> but at the top level 27% of developers are using Ubuntu desktop.
That seems very unlikely. The sample is of course probably extremely unrepresentative. Also I wouldn't be surprised if selection bias might have a significant effect (a lot of Linux users can't help but try and publicize the fact even if nobody is asking, so imagine what they do when someone does ask... Windows is likely the opposite).
I'm sure stack overflow is a biased subset of users, and I imagine the people that fill out tech surveys are going to be more likely to use Linux. But it's the best data source on this that I know of. Regardless of the exact market share percent, I think it's a reasonable assumption that there are millions of developers using it as a desktop OS.
Along those lines too, I don't have evidence but I'd bet that the type of developer who uses Linux is more likely to integrate new tech like LLM's into their workflow than someone whose on Windows.
That seems very unlikely. The sample is of course probably extremely unrepresentative. Also I wouldn't be surprised if selection bias might have a significant effect (a lot of Linux users can't help but try and publicize the fact even if nobody is asking, so imagine what they do when someone does ask... Windows is likely the opposite).