The Fork in the Road: Understanding Community Dynamics

author-image

作者

 

In this episode of Open at Intel, host Katherine Druckman talks with Taylor Dolezal, the head of ecosystem at the Cloud Native Computing Foundation (CNCF). Taylor’s role involves connecting users, vendors, and community members within the CNCF, and he works closely with end users who utilize CNCF open source projects like Kubernetes, OpenTelemetry, and Argo to build platforms and workflows within their companies. Taylor has a background in open source development and has worked in various roles, including developer advocate and site reliability engineer. He has seen projects fork, such as Terraform forking into OpenTofu, and believes that forking can create healthy competition and benefit users by offering new functionalities and features. Taylor and Katherine discuss the challenges of prioritization, sustainability, and passing on knowledge within open source communities. Taylor encourages open dialogue and collaboration to address these challenges and make progress together. 

“Let's talk about what we're going to do before you jump right into the code, right? It's the measure twice, cut once kind of methodology. It's so helpful to do that. A lot of folks do lose sight of that because I get it, right? There's an enthusiasm in open source.”

– Taylor Dolezal 

 

Katherine Druckman: Hey, Taylor Dolezal from the CNCF, that is the Cloud Native Computing Foundation for anybody who might not know, but I think probably most people listening to this do know. Thanks for joining me. I look forward to hearing your stories. 

Taylor Dolezal: I can't wait to tell them. I'm excited to see where we're going to go and what we're going to talk about and thank you so much for having me. 

Katherine Druckman: Yeah, this is going to be great. I would love for you to start by telling us what you do at the CNCF. 

Taylor Dolezal: That's something that I try to figure out every single day. 

Katherine Druckman: I can identify. 

Taylor Dolezal: I'm the head of ecosystem at the CNCF, and what that constitutes is working closely with our end user community. Now, end user, what the heck is that? Those are folks who are taking CNCF open source projects like Kubernetes, OpenTelemetry, and Argo, and then building up platforms and workflows within their companies. It's not folks that are reselling those products, it's not consultancies; they're not reselling or repackaging up the experience. It's end users like Mercedes-Benz, Intuit, or Apple taking these projects and then building out that platform. I also talk to vendors and cloud service providers, and then we just stitch everybody together and help out with those conversations. We'll see, "Oh, you're having the same conversation in two different places, let's bring them together." That's part of the job as well. And then building out programs or just being a human router and connecting people together. I get to talk to cool people all day, and I love my job. It's so much fun. 

Katherine Druckman: Oh, awesome. I appreciate you talking to me. I feel like I'm in good company. This is a privilege. Tell me a little bit about your open source origin story. 

Taylor’s Open Source Origin Story

Taylor Dolezal: Yeah, I'm very happy. Honestly, when I was growing up, I was lucky in that I got a sneak peek very early on into what I wanted to be when I grew up. I remember it distinctly, being in math class and holding a TI-83 Plus calculator. We were in the final five minutes of this class, and I had someone come up to me and say, "Hey, Taylor, did you know you could play games on that thing?" 

I was like, "Wait, what?" And so we were playing Nibbles, the snake game, and I was like, "This is so fun." 

Then this person did a "and one more thing" to me, and they said, "Oh, these are…" I think it was the number eight or asterisks or something like that that made up the segments of the snake, that as you got the little niblet thing, you would keep growing and try not to bend upon yourself, and so they changed it to another character or question mark, I can't remember what it was. 

But once I saw them go into the code and change that, I was like, "This is cool." 

Katherine Druckman: The power. 

Taylor Dolezal: It was all over. It was all over after that. 

Katherine Druckman: That's awesome. 

Taylor Dolezal: For the rest of my high school career and beyond I was working in the basic programming language, building out things to help summate different math lessons that we had. I built equation solvers. It was just so much fun. I had no idea about licenses and dependencies and all these things at that point in time, but that's what sparked me and got me really interested in working in an open source way because I would be trading these programs and these sets of instructions with classmates and other people. I would ask folks about what's helpful, what's not. It was something I liked doing, so I liked getting feedback and sharing. You can see I was being forged in this open source crucible without even knowing it way back when I was 13 and onward. That's really where it all started. 

