For me, it was the second application I ever released, when I was a student at university and still didn't really know how to program properly. The application

For me, it was the second application I ever released, when I was a student at university and still didn't really know how to program properly. The application was Dash Board[1] for Newton OS, and it only ran on the final generation of Newton hardware (created by Apple, but spun out as a separate company in its final days, before being killed by Steve Jobs shortly after his return). It "only" sold a few thousand copies. (But it was during the warez heyday, and I am pretty sure there were also tens of thousands of bootleg copies being used, thanks to the registration code generator by "DocNZ" that was widely shared on Hotline back then.) But that was really pretty great, since the final MP2000/2100 generation of hardware it required was thought to have only sold about 200,000 devices in total. I have since had a fairly normal software engineer career, and have worked on apps that shipped far more copies, and today I work on customer facing web applications and API SDKs that have more users, and arguably do stuff that is more "important" (e.g. help companies manage large fleets of machines/robots/IoT stuff) than what Dash Board did — which was basically just improve the user interface of the Newton. But it's 100% clear to me that the magnitude of user impact of Dash Board was much higher than any other thing I've built. People really loved it — I know because hundreds of them actually wrote to us to let us know. (LOL I mean wrote to me "me" — old habits of pretending the company wasn't just one student in his tiny apartment die hard). Of course, I made more money later, and worked on things that touched a much larger number of people's lives. But "impact" has both X and Y axes. It was the depth of the users' fondness for Dash Board that makes it eclipse everything since. I don't think there are that many chances to just go for "user delight" as the number one metric. For me, developer satisfaction is a function of that user delight more than anything else. [1]: http://www.fivespeedsoftware.com/dashboard http://www.fivespeedsoftware.com/dashboard [2]: 15 years later, I open-sourced the code and gave it a proper retrospective: https://github.com/masonmark/Dash-Board-for-Newton-OS https://github.com/masonmark/Dash-Board-for-Newton-OS