licenses designed solely to protect freedom, had been irrevocably complicated. The license was now undeniably a strategic weapon, deployed to defend economic territory in a war over who captured the value of open collaboration. The open-source ecosystem, now armed with intentionally restrictive licenses, entered a new phase of commercial and legal fragmentation. The battle lines were drawn not between open and closed, but between competing visions of what openness must sustain: pure liberty or productive capital.

This fragmentation did not destroy open source, but it permanently altered its social contract. Trust now required legal scrutiny; adoption came with strategic calculation. The cost of this defensive turn was a loss of innocence and unity; its benefit, for its advocates, was simply the chance to fight another day. As this legal dust settled, it was the individual developer—the coder writing commits in this newly weaponized environment—who would next feel the full weight of this industrial reorganization upon their daily work and professional life.