And then as I grew up, my parents were divorced. We weren't an affluent family, and it wasn't an easy thing. I was working in VB 6 and things like that. People who aren't familiar with these terms should feel free to look them up. Really, all I'm doing here is dating myself, but it was around the time when Microsoft came out with .NET and Visual Basic and C# and all of these languages, but you had to pay pretty steep fees to be able to get them. I'm so happy that we don't have that today, and a coding interface or IDE is lot more accessible. It's just not as financially limiting. 

I think that was the first time I hit an impediment that I wasn't used to. I had a calculator I could code. When I wanted to move onto the desktop, and that was around the time that iPhones and mobile development were coming up. I saw these cost barriers and I was too young to be able to afford all these things. My family didn't have the means to do this, and so I was like, "That's it. I'm done. I'm 16 and my life is over. I can't afford an MSDN subscription. What am I to do?" That's when I came upon PHP and all of these other communities. I was like, "There's a way." 

Katherine Druckman: That's an excellent open source gateway. 

Taylor Dolezal: It was so cool. I agree. It was so cool, going to some of those forums and just saying, "I don't know how to do this." And being able to pose a problem like that and get clarity on it from someone else that I may never meet. Some folks I did meet and some I never will, and I loved the intention behind the coding and people seeing one another from a human perspective. "You want to do cool things? Here, let me help you." Here's a breadcrumb on your journey kind of thing. I love that, I love paying it forward. 

And then, yeah, ever since getting into Ruby on Rails, making my first contribution, getting into Go. Working at a hospital, upleveling them to the cloud versus pen and paper. There's so many fun little tidbits from my story that I love to talk about and tell. It's the one thing I love doing with people—if you're on the fence about open source or wondering, should I? Here's a friendly nudge. Yes, you should. Just give it a try, right? 

Katherine Druckman: Yeah. 

Taylor Dolezal: Take a look at it. It's fun. 

Katherine Druckman: First, I have to say that I love that a TI calculator is part of your story. I had a TI-85, and at the time, that was maybe the coolest device I had ever held in my hand. I mean, it was amazing, it really was. I previously had dabbled with other TI devices, but if I share those, I would definitely be dating myself, and I think I'll leave a little mystery. But yeah, that's such a good story, and then you mentioned PHP, where I had a similar experience. My entry into the open source world was through web applications. I think that there's a common thread. I find that with so many people, especially in the cloud native community, a lot of people got their start in building simple web applications. And of course, that evolved into the complicated world that we live in today. So yeah, I think it's fun to hear how people ended up where they are now. 

Taylor Dolezal: You think back to some of these early communities too. I think of myself as an elder millennial so the behavioral psychology of all of these little micro or macro communities that popped up at these very specific times of our lives fascinates me too. I remember Myspace and styling your page and CSS. 

Katherine Druckman: Oh my gosh, yes. 

Taylor Dolezal: So fun. 

Katherine Druckman: Remember when Cascading Style Sheets were new and even a little controversial? What do you mean I don't use a table? I don't understand. It's wild. 

Taylor Dolezal: And it's not really cascading, what is this important tag doing here? 

Katherine Druckman: Yes, what does that mean? I love it. This is very cool, and from those beginnings, here you are at one of the most significant organizations in the open source world right now. These things ebb and flow, but I would say right now that CNCF is pretty critical stuff. 

Taylor Dolezal: It's a little thing called cloud native. That could be the memoir, who knows. 

Katherine Druckman: A little thing called cloud native. 

Taylor Dolezal: It's wild to see the impact that cloud native has had, and I think that I could take the modest approach and say, "Oh no, it's all you." It really is everyone working together. It really is. It's not just the foundation, it's not just the maintainers. It's this amalgamation of people with this amazing intention to be able to bring this knowledge to other people, and a lot of courage on the part of a lot of people. 

