Folks, there is a pressing issue with open source software. And no, it’s not the problem that some may think. People often criticize corporations for misrepresenting their code as open source, and there are times when they are right. Others lament the rise of venture-backed companies that dilute the true meaning of open source to serve their own corporate interests, and sometimes they are right as well.
However, the real problem is not solely with the companies. While businesses leveraging the open source label for commercial gains is nothing new, in recent years, free and open source software has veered off course, leaving developers and businesses with only one choice: permissive, Apache-style licensing. The original form of open source licensing, as its firm adherents claim, was not truly “open source” but rather copyleft, free software licensing like the GPL. This shift has compromised the ability to protect user freedom.
Blaming corporations is easy, but the fundamental issue is that in the rush to make open source software more appealing to corporations, we have sacrificed the power to safeguard user freedom.
No give, all take
There is no such thing as “open source AI,” despite our desire to believe otherwise. Even the Open Source Initiative (OSI), which defines the Open Source Definition, has been grappling with how to define “open source” in a world dominated by artificial intelligence. Stefano Maffulli, OSI’s executive director, has acknowledged that open source missed the evolution of software distribution and execution and is working towards addressing this by October 2024.
Confusion reigns in the AI realm, exemplified by Meta’s “open sourcing” of Llama, its large language model (LLM). While it is available for free research and commercial use, there are restrictions that deviate from standard open source definitions. This disparity highlights the need to rethink licenses to accommodate the complexities of AI models while upholding the principles of the open source community.
The focus on downstream developers’ rights has overshadowed the importance of protecting upstream developers’ freedom. This licensing dilemma, which I once dismissed as trivial, now appears more significant.
Both sides now
My perspective on open source licensing has evolved over the years. In 2005, I advocated for the GPL, praising its role in making open source commercially viable. By 2009, I shifted towards the Apache Software License, and by 2014, I acknowledged the changing landscape by proclaiming a “post-open source world.” Each stance had its merits and drawbacks, reflecting the complexity of these issues.
While I have leaned towards permissive licensing for community development, the governance structure of projects like Linux under the GPL showcases the value of software freedom in maintaining community cohesion. The tension between free software and open source in the past was beneficial for software development, offering project maintainers choices that are lacking today.
It’s crucial to consider the developer who values keeping their software free. The limited options in open source software licensing today are not conducive to long-term interests. We must revisit the importance of software freedom, not just developer/user freedom.
It’s time to shift the focus away from companies and prioritize the preservation of free software in the cloud and AI landscape. Developers deserve more choices beyond the current permissive licensing model. Open source and free software are essential, and it’s imperative to bring them back to the forefront.
Copyright © 2024 IDG Communications, Inc.
