 Next, we actually have another great speaker who has a lot of experience in the open source world. From Microsoft, we have Sarah Novotny, who's joining us. She heads up their open source strategy and ecosystem. And so welcome, Sarah. Thanks so much. It's great to be here. Now, Sarah, as I mentioned, you have a lot of open source experience. You've been in the open source world for quite a while and a lot of different contexts, small companies, big companies really guiding a lot of people through engaging in open source communities. And I'd really like to start there. Just what are some of the key patterns for success that you've seen when companies engage in open source communities? Oh, this is such a fun topic. It's something I just think about and talk about a lot. And honestly, what the first stop I made when I'm talking to a company or even a product group or some group within the industry is to ask why they want to open source something because that will guide how they need to engage in the ecosystem or in the open source communities in order to be successful. So maybe your goal is to recruit from open source. So you want to open source some tools so that you can potentially work with different communities or different groups that may give you an insight into whether or not they're right for your company to hire. So that could be one thing. We all know the use case of engineering economics, which is nobody writes a math library anymore. You use an open source library for that. It's pretty straightforward. But if you want to open source something, are you trying to change the industry like Kubernetes and OpenStack did, for example? Or are you trying to just potentially find ways and adjacent areas where you can draw in a partner ecosystem? You can build those all out. So basically, I always, always start with why do you think you want to open source it? What is the goal that you have for this project? And then I walk through and talk with them about how successes look different for the different types of goals. And open source success can mean moving an industry. It can also mean just being a commonly used library. And since success looks so different, a company really needs to know what they're looking for in order to have that conversation. I think that is a really great place to start. I think just kind of referring back to licensing. Sometimes one of the things that I think companies have found difficult is they sort of started down an open source path, which means you're sort of giving away your code to the world. And then maybe didn't realize why they did that or the impacts that it would have on their business. So I think you're absolutely right. You really have to start with that why. Understanding what the software is that they might be open sourcing is another piece. We talk about this, I think Stephen Wally came up with this sort of framework. We talk about it in core context and complement. And a core is the whole of the project or the product area that you're giving away. Context is something like Spinnaker where Netflix built it and needed it and used it, but they didn't actually have it as their core business. So they could use it as a recruiting tool. They could use it as a way to be engaged in open source. But all it did was give them information about what was going on around their core business. And then complement is something that is open source. You're never really going to build an ecosystem around it, but it needs to exist in order for your product to be picked up, engaged with, and your customers have a good experience. So that might be an SDK that's open sourced. Yeah, that's great. That's a really useful framework. What are areas where you've seen kind of people stub their toes or get off to a rocky start that are maybe the anti-patterns to watch out for? Yeah, control for me is the anti-pattern to watch for. Anytime I'm talking with a company or an organization and their focus is immediately on how do I control this project. I want to not lose control. I work them back and try to say, well, if you're really worried about control, why are we open sourced? Let's look at it from a more holistic standpoint and go through what are you trying to get out of this? I think that's another great point. Control, especially in open source, is an illusion. And in a lot of ways it's counterproductive to, oh, yes. It's all about influence. Exactly. I was actually going to say that. That was one of the things that was really interesting for me as I moved from kind of a more of a traditional commercial world into really much more of an open source world. You really do go from control and direction to having to have relationships and influence. And it can be a shift, but it's also much more powerful over the long term and really also just kind of more fun in terms of the way that we get to work. We first started working together over a decade ago when we were launching OpenStack and a lot of people in open source communities carry that through. So my final question is, as you think about kind of the phases of open source, we've gone from it being a new concept 20 years ago to it being something that is pretty widely accepted within business. But we've also seen changes in the way that people produce code, host code, license code. What do you think are some of the big trends to look out for over the next decade of open source that people should be thinking about? I think there are two that are really interesting ahead of us. One of them is provenance of data and understanding how your data has been transformed. So I have a collection of data. It might not be all that clean. I need to do things to it to transform it. I need to know what that is by the time I give out another data set. So open data sets I think are going to be a very important thing. And getting a cross industry, cross worldview of that data in a meaningful way while understanding where we as humans may have introduced risk through our own biases, through our own cleaning things up, that kind of work. We need to understand that provenance and we need to understand how those data sets evolve. And that of course ties into the other half of it, which is all of AI. Because we need to understand how models in AI involve and how models are trained on new data. And we need to understand that and then be able to go back and legitimately audit the biases or the concerns that we may have about data that it was used to train a model. So we need to have, there are open source, or there are startups. I don't think there are open source, but there are startups that are looking at this kind of space now. So it's almost like a get hub for model data, data specifically, and then models where you can see the transformations that are made in the data and then the model changes that are going forward. So I think that's a big spot that's going to have new focus and attention in the coming years. Okay, well, that will be something to keep our eyes on. Thank you very much for joining us. It was great to learn a little bit from your wisdom. Happy to join you all. It's always fun to see. And yes, it has been 10 years. Holy cow. All right. Well, thank you, Sarah.