I was privileged to go to the Kubernetes 10th birthday party. It was held at the Google facility in SF, and it got to the point where there were people crying on stage because they remembered how difficult it was, and they were so moved by what the community has done and how it has come together to solve all of these gaps. I love that people are pragmatic and are thinking about how we don't all do the same thing 18 different ways across each of our own organizations. I love that pragmatism and saying, "There's got to be a better way. No one wants to do this thing. Hey, do you like renewing TLS certs? Probably not. Let's figure out how to automate that.” So cool, I love that. 

Katherine Druckman: Automate the boring stuff. Yeah, I like it too. I do remember a world before Kubernetes existed. I remember the conversations a long time ago that were like, "Ooh, containers. Containers are the future. What's a container?" It's wild. 

Taylor Dolezal: We figured out how to make things that we never even thought about, and how to make them more complex too. It is a double-edged sword. 

Katherine Druckman: It's a double-edged sword, yeah, for sure. We heard your origins, and now we're here, but a lot of things have happened in the meantime. To oversimplify, there's something that we've talked about previously before this recording and that I really wanted to pick your brain about. You've been involved in a lot of different types of work. You were a developer advocate for quite a while, I believe. You did some site reliability, is that correct? 

Taylor Dolezal: Yeah. 

Project Forks and Community Dynamics

Katherine Druckman: So, you've seen things and you've experienced, let's say, the more difficult parts of community-based development. One of the things that I wanted to talk about with you is when projects fork. Now, I have minimal experience. I've seen it as an outsider, especially in projects that I've used. I've seen CMSs fork and create new ones. Usually, that has to do with the vision, right? The vision for the project, what one group of people may see as more beneficial to a certain group of users, and then something will fork, and that's great. 

Or distributions, it's not a fork, but it's applying a different vision for something that is underlying the distribution. That could be Linux, that could be something else. Drupal has distributions, for example. But there are some more controversial ones, such as a license change, and there may be some other things that I haven't even thought of. I would love to hear your experiences about when projects fork, why they fork, and should they fork? 

Taylor Dolezal: It's something that I love, and it delves a little bit into the psychology front as well, right? Because at the core of it, which I feel is a cliché, but it is incredibly true, it's that these communities are people. And with people comes perspectives and intentions and just hopefully respectful disagreements about the things that we do when it gets intense. Sometimes you have forcing functions where projects have to fork because, like you said, it's a licensing thing, or there's a subset of the community or a large part of the community that just really wants to do something one way. There's a disagreement, and that's the only thing that makes sense. 

Within my career, I've been lucky to jump around to different verticals, and so as chance would have it, I worked at HashiCorp before coming to CNCF. I have so many thoughts and perspectives, and I'm past my NDA, so I can talk about it. From the perspective of my HashiCorp and ex-HashiCorp experience, my CNCF experience, or just Taylor, when it comes down to it, the story of HashiCorp is really interesting. I joined when there were sub 1,000 people there, and I was employee 996. I didn't really hold too much clout there. 

Katherine Druckman: Cool. 

Project Forks and Community Dynamics 

Taylor Dolezal: What I did there was work on the DevRel team, so I got to meet with the community and have these really interesting chats and then work across groups, and I know I'm getting ahead of myself, but I promise I'll jump back into forks. I would work across marketing, product, and engineering, and directly with the community and customer success teams. It was an incredibly multifaceted role, and I developed a lot of empathy working with each of these people and getting insights and tidbits of what they did day in and day out to flesh out the entire picture. That's what made this a lot more impactful for me, seeing Terraform fork into OpenTofu, and we've seen others like Valkey and Redis, and OpenBao and Vault, and it won't be the last one. 

Historically, IO.js and Node.js was a fun one way back in the day. I think that things like that happen because there are just a lot of factors. HashiCorp ended up growing like crazy. I think that we were around 2,500 or more employees by the time I ended up leaving. And this was all happening during the intensity of COVID, with everyone working remote. You can't see the company grow at that kind of rate or scale when you're at home, and it's hard to visualize when you're jumping on Zoom calls. The company was scaling up, getting ready for an IPO. We didn't know that at the time. We thought that was where it was going, but we didn't know for sure. You're seeing all these people come in and more process getting added. The company is fundamentally changing, which does slow things down, and it's becoming a different company. So that was one huge effect of a lot of this. 

Meanwhile, Terraform was open source under an MPL license, so people could take it and copy it, and that's where I think that HashiCorp did the right thing in terms of what worked for them and with switching to the business license. They were slowing down trying to figure out processes. Not so much adding new features, bolstering the SaaS platform that they were working on. You can't do all of these at once, and you really have to be strategic and pick and choose on that front. Unfortunately, a lot of that fell to product and engineering, and they just couldn't do everything at once, so it made sense for them to do a defensive rather than an offensive move, which would've been matching features, pricing, a combination of other things. 

That's why I think they ended up, I don't know for sure but based on everything that they published and people I've talked to, the focus seemed to be, “how do we make this defensible and not just give everything away for free?” They still need to make money and they absolutely should. Closed source isn't a bad word, inner source isn't a bad word. It's just how you use these things, and you can be effective with all of them, but it takes a lot of dedication, time, and strategy. You're not always right. People are human, this happens, but that's where I really think that sharing these stories is so important because we can learn from one another. It's not a zero-sum game. I refuse to believe that. There are so many other ways to tackle these things and for everyone to truly be successful. It was very interesting leaving HashiCorp, coming into CNCF, and seeing the OpenTofu folks come up into the Linux Foundation. 

I was eating a lot of popcorn at that time. I was just like, "Ooh, what's going to happen next?" Lots of good discussions. 

Katherine Druckman: Action film. 

Taylor Dolezal: It was fun. Hopefully it'll be on Netflix at some point in time. 

Katherine Druckman: Why not? 

Taylor Dolezal: I want more techno thrillers like that, that'd be cool. 

Katherine Druckman: Right 

Taylor Dolezal: It really hurt, too, to see people standing on either side of this. I truly was right in the middle and would talk with people on the HashiCorp side and on the OpenTofu side, and just really try to pull them together. I think the best-case scenario with a fork is that it creates competition. You see the intent behind the decisions that led to the fork, and then at some point in time, hopefully like with IO.js and Node.js, both the fork and the original project see the light, and then they can merge back together. 

OpenTofu is still open to that. I'm not 100% sure about Terraform, but we'll see what happens with the IBM acquisition. Hopefully they come back together. I would love to see that happen. Users win because they get all these cool new functionalities, features, and workflows. It just creates less work; the difficulty for the user is gone. At the end of the day, that's the most important thing. Don't create friction for your ecosystem and for your users, and you're going to have an easier time for sure. 

Navigating the AI Frontier

Katherine Druckman: Something that I feel is peripherally related to this conversation around forks and communities, and the reason I think we both have an open bias is because we've seen the benefits of solving problems together, which is the spirit of open source. You're all collaborating on the common issues and problems, which lifts all boats, and relieves maintenance burden for everybody. There are a lot of great reasons to participate in open source communities, but let’s take all of the things that you and I have both seen over the years and then apply that experience to looking at the emerging world of, for example, AI. I feel like everybody's got to talk about AI right now, right? Because we're all talking about it. 

And so now it's the Wild West again. We're pioneers yet again in a new world, and those of us from the open source community are trying to redefine things and figure out how these things will fit together, and how we can ensure some openness and interoperability and solve the big problems together. But in, let's say, times of dynamic change, these things can lead to competing needs and interests. Now, of course, again, we come from a place of open bias. 

I've got to throw this out. Intel, along with the Linux Foundation, is involved in a new thing called OPEA, which is the Open Platform for Enterprise AI. These are people trying to work toward preserving an open environment. But I do foresee a rockier path because every time you break new ground, it becomes complicated. Anyway, this is a very roundabout way of asking about your thoughts on what the future looks like for defining these new communities around AI, based on what you have seen in open source communities.  

Taylor Dolezal: It's just like you said. I live in Los Angeles, and I like making the comparison between cloud native and AI to how a lot of the neighborhoods are set up here. They're sprawling. Everything is interconnected, but it is so fundamentally different if you just go a couple of streets down. Very different kind of vibe—people, hobbies, and things like that. It's very much the same thing within all these little micro-ecosystems we run into, and with something new like AI coming up, that's a double-edged sword there too, right? Everything is a trade-off. There's absolutely nothing in the world, unfortunately, that's going to be 100% good or 100% bad. It's always a mix.  

But with AI, we're still in that framing period where people are trying to figure out what is the killer feature? What is the absolute reason I can't avoid this? Or, here's the reason I have to adopt this. We haven't really seen the answer. We've seen features that will help summarize stuff and a lot of products around GenAI. How you adopt it into your organization is still mostly being figured out. The good news is that all the infrastructure supports it. Kubernetes has all these features, and it's like, it's ready yesterday. It's a platform to build platforms. 

Katherine Druckman: That's always a great answer. 

Taylor Dolezal: It's so cool, but the reality is that we don't have specifications. 

Katherine Druckman: There you go, yeah. 

The Challenges of AI Standardization and Vendor Neutrality 

Taylor Dolezal: We don't know what the de facto organization is. We saw OpenAI come out with a model spec. Huge applause for that. I'm biased working at the CNCF and the Linux Foundation, but I would love to see a foundation own that specification, so that it gets developed in a vendor-neutral way to ensure that people's voices are heard, right? There's always that risk of placating the original organization if it's still held internally versus in more of a vendor-neutral space. And for folks who have not seen the XKCD comic, please check that out. It's a really funny comic. It's math, science, computer stuff, and other things. 

Katherine Druckman: So good, yeah. 

Taylor Dolezal: And there's one comic where somebody is saying like, "Hey, there's 14 competing standards about this thing. We should make one that unites all of them." And then the next frame, it says, "There are 15 competing standards." 

Katherine Druckman: That is perfect. 

Taylor Dolezal: That's what we're seeing in AI today. We just don't know where to go, but what is helpful to really frame things is, what part of AI are you talking about? It's really split into two main focuses. One is training, which is model creation, and that's using PyTorch and JAX and other frameworks for data science or data classification. That's not what the world is obsessed about right now when it comes to AI. They're more focused on the productization, which is inference. GenAI, ChatGPT. You've built the model, and you've worked with the data. Now, how are we applying it? 

Katherine Druckman: Now, yeah. 

Taylor Dolezal: And then GPUs, all that fun stuff, and can we make a CPU that can process things in a similar way? How does all this fit together? What do we do? That's where the churn and the interest are right now. I'm happy to see that it's benefiting a lot of people who have spent a lot of time in ML and data science. 

At CNCF, we're focused on the infrastructure side of things. We're starting to get a little bit into the app development and deployment side as well, but when it comes to training models and things like that, there is the PyTorch Foundation, under the Linux Foundation. There's the LF AI & DATA Foundation headed by Ibrahim Haddad. 

Katherine Druckman: So, a couple of things. One, there's always a good XKCD to use as reference material, and I love that you brought that up. And then you said something that is music to my ears and that's vendor neutrality, and I think that's so important in this conversation. I feel like , we're at one of the many inflection points that we've seen in the tech world over the last, oh, I don't know, decades. 

Here, things are moving so fast and people are focused on innovating quickly and doing things as fast as they can. And what we end up doing, because we're humans, is we innovate very quickly in silos, and we can't help it. Even though, back to the open source ethos, which is, let's do it together and collaborate on the hard stuff, we come up with 15 standard specifications. But yeah, if we could come together a little bit and solve these problems, I think we'd all be better off, right? Take a little bit of a pause instead of spinning our wheels in our own little bubbles. 

Balancing Priorities in Open Source

Taylor Dolezal: Yeah. Let's talk about what we're going to do before jumping right into the code, right? It's the measure twice, cut once kind of methodology. 

Katherine Druckman: Yes. Love it. 

Taylor Dolezal: It's so helpful to do that. A lot of folks do lose sight of that. I get it, right? There's an enthusiasm in open source. At the end of the day, I'm very, very happy within the open source space and very bullish on it, but there are things to watch out for too. I'm more than happy to share, and that's with that positive intention. 

It's very easy to make this laundry list of features and all this cool stuff that you're going to add, and it can be difficult to figure out prioritization on that front and what should be in the backlog. How do we ensure a good developer experience for people—power users as well as people coming into the project or using it for the first time? There are all these things you have to balance. The reality is that good prioritization is really hard to do, especially with a lot more people in the mix across a larger ecosystem that includes many different companies with many different things that they consider important for what they're doing day to day. 

They might be in the manufacturing space, they might be in finance, they might be selling ads. All these different perspectives make it very difficult to prioritize in some contexts as well as to identify people to do the work, right? Life happens. When you thought you had free time, that might evaporate. You might have a new addition to your family, something unfortunate might happen. There might be a job change. Life isn't always predictable. As much as we want our infrastructure to be declarative, life isn't, so navigating that is yet another thing that we have to think about, and it makes it difficult, but we do need maintainers. We need people who want to show up and stick around. 

You do have benevolent dictators for life, like Linux Torvalds with Linux. That's good. It's helpful to have Linux as a very opinionated thing. It doesn't restrict people from forking and creating other versions. For Linux, for Kubernetes, it works well in some groups to have someone with all the knowledge, but we do need to think about sustainability over the long term, how we educate folks, and how we pass the torch off to the next generation. How do we do all this? How do we capture all these fun stories that happened? Like, "Oh, that line of code? Oh, sit down. This is a long story." Comments don't do it. 

Katherine Druckman: That's really funny. Novels and novels for a line of code. We could go so many places with this, but I don't want to keep you the whole day. I feel like we're going to need some follow-up conversations here because the conversation around sustainability and the open source community is so critically important. There are common pain points for all organizations. Life, period, right? Open-source technology is critical to just about every single person on the planet, right? We’ve got to take care of our software environment just as we do our planet. And we’ve got to take care of the people who are creating it. The more we can address these issues together collaboratively, the better it is for all of us. So yeah, these are definitely conversations I would love to continue. But for now, I really appreciate the one we've just had. 

Taylor Dolezal: Yeah, thank you so much, Katherine. I love talking about this. I'm a person who tries to be as accessible as possible. I've burnt out several times, so I've figured out how to safeguard myself against most of that. But I still encourage everybody, please reach out on LinkedIn, on any platform where you find me @onlydole. Please, please, please reach out. Happy to share my email with you. I love having these conversations. I love getting emails like, "This is great. This is helpful." I love hearing that. I love hearing about, "Taylor, this is really difficult. How do we fix this?" Right? 

I love both of those because it's all culminating into progress, how we refine things and make them better. If everything's great, great. If something needs to be fixed, let's do it. And trying to show people how they can have a hand in making that happen makes me happy. I love to see that spark of somebody finding value in the ecosystem and the community and seeing that they're going to continue carrying that experience with them. Yeah, let's do this together. It's not always easy. So let's work together, let's do it. Please @ me. I love having these conversations. 

Katherine Druckman: Fabulous. Well, that's great. Yeah, I hope that you get some good @s out of this conversation and the people listening to it. I really appreciate this. I hope we'll do it again soon, so everyone has something to look forward to. 

Taylor Dolezal: Thank you, thank you. Have a good one, everybody. 

About the Guest

Taylor Dolezal, Head of Ecosystem at the Cloud Native Computing Foundation (CNCF) 

Taylor Dolezal navigates the cloud native universe with a knack for puns and a keen eye for psychology. Living in the heart of LA, he blends tech innovation with mental insights, one punny cloud at a time. Avid reader, thinker, and cloud whisperer. 

About the Author 

Katherine Druckman, Open Source Security Evangelist, Intel  

Katherine Druckman, an Intel open source security evangelist, hosts the podcasts Open at Intel, Reality 2.0, and FLOSS Weekly. A security and privacy advocate, software engineer, and former digital director of Linux Journal, she's a long-time champion of open source and open standards. She is a software engineer and content creator with over a decade of experience in engineering, content strategy, product management, user experience, and technology evangelism.