I've had this conversation a few times with people today so I may as well have it publicly here now too.<p>I feel especially privileged to have lead the Heroku Add-ons/Ecosystem team at a pretty pivotal time in devtools history. There was a sudden emergence of people/companies inventing entirely new things (e.g., databases, logging systems, telemetry, etc.) and so much of it was OSS. The overwhelming majority of these companies took the approach of "we'll make this thing free and successful, and we'll build a business off the back of enterprise support contracts and maybe some feature discrimination in a private 'enterprise' version" (e.g., clustering/HA support). I think in part it was because back then the only open source success story anybody had as a reference was RedHat, and cloud adoption wasn't as ubiquitous as it is today. Certainly not in the enterprise segment. So in the vacuum that was left emerged a whole industry of smaller startups that would provide said technology as a managed service. Go check out the Heroku Add-ons Marketplace circa 2012-2015 to see what I mean. Belatedly the creators of these technologies realised the enterprise support contract business was a terrible business to be in, and realised managed services was where they should have been all along. Absolutely none of these companies had any problem muscling in on the ecosystem of managed providers that had contributed to their success in a meaningful way. Some of the startups got acquisition offers on pretty lowball terms, others were essentially forced to accept partner terms that were so onerous it was doubtful they could ever turn what they'd built into a successful high growth business now. Many saw the writing on the wall and found an exit at a larger cloud/platform company that could roll them into their broader product portfolio.<p>Fast-forward a few years and AWS starts offering some of these technologies as a managed offering (disclaimer: I later worked at AWS for a couple of years). Suddenly these same companies don't like having similar market pressure exerted on them, and so begins the slow trend of license changing away from APL/MIT/whatever towards something that is trying to neutralise a legitimate competitor. Rules for thee, not rules for me.<p>My time at AWS gave me some new perspective on this whole sorry saga though, some things I'd observed but couldn't quite articulate why it didn't feel right. AWS taught me that at a certain level of scale almost everything ultimately becomes a logistics challenge. Trying to ensure that the infrastructure that's supporting tens of thousands of customers globally is constantly running, highly available, able to support the continued growth, etc.? It's as much a problem of capacity planning and co-ordination as one of software. And the more successful you get the less the problem becomes the specific nuances of running a given OSS product and the more it skews towards just knowing how to coordinate millions of anything.<p>What this surfaced for me is that in the vast majority of cases that I was personally familiar with, the companies in question barely used their own products. I don't mean in way that suggests they didn't believe in their value. It's just that their day-to-day needs of building said product very rarely intersected with the need to be the most sophisticated user of said product. They had very limited experience at operating it at scale, they all had customers (or managed service partners) who had orders of magnitude more experience about the realities of operating it. And high on their own hubris they'd decided that because they'd invented the technology they were now suddenly expected to be the world leaders at running it. They weren't. And they were never going to be, because the moment you hit that inflection point of success AWS/Microsoft/Google/so many others are better at running software than you are... and a license isn't going to change that reality. The "we'll run this for you" is just a bad business to be in.<p>A better business is "we'll provide you a UX and workflow and features _on top_ of that thing that makes it even better". There's a whole industry of companies who exist solely to make your AWS bill comprehensible, because AWS are organisationally incapable of providing good UX for most things. In it's most reductive and cynical take Heroku is "just" a UX on top of the core AWS commodities, one that has been largely unchanged for 5-10 years depending on who you want to ask (the slow decline there is a whole separate topic).<p>Which is why I was excited to take on a product leadership role at HashiCorp to help launch Terraform Cloud a few years ago (I left last year). Here you had an OSS product with a big community, and a set of features and capabilities that extended that to try and make it even better. Especially in situations where you're having to work with other people or across multiple teams. The fact that Spacelift, Scala, Harness, Pulumi, Terrateam, etc. existed didn't bother me much. If they copied what we were doing it was often good validation, if we lost a customer to them it was a good data point for things we were lacking or needed to fix, in some cases they just had wildly different takes on fundamental things which were a great reason for some self-reflection and to question why our conviction on a different way was so strong... were we right? How did we know?<p>OSS is good for so many reasons, but as a product person one of the things I loved most was the way it could help shape what the product could be in the future. Because of the ecosystem that erupts around it. You've already got such a huge advantage as the steward of the project, the most recognised brand in the ecosystem you created, <i>the</i> brand recognition in an enterprise conversation, and so with all of that head start I felt like we should just win on our merits. And if you can't win given all of that advantage then maybe you don't deserve to.