Kubernetes' Tim Hockin on a decade of dominance and the future of AI in open source
Going back to a time before autocomplete
Interview Tim Hockin has been working on Kubernetes since before it was announced. As the Cloud Native Computing Foundation (CNCF) takes a sudden lurch into the world of artificial intelligence, Hockin spoke to The Register about trends, licensing, and his love of Vi.
Next year will mark a decade since the initial release of Kubernetes, and Hockin, a near-20-year veteran of Google, has been heavily involved in the container orchestration technology. While he is proud of the project, he also has words of warning for the community that has sprung up around it.
"We've reached a level of success that I probably would never have dreamed of," he says. "Ten years ago I did not have the confidence that we would end up here … I think we are the de facto choice today, but we could not be tomorrow."
The IT world has changed very suddenly, and the CNCF, which counts Kubernetes among its projects, has leaped aboard the AI bandwagon with impressive agility.
"I think this whole AI/ML thing – this juggernaut," he says, "is changing the conversation a lot."
Hockin has been around for some time and witnessed the latest and greatest tech subsequently fade into obscurity as fast as it appeared. He noted the hype around blockchain and web3 with a wry smile. "Now it's like, 'We'd like our GPUs back now, thank you!'"
AI, however, is different. Although things like WebAssembly might get technologists excited, it has not had anywhere near the same cultural impact as AI. Hockin described talking to other parents at soccer games who could not give two hoots about Kubernetes, but were very much aware of AI. "I think the magnitude of the AI revolution is about the same order of magnitude as the internet," he says.
A glance around the Solutions Showcase at the recent North American KubeCon shindig shows no end of companies keen to brag about an AI tinge to their products. Hockin is a little cynical about such approaches, saying: "There's no magic, there's no pixie dust you can sprinkle on something and say 'now with AI!'
"There's a lot of people who are trying to raise money by doing that, for sure."
We note that Hockin's employer, Google, might also fit the bill.
Hockin compared today's AI tooling to the dial-up modems of yesterday, musing: "Could it be overhyped? Yes. I've seen a lot of things, but I think this one has real sticking power."
Another issue very much in the here and now is around licensing and the future of open source, particularly in light of a recent high-profile shift to a source-available license. "I watched the HashiCorp thing happen with the bowl of popcorn meme in my mind," he says.
"It doesn't affect me personally, but it was very interesting to see how many people it did affect – or rather, how many people who thought it affected them, and to see the passion of people that engage in the conversation."
However, Hockin takes a pragmatic approach. "You have to understand your business. You can't just write code, throw it out there, and hope to be successful ... It might make you lots of great friends ... but it's not the business.
"Companies like Hashi deserve to make a living. They write great software, and their fans are passionate advocates.
- Intel's Arun Gupta on open source pragmatism and fanatics
- The Cloud Native Computing Foundation leaps aboard the AI bandwagon
- CNCF's chief techie talks WebAssembly, AI and licenses
- If the Linux Foundation was a software company, it'd be the biggest in the world
"Terraform, for example, is a wonderful success story of organic growth. Now, I know Hashi put a ton of energy in making the organic growth happen, but people love it: It solves real problems and solves them well.
"And then the rug got yanked out.
"So I understand how people are frustrated with that. If it was at the beginning, it would have been different."
Even behemoths like the Kubernetes project have their issues, and Hockin notes issues with contributions and maintainership, where some put in vast amounts of effort and energy. In contrast, others tend to be "players who take a lot, and they put energy into just the places that help them exclusively."
As for where foundations fit into the picture – and companies such as HashiCorp have queried their role in recent times – Hockin says: "CNCF believes – and I agree with them – that it's not their job to make the winners or to bang heads together and force people to talk to each other. It's to provide fertile ground for ideas to grow."
And what of the addition of the OpenTofu project in the face of the Terraform licensing change? Hockin was diplomatic: "I don't know how I feel about OpenTofu. There's a lot of drama there that I don't really understand because it doesn't affect me personally, and it doesn't play in the space that I'm really familiar with. But I'm interested to see how it plays out."
What about the abrupt AI-ization of CNCF? "I think they've done a great job pivoting to it," but "in nine months, it'll be just another thing in the portfolio."
Finally, when questioned over the use of generative AI in day-to-day programming – Hockin was most concerned about knowing where the code generated had been derived from as well as the issue of trust – he revealed that even services that many today take for granted, such as autocomplete, do not always apply to him.
"I still use Vi."