The Crippled Llama: What is going on with enterprises and open source?
In July of 2023, two spats about enterprises controlling open source projects have arisen: Meta Llama 2 and RedHat downstream source code. What is going on here?
There is no ending, there is no beginning, but we have to start somewhere to tell a story
The stories don’t start in July 2023. One starting point is the choice by Google to move the control of Kubernetes to The Cloud Native Computing Foundation (CNCF).
I can safely say that Kubernetes was an inflection point of the interaction between enterprises and open source. I feel pretty confident claiming that after K8s became a foundation of new app deployment, enterprise use of open source increased — as did enterprise contributions to open source projects. (Does anyone agree or disagree with this? Let’s call it Point 1 and feel free to let us know.)
Google took a risk and boldly released control of Kubernetes to the CNCF. Did it pay off? There is probably a business case that it did not. Google could have monetized Kubernetes. Not necessarily just in licensing but as a unique service in GCP or some other weird tech model. They could also (as Meta has done with the Llama 2 license) just prevent their rivals from monetizing their intellectual creation.
It is open source if I say it is open source!
The news and kerfuffle about Llama is following on the news that Red Hat, directed by its parent company IBM, clamped down on access to its source code. They knowingly stepped on competing business models. In the process, they have threatened many business server architectures that rely on CentOS or other distros built from RHEL like Rocky and AlmaLInux.
Business minds at IBM seem to make this argument with Red Hat; money is being left on the table or just outright given to direct competitors. While raising the ire of some in the open source community, the strategy seems to have worked for the most recent quarter.
So Meta has released a large language model (LLM) AI that they label as “open source.” But, their license contains limitations in use that would not be acceptable to any of the extant open source licenses. It was also not created, curated, or maintained by a community. So, folks like Stefano Maffulli, quoted in The Register, note it is not really open source. I am putting an extensive excerpt here for your convenience and to support my point, but do read the whole article, Meta can call Llama 2 open source as much as it likes, but that doesn’t mean it is.
Stefano Maffulli, the OSI’s executive director, explained: “While I’m happy that Meta is pushing the bar of available access to powerful AI systems, I’m concerned about the confusion by some who celebrate Llama 2 as being open source: if it were, it wouldn’t have any restrictions on commercial use (points 5 and 6 of the Open Source Definition). As it is, the terms Meta has applied only allow some commercial use. The keyword is some.”
Maffulli then dove in deeper. “Open source means that developers and users are able to decide for themselves how and where to use the technology without the need to engage with another party; they have sovereignty over the technology they use. When read superficially, Llama’s license says, ‘You can’t use this if you’re Amazon, Google, Microsoft, Bytedance, Alibaba, or your startup grows as big.’ It may sound like a reasonable clause, but it also implicitly says, ‘You need to ask us for permission to create a tool that may solve world hunger’ or anything big like that.”
Open Source and the Public Company
As a fan of open source, that is disappointing and perhaps it seems like a bad idea. But IBM and Meta are publicly held companies. Their only way of keeping score is profitability. There were likely very involved conversations about how to position Red Hat in the IBM business and how to set up the Llama licensing so that people in the organizations did not get fired by the board for giving money away.
A cloudy view into the future
The open source perspective is compelling — a truly open source Llama could become the LLM of choice for all AI development. It could outlast and perhaps choke OpenAI and other cathedral-modeled platforms. But would Meta benefit from that? I think it would be hard for the employees to make a case for that to their bosses.
Llama does have distinct value as a moderately open platform, and I am glad that there is a fish swimming in the AI pool whose genetic code we have access to. Is it open source? not truly. That makes business sense, but cripples the Llama. As a result, it will not benefit from the enthusiasm and innovation that marks a truly open source project.