Sure, that seems plausible, but I still don't think that translates into the kind of user that would take the StackOverflow survey, which I suspect to mostly be

Sure, that seems plausible, but I still don't think that translates into the kind of user that would take the StackOverflow survey, which I suspect to mostly be users that are quite active on StackOverflow. More generally, I'm seeing StackOverflow metrics underperform on Julia compared to other languages (as opposed to things like Discourse usage for